The Journal 2024

JHEOURNAL 2024

FROM THE HEADMASTER

Hampton Court is the self-proclaimed “most famous maze in the world”, and justifiably so. It is the UK’s oldest surviving hedge maze; it was commissioned around 1700 by William III and covers an impressive third of an acre. All visitors start from the same place and then through a series of personal choices make their way to their destination: different pathways, shared goal. As I read through The Journal it is the rich diversity of the subject matter which particularly resonates for me. Articles focus equally on STEM as well as the Arts and Humanities. They range from the downfall of the Parisian suburbs to attitudes towards China, from the influence of Classics on white supremacy to a case study of Google and Amazon from an Economics perspective, from the rise of smart cities to hypersonic missiles, from machine learning to the rise of obesogens or pain processing in the spinal cord.

At the RGS we genuinely and passionately believe in our philosophy of Scholarship for All; every single student has considerable potential, every single student enters the maze. It is, however, the next steps that fascinate me. Our students are not steered in just one direction. Rather, all our students have the same targets but the pathways they choose, the directions they take, are unique. Students at the RGS genuinely have the potential to be who they want to be; irrespective of their interests they have the opportunity to nurture and develop them. The Journal illustrates this richness beautifully.

I would like to take this opportunity to thank Mrs Tarasewicz, our Head of Scholarship, for all her hard work and inspiration in compiling this impressive publication, to Mrs Webb for designing and producing this, and – most of all – to all those students who have contributed articles which are the result of so much passion, research and reflection: different pathways, shared goal.

Dr JM Cox Headmaster

FROM THE HEAD OF SCHOLARSHIP

Welcome to the 2024 edition of The Journal. The Journal was launched as a place to celebrate some of the finest examples of scholarship produced by RGS students as they approached the end of their time with us. It was modelled upon its namesake, a typical academic journal, with essays published in full for the readers’ perusal.

For this year, I am delighted to launch a new look for The Journal. The task that we set ourselves was to create a layout that was more visually appealing than in the past, breaking up the long blocks of text with colour and visuals, but without in any way compromising the academic quality and integrity of the content. I hope that you feel that we have been successful in our aim.

Our aspiration to create a more appealing, readable Journal mirrors our wider aspirations around scholarship at the RGS. Scholarship is one of our core school values and, as such, is not something that we believe should be limited to just an elite few. We are all scholars. This belief lies at the heart of our Scholarship for All programme, which seeks to offer opportunities for enrichment and challenge to all of our students, as well as supporting them in developing the habits and values of a scholarly mindset.

In the following pages you will find full versions of Independent Learning Assignments (ILAs) and Original Research in Science (ORIS) projects produced by students during their Lower Sixth Form year. Every year, the best of these projects are shortlisted to be presented at the annual ILA/ORIS Presentation Evening, from which two winners are selected, one each in the STEM and Arts/ Humanities categories. It is these shortlisted projects that have been reproduced here. Covering topics from pain processing to historically informed musical performance, and from machine learning to the Parisian suburbs, it offers an insight into the diversity and depth of scholarship here at the RGS.

Details about other Scholarship for All initiatives and how to become involved can be found on The Hub (our intranet site for students), The Unchained Library (our external scholarship website) and the RGS Events page.

My particular thanks to Georgina Webb, our Partnerships and Publications Assistant, for her hard work in producing this year’s Journal and for helping to make our vision a reality.

Mrs HE Tarasewicz Head of Scholarship

With thanks to: Mr Dunscombe, Mr Lau, Mrs Farthing and all of the ILA supervisors for their support with the ILA and ORIS programmes.

SHREY BIJLANI

ROHAN MCCAULEY

FINLAY SANDERS

BEN TABBERNER STEM Winner Highly Commended Short-Listed Short-Listed

PAIN PROCESSING IN THE SPINAL CORD

Validating Multi-Electrode Silicon Probe Placement in the Rat Spinal Cord Using Fluorescence Microscopy for Analgesic Drug Development

SHREY BIJLANI

This report was based on an Original Research in Science (ORIS) placement at the University of Bristol and went on to win the ILA/ORIS Award in the STEM category.

ABSTRACT

The N13 potential is a response in the spinal cord in humans that is thought to reflect post-synaptic activation of neurons. Recordings from rats show the N13 potential has an analogous N1 potential in the rat and might be useful for developing new drugs, however the precise location this potential originates from remains uncorroborated. Electrophysiological experiments were performed to analyse this, and subsequently, tissue was extracted from rats and

stained accordingly. The results show that the probe was positioned approximately 1.2mm deep and 0.5mm laterally with the tip of the probe ending in lamina V. When compared to electrophysiological data, this confirms that the origin of the N13 potential is located within lamina V, supporting the conclusion that the initial response was mediated here.

INTRODUCTION

The spinal dorsal horn is organised into different lamina which are responsible for processing different aspects of sensation (Todd 2010). There are different types of peripheral neurons that respond to differential stimuli. C fibres are responsible for conveying noxious, high-intensity stimulation and synapse with neurons in laminae I and II (Todd 2010). Aβ neurons are sensitive to low-intensity stimulation such as innocuous touch and synapse with neurons in laminae III and V. Investigating the properties of these neurons in the spinal cord can provide insight into the mechanisms of analgesia for pain killers, as well as the pathology of acute and chronic pain conditions.

“ The use of rats allows for an ethically invasive exploration of the CNS as rats have a similar anatomy and physiology to humans owing to their similar spinal structure.

The human N13 potential is a measure of postsynaptic neural activity in the spinal dorsal horn and the amplitude of the N13 potential is thought to be sensitive to analgesics (Di Lionardo, Di Stefano and Leone 2021). Having measures of neural activity in the spinal cord which are sensitive to analgesics would increase our understanding of the spinal mechanisms underlying analgesia. According to prior research findings and clinical trials, it has been established that the N13 potential is modulated by Aβ neurons located in the periphery. However, the precise location within the spinal cord responsible for generating the N13 potential remains unknown and there has been no supporting evidence of the origin of the N13 potential using multi-electrode probes.

Examining the generation of the N13 potential necessitates exploration and employing rats as an experimental model is optimal for this investigation for a number of reasons. The N1 potential generated in the rat lumbar spinal dorsal horn is similar to that of the N13 potential generated in humans. The use of rats allows for an ethically invasive exploration of the CNS and rats have a similar anatomy and physiology to humans owing to their similar spinal structure. (Toossi 2021). Rats are also cost effective as they are small animals which are able to breed quickly and cheap to house.

Therefore, following electrophysiological experiments investigating the properties of the N1 potential in rats and their sensitivity to analgesics, I sought to determine where multi-electrode probes were placed in the spinal dorsal horn, whether or not these probes were placed accurately and how their positioning corresponded to collected electrophysiological data on the spinal N1 potential.

AIM

To determine the origin of the N1 potential in the rat lumbar spinal cord by validating the location of multi-electrode silicon probes with fluorescent markers.

Figure 1: Summary of the methods used as part of the study to accurately determine the location of multi-electrode silicon probes in the rat lumbar spinal cord.

OBJECTIVES

● Sectioning of the lumbar spinal cord at a constant thickness.

● Staining the resulting tissue for better microscopic visualisation.

● Using fluorescent microscopy to image the lumbar spinal cord sections which show evidence of probe markers.

● Scaling and measuring the length of the probe in the sections using vector graphics.

● Calculating the accurate location of the probes.

● Comparing the probe positions with electrophysiological data.

HYPOTHESIS

Summary of surgery and perfusion for investigation of SEPs A 1,1'-Dioctadecyl-3,3,3',3'-Tetramethylindocarbocyanine Perchlorate (DiI) fluorescent marker will show probe placement in the spinal dorsal horn in lamina I-V and the resultant images can be used to accurately confirm the origin of somatosensory evoked potentials in lamina IV-V as part of an overarching and ongoing electrophysiological study.

METHODS

Summary of surgery and perfusion for investigation of SEPs

The tissue used in this study was collected as part of an ongoing investigation of somatosensory evoked potentials in the rat lumbar spinal cord. As such, a brief summary of the methods used in the overarching electrophysiological study has been summarised below.

In isoflurane anaesthetised Wistar rats (n=60, 250-375g), a laminectomy was performed on the 13th thoracic and 1st lumbar vertebra to expose the dorsal surface of the spinal cord. The left sciatic nerve was exposed via blunt dissection and the animal was secured in spinal clamps. A bipolar Ag-AgCl stimulating electrode was hooked onto the sciatic nerve to stimulate low-intensity SEPs that mimicked protocols used in clinical trials to observe N13 potentials in humans. An acute 64-channel silicon probe (Cambridge Neurotech, UK) was inserted into the spinal dorsal horn using a micromanipulator and stereotaxic frame to record SEPs from the lumbar spinal dorsal horn. The probe was painted with a DiI stain to allow for post-hoc investigation of the probe placement using histology and fluorescent microscopy.

refrigerated at 4°C.

Data from the electrophysiological experiments were analysed in MATLAB R2022a using custom written code and the statistical comparisons visualised in GraphPad Prism v9.4.1. The responses of the SEPs following low-intensity stimulation were averaged and the data was extracted from each channel for the 30-minute baseline period.

Sectioning and Staining

After extraction of the spinal cord from a rat, the lumbar region was removed for sectioning using a single-edge razor blade. The DiI fluorescent probe marker (a point on the spinal cord showing where a silicon probe was previously inserted) was identified as a pink mark on the dorsal surface of the lumbar spinal cord and the tissue was cut 2mm either side of this mark.

Baseline responses to low-intensity stimulation were recorded for 30 minutes before either a vehicle or drug solution was administered (1ml/ Kg, I.P. injection). Post-dose, spinal SEPs were recorded for up to 90 minutes to observe changes to the amplitude of the SEP until the experimental endpoint, at which time the animal was killed and perfused with phosphate buffered-saline (PBS). The spinal cord was removed from the vertebral column and fixed in paraformaldehyde for 24 hours before being transferred to a 30% sucrose solution and

Figure 2: Mounted slide. 24 spinal cord sections from the lumbar region at a 50μm thickness. Sections with a fluorescent probe marker have a pink mark and the section where the probe marker is the deepest is shown in a red box. The section has been enlarged and shown as a brightfield image.

The spinal cord was sectioned using a freezing microtome. The spinal cord was positioned onto a cooling Peltier using Cryomatrix embedding solution with the rostral end of the cord directed vertically upwards. The Peltier began to cool to approximately -20°C. Additional Cryomatrix was added on top of and around the spinal tissue to secure its vertical position during the freezing process.

3: Chemical structure of 4′,6-diamidino-2-phenylindole. Produces fluorescence given by minor groove binding and intercalation in the DNA pocket in preference to adenine-thymine clusters.

It was important that the length of the spinal cord was not much longer than 4mm, ensuring that the Cryomatrix would freeze on top of the spinal cord. Once the Cryomatrix was frozen (forming an opaque white solid), a sharp blade was attached to the microtome and adjustments to the blade’s height were made to match the height of the Cryomatrix around the spinal cord. 50μm sections were then cut manually and individually transferred to wells filled with PBS in a microtitre plate using a thin paintbrush. As a buffer solution, PBS has a similar ion concentration, pH and osmolarity to that of biological tissue, preventing sections from becoming either turgid or flaccid. On completion of tissue sectioning, microtitre plates were covered with aluminium foil and refrigerated. Aluminium foil was used to ensure that the DiI stain maintained its fluorescence, which would be otherwise compromised in the presence of light.

Prior to the mounting process, it was essential to identify a critical area within the sections. This refers to a group of sections where the probe marker can be identified, highlighted as a pink line by the DiI stain. Once identified, sections were mounted in series onto 25mm × 75mm × 1mm microscope slides (Epredia Superfrost, UK) according to a prescribed ratio: sections outside the critical area were mounted at a 1:5 ratio, while sections within the critical area were mounted at a 1:1 ratio.

Approximately 24 sections were mounted on a single slide. This was achieved by dipping the sections in distilled water to wash excess PBS solution from the surface of the section and transferring them to a slide using a thin paint brush. Washing in distilled water is vital in preventing the solidification and build-up of salt around the section on the microscope slide. Once all sections were mounted in this way, the slides were kept in darkness at ambient conditions for 24 hours.

Once the mounted sections were dry, they were stained using 300nM 4′,6-diamidino-2phenylindole (DAPI), a fluorescent stain that binds strongly to adenine–thymine-rich regions in DNA to highlightneuronal cell bodies which appear blue under ultraviolet light. 200μm of DAPI was added to each slide using a pipette and spread across the slide, ensuring that all sections was fully covered. The sections were left to stain in darkness for approximately 5 minutes and excess stain was washed off in a slide bath. They were then left to dry for 1 hour in an incubator at approximately 37°C. Once dry, 3-4 drops of Fluorsave reagent was added onto each slide using a pipette and a 24mm × 50mm coverslip was placed on top. If air bubbles were present on a particular slide, the pointed tip of tweezers were used to tease the bubbles away from the sections, allowing for a clearer final image.

Figure

Bioimaging and Microscopy

All slides were initially visualised under a Leica DM 4000B fluorescent microscope (Leica Camera, Wetzlar Germany) at a 5× magnification. An ultraviolet filter was used to show the DAPI staining of neuronal cell bodies, and a green chromatic filter was used to produce a bright red fluorescence in areas of high DiI concentration, thus confirming whether or not a probe marker was present. After identifying a single section per spinal cord that showed the brightest DiI red fluorescence, the slides were scanned at a high resolution on an Olympus Slideview VS200 Slide Scanner (Olympus Microscopy, Japan).

Image Analysis

Once scanned, the microscope slides were digitised as tiff files and initially appeared in greyscale, showing the binary intensity of fluorescence for each particular stain. To better visualise the fluorescent marker, FIJI, an open-source image processing program (LOCI, University of Wisconsin), was used to merge the blue colour channel (highlighting the

5: Output image after running the macro instruction. A single lumbar spinal cord section at a 50μm thickness under a fluorescent microscope showing a white line providing the angle of probe insertion and estimated probe depth.

neuronal cell bodies of the section) with the red colour channel (highlighting the area where the probe was inserted) to create an overlaid image.

To determine the probe position in sections with probe markers present, a macro instruction was created. It described a threshold value that established a limit after which no more fluorescence was detected. The macro also described drawing an elliptical region around the fluorescent area of the image, quantified by the threshold value, and then drawing a line connecting the two most distant points within this ellipse. This process yields two key pieces of information: the length of the fluorescence within a particular section (and therefore an estimate for the depth of the probe), and the angle at which the probe was likely inserted.

Figure 4: Using FIJI to create an overlaid image. A single spinal cord section from the lumbar region at a 50μm thickness under a fluorescent and brightfield microscope showing the relative probe position. A. Brightfield image. B. Yellow light fluorescence appearing as greyscale. C. Ultraviolet light fluorescence appearing as greyscale. D. Merged colour image with neuronal cell bodies in blue highlighted by the DAPI stain and the area of probe insertion in red highlighted by the DiI stain.

The final step in the image analysis was to perform accurate measurements determining the precise position of the probe within the lumbar spinal cord section using Inkscape, an open-source vector graphics editor.

Figure

To perform the calculations of the probe position, multiple additional images were uploaded to Inkscape as individual layers, including the original overlaid image (Figure 4, D), the output image from the macro instruction (Figure 5) and an image of the multi-electrode silicon probe. Calculations were performed to scale the initial size of the probe in Inkscape to the appropriate size.

Below is an example of how the depth of the probe was calculated from the top of the cross-section using Inkscape:

RESULTS

The results of this report show the identification of multi-electrode silicon probes inserted into the spinal dorsal horn to record somatosensory evoked potentials.

The measurements in Table 1 were recorded during the electrophysiology experiment using a micromanipulator to identify where the probe was positioned in the rat lumbar spinal cord. They all show a rostral-caudal measurement of -7mm, a medial-lateral measurement of -0.3mm and a dorsalventral measurement of -1.0mm from the T12-T13 vertebral boundary.

Table 1 The rostral-caudal, medial-lateral and dorsal-ventral probe positions in 5 rat lumbar spinal cords.

Figure 6 Final images produced using Inkscape vector graphics. 5 spinal cord sections from the lumbar region at a 50μm thickness under a fluorescent microscope with an animal ID showing the relative probe position as compared to the structure of an L5 rat spinal cross-section. Neuronal cell bodies in blue highlighted by the DAPI stain and the area of probe insertion in red highlighted by the DiI stain.

Table 2: Means and standard deviations of probe placement in the spinal dorsal horn. Major axis length: length of fluorescence produced by the macro instruction. Probe depth: depth of the probe from the top of the section. Medial-lateral: Horizontal deviation of the probe from the middle of the section. Orientation: Angle at which the probe was inserted taken counter-clockwise from the horizontal. Lamina: Shows the lamina number in which the probe tips reached.

“ These results can be used to accurately confirm the position of the probes when used in conjunction with the electrophysiological data.

Values for the probe positions in sections A-E were calculated using Inkscape and the means and standard deviations for the measurements are shown in Table 2. Individual measurements for the probe depth are represented in Graph 1 below. These results can be used to accurately confirm the position of the probes when used in conjunction with the electrophysiological data.

Figure 7 shows 3 of the 5 sections where the corresponding electrophysiological data was provided. Each image consists of an Inkscape image showing the individual electrode sites on the surface of the probe, a heat map showing amplitude against time in a colour chart to display where the peak is, and an SEP waveform showing the shape of the signal over time, emphasising the N1 potential. In addition, other slower conducting peaks are shown on the heat map in light blue.

Graph 1: Bar chart showing the values of the depth of the probe in five rat lumbar spinal cord sections. Mean value of probe depth is shown by the blue bar, standard deviation is shown by the thin black bar with measurements of each individual probe depth plotted vertically as red dots.

Figure 7: Comparing probe placements with electrophysiological data. A heat map showing the amplitude of signal and latency, and a waveforms graph showing the electrode channels where electrical activity is recorded are presented alongside 3 images produced on Inkscape.

Across the 3 sets of data presented below, the peak responses were recorded on electrodes in rows 20 to 40 as shown on the SEP waveforms. When compared to the images created on Inkscape (Figure 6) and the probe depth measurements (Graph 1), this directly correlates to lamina IV-V in the lumbar spinal dorsal horn. However, the probe in section A has been inserted deeper which explains why the peak amplitude of the N1 potential is recorded towards the top of the probe. The peak amplitude in recorded in sections B and C is more central as the probe is shallower in the spinal cord. Therefore, the N1 potential is recorded on medial channels.

DISCUSSION

Experimental procedure and wet lab techniques used to obtain the final results had to be regularly assessed and modified, and were not without its limitations. Having started off with a total of 9 spinal cords, only the sections from 5 spinal cords were usable as part of the study. Some spinal cords became especially fragile having been perfused from the rat vertebral column and were unable to be sectioned. Of those that were, many sections became damaged while using the freezing microtome, including those showing evidence of fluorescent probe markers. Furthermore, condensation on the microscope slides and coverslips after refrigeration may have been responsible for the lack of clarity of several images after being scanned.

During the sectioning process, it became clear that using a section thickness of 25μm left sections susceptible to tearing, which reduced the quality of the spinal tissue. Therefore, changes were made to the methods to increase the thickness of the sections from 25μm to 50μm. This was shown to increase their durability during the mounting process, which ensured that tissue was successfully stained and mounted while preserving the integrity and clarity of the resultant images, highlighting the significance of thickness selection in minimising tissue tears and breakages.

After mounting the first spinal cord section, I came to realise that the direction in which the sections were facing on the surface of the slide was not kept constant, resulting in the exact distance between each section being different. In order to maintain a single facing direction for all sections moving forward, I employed a simple scoring technique which involved making a shallow cut down the contralateral ventral surface of the spinal cord in advance of sectioning. When the sections were then mounted, they were placed on slides with this cut positioned on the left-hand side, allowing for the rostral end of each section to be constantly directed upwards. This, in turn, meant that the probe marker always appeared on the same side regardless of the section, making the microscope images more accurate when analysing. However, introducing an additional cut to sections made it difficult to mount onto microscope slides and some were damaged.

The rat spinal cord perfused from the vertebral column in a dish of 30% sucrose solution with peripheral nerves branching out from the cord.

The fluorescent dye successfully stained the position of the probe and the cell bodies of neurons within the spinal dorsal horn across 5 rats. The anatomical features of the sections identified with DiI fluorescence closely match those of lamina V of the spinal cord, confirming the expectations of the electrophysiology. Figure 6 also shows a similarity regarding the shape and size of the sections. This holds true when compared to the cross-sectional image of the rat L5 spinal cord from the rat spinal cord atlas, especially with regards to the shape of the grey and white matter.

By considering Table 2, it can be deduced that the major axis length provided by the macro instruction served solely as an estimate for probe depth, where the length of the fluorescence was notably larger than the depth of the probe calculated using Inkscape. This may suggest a possible limitation in using computation to define a region of fluorescence

based on a single threshold value used across all 5 sections. Therefore, fluorescence would be better used as a visual indicator of probe position rather than a quantitative value which serves to estimate probe depth. It is worth noting that the DiI stain which gives rise to fluorescence under the microscope may have diffused through the spinal tissue to a certain extent, further undermining the values for major axis length in reflecting probe depth. A similarly large discrepancy can be seen when comparing the mean medial-lateral value in Table 2 to the medial-lateral values in Table 1. It is now clear that the reason for this was due to the fact that measurements obtained using the micromanipulator were taken from the left hand side of the central vessel instead of the midline of the spinal cord as done on Inkscape. This would have accounted for the difference of approximately 200μm between the two results.

The measurement of the depth of the probe recorded in Table 2 showed a strong similarity to the dorsal-ventral position calculated as part of the electrophysiology in Table 1. A minor difference between the two may be due to the inclusion of the result obtained from section A in Figure 6. While he depth of the probe in section A was deeper compared to sections B, C and D, this result is clearly reflected in the electrophysiological data shown in Figure 7. Similarly, the value of orientation is important in showing that the probe insertions across all 5 sections were similar and inserted nearly perpendicularly. However, insertion at an angle may have caused a greater number of electrodes to be present in certain lamina compared to others.

“ This study has contributed to the understanding of the origin of the N13 potential.

Analysis of features of the electrophysiological data show that the N13 peak has a latency of less than 5ms and an average maximum amplitude of over 1000μV. This this suggests that the observed phenomena are the result of post-synaptic neural activity in the spinal dorsal horn to an Aβ-mediated response. Aβ primary afferent neurons conduct sensory information at velocities less than 100ms which matches the time course of the evoked response here. Moreover, these results show that the origin of spinal SEPs occurs in laminae IV-V of the spinal dorsal horn, which is expected because Aβ fibres form synapses with neurons in laminae III-V. The data in Figure 7 also shows a second peak in the heatmaps and waveform plots, suggesting a second component of the SEP which could reflect the activation of slower conducting peripheral neurons or activation of second order neurons in the spinal

cord. It is thought that these wide dynamic range (WDR) neurons, which respond to both painful and innocuous touch, contribute to the generation of the N13 potential in rodents and humans (Di Pietro, Di Stefano and al. 2021). The electrophysiological data proposes that WDR neurons may contribute to the generation of SEPs in the spinal dorsal horn which in turn may be sensitive to analgesics such as Tapentadol. More work, including single unit analysis of this data, is needed to understand which spinal neurons are responsible for the generation of the N13 potentials. However, this study has contributed to the understanding of the origin of the N13 potential in rat, which has important ramifications for ongoing studies that investigate the N13 potential as a biomarker of analgesic drug discovery.

After constructing images of the 5 sections in Figure 6, corresponding electrophysiological data was unavailable for all 5 recordings. 2 of the 5 recordings were discounted from the analysis pipeline and these figures were not pursued. This was likely because of dirt on the stimulating or recording equipment. However, it is worth noting the placement of the probe in both section B and E (Figure 6). The value for the depth of the probe in section B is very low compared to the other 4 sections and the majority of the electrodes lie between lamina I-III. In section E on the other hand, the probe tip ends in lamina V as originally planned and the majority of the electrodes on the probe lie between lamina III-V. We can therefore expect the electrophysiological data corresponding to section E to be similar to sections B and C in Figure 7.

CONCLUSION

This study has validated the position of probes in the spinal dorsal horn with DiI fluorescent probe markers using a range techniques within histology, bioimaging and vector graphics. This serves to contribute to a wider study where multi-electrode electrophysiology has been used to show where the N1 potential is generated in rats for the first time, providing the groundwork for novel testing of analgesics on patients suffering from chronic pain. This has important implications for understanding the generator of the human N13 potential and future analgesic drug development. Nevertheless, further investigation of the N13 potential in humans and the N1 potential in rats is needed, and future studies should use lamina-specific markers to more clearly identify anatomical landmarks to assist with the tissue analysis.

An image taken from Inkscape showing how lines and shapes were used to calculate the depth of the probe.

Experimental set-up for electrophysiological experiments on rats using electrodes to monitor and record electrical activity.

LITERATURE CITED

1. Di Lionardo, A, Di Stefano, G and Leone C et al. Modulation of the N13 component of the somatosensory evoked potentials in an experimental model of central sensitization in humans. Sci Rep, 2021.

2. Di Pietro, G, Di Stefano, G, and Leone C et al. The N13 spinal component of somatosensory evoked potentials is modulated by heterotopic noxious conditioning stimulation suggesting an involvement of spinal wide dynamic range neurons. Neurophysiol Clin., 2021: 517-523.

3. Todd, A.J. Neuronal circuitry for pain processing in the dorsal horn. Nat Rev Neurosci., 2010: 823-836.

4. Toossi, A., Bergin, B., Marefatallah, M. et al. Comparative neuroanatomy of the lumbosacral spinal cord of the rat, cat, pig, monkey, and human. Sci Rep, 2021.

5. Watson C, Paxinos G, Kayalioglu G, Heise C. Chapter 15 - Atlas of the Rat Spinal Cord. The Spinal Cord, 2009: 238-306.

ACKNOWLEDGEMENTS

I extend my formal appreciation to the following individuals and facilities for their contributions to this project. Kenneth Steel for the day-to-day supervision and experimental planning, Dr Tony Blockeel for his expertise of somatosensory evoked potentials and Professor Tony Pickering for his support in securing my research project. I also thank the Histology Services Unit for granting access to their freezing microtome and the Wolfson Bioimaging Facility, particularly Katy Jepson for granting access to the Olympus Slideview VS200 Slide Scanner and Dr Stephen Cross for helping to develop the macro instruction. Their collective efforts greatly facilitated the success of this study.

THE RISE OF OBESOGENS

Could synthetic chemicals be the hidden catalysts of the obesity epidemic?

STEM ROHAN

M C CAULEY

This Independent Learning Assignment (ILA) was highly commended at the ILA/ ORIS Presentation Evening.

ABSTRACT

Obesity is a huge problem in both the developed and developing world. Rapidly rising levels of obesity mean that every year, a greater proportion of the population is at risk from diseases such as type II diabetes and various cardiovascular disorders. My ILA aims to explore the role of obesogens, endocrine-disrupting chemicals that contribute to obesity, by examining their impact on factors such as adipocyte differentiation and Appetite control which lead to weight gain. Understanding obesogens is crucial for effective policy-making and prevention strategies, although it is evident that factors such as diet and exercise are ultimately more significant, and that tackling the obesity epidemic is an incredibly complex issue which requires the consideration of a broad variety of contributing factors.

INTRODUCTION

An Overview of the Obesity Epidemic

In 1997, the World Health Organization (WHO) declared obesity to be a global epidemic1

Since then, numbers have only continued to rise increasingly rapidly. The largest observed increase has been in the US; National Health and Nutrition Examination Surveys (NHANES) have found that the proportion of obese adults in the age range 20-74 has increased from around 14.5% of the population in the period 1976-19802 to 30.5% in the period 1999-20003 - an increase of approximately 110% in only around 20 years. As of 2018, the figure sits at around 42.8%, and by 2030, just under half of all US adults are expected to be of obese weight status4

“ Rapidly rising levels of obesity mean that every year, a greater proportion of the population is at risk from diseases such as type II diabetes and various cardiovascular disorders.

The commonly proposed reason for this astronomic increase, and a fixture of the health and wellness industries, is the concept of ‘calories in, calories out’. On the surface, this model of weight gain has its merits - as diets and, more broadly, lifestyles, have shifted in the developed world to favour energy consumption over expenditure, so too has the incidence of obesity dramatically increased5. At the same time, this concept is insufficient as it massively simplifies a problem rooted within a complex spectrum of socioeconomic factors, belying a huge range of confounding variables such as the quality of those calories consumed as well as a demographic’s access (or lack thereof) to good quality food.

It would also be wrong to assume that this problem is limited to developed countries. Whilst the prevalence of obesity is generally lower across African and South East Asian countries, more recent trends show that the mean body mass index (BMI) in many of these developing countries is on a sharp rise, and with it, the proportion of adults and children that are obese6,7. A case study in The Gambia published in 2020 found that obesity rates in Gambians aged 16 years and over had increased from an estimated

2% in 1996, to a prevalence of 8% in men and 17% in women in recent years, particularly in urban areas8. This global rise is of huge concern as there are well-established links between obesity and a huge range of further health complications. Obese individuals are more likely to suffer from a wide range of health complications including (but not limited to): type 2 diabetes; coronary heart disease, heart failure and strokes; respiratory problems such as asthma; weakened immune systems; cancer; and kidney disease9. This comes not only at a great personal cost for the sufferer, but also as an extreme burden on health services globally - as of 2014, the estimated annual cost of obesity sits at around two trillion dollars, due to the direct costs of healthcare as well as the indirect costs of lost economic productivity10,11 . It seems clear that as we have transitioned to an increasingly automatised and sedentary lifestyle, and as the volume of readily available, high-calorie food has increased, the prevalence of obesity has followed suit. However, the question remains; within the complex web of factors and variables responsible for the expansion of this epidemic, could one comparatively inconspicuous factor be driving the increase at an ever faster rate?

INTRODUCTION TO OBESOGENS

An endocrine-disrupting chemical (EDC), also called an endocrine disruptor, is any chemical, natural or artificial, that can interfere with the endocrine system, usually by mimicking or blocking the action of the body’s own hormones12,13. Just under 1500 potential EDCs have been classified14, and these come from a wide variety of sources, including pesticides, flame retardants, cosmetics, and compounds used to produce plastics.

The possibility that EDCs may be able to promote obesity in humans was first hypothesised in 2002 by Dr Paula Baillie-Hamilton15,16, and in 2006 the term ‘obesogen’ was coined in a paper published by Felix Grün and Bruce Blumberg to describe such a molecule. In the paper, Grün and Blumberg define obesogens as “molecules that inappropriately regulate lipid metabolism and adipogenesis to promote obesity”17. In other words, they are substances that promote the differentiation and proliferation of white adipose tissue responsible for the characteristic weight gain seen in obesity.

White adipose tissue (WAT) is made up of white adipocytes, cells responsible for the storage of energy as triglycerides. When the body’s energy expenditure requirements outweigh its energy intake, the adipocytes break down the triglyceride store via lipolysis, releasing free fatty acids which can be oxidised readily for the production of large amounts of ATP18.

Much of the WAT is subcutaneous, meaning it is stored under the skin. This allows it to act not only as an energy store but also as heat insulation and a protective buffer against impact. WAT is also present as visceral adipose tissue, which is packed around the intraabdominal organs such as the stomach, intestines, and kidneys20. However, problems can arise when these cells proliferate too expansively, either through an increase in adipocyte number, called hyperplasia, or an increase in the size of individual adipocytes, hypertrophy21. The storage capacity of the subcutaneous WAT, the largest depot of adipose tissue, is limited. This means that excess adipose tissue accumulation increases the load of the visceral adipose tissue, and can also lead to fat accumulation in abnormal areas, such as in excessive quantities around the liver and heart21. A potential consequence of this is the accumulation of toxic lipid

compounds in non-adipose tissue which can lead to cellular dysfunction and in some cases cell death, a condition known as lipotoxicity22

Furthermore, it has been observed that the proliferation of WAT induces a dangerous inflammatory response23. In 1994, the discovery of leptin, a hormone secreted by adipose tissue24 revealed that WAT functions not only as a storage tissue, but also as an active endocrine organ, and in addition to the wide variety of cytokines, hormones, and other products secreted by WAT (jointly referred to as ‘adipokines’), when under stress - as they are in the case of obesity - adipocytes secrete inflammatory mediators and chemoattractants such as monocyte chemoattractant protein-1 (MCP-1)25. These attract and recruit macrophages to the tissue, which themselves secrete, among other things, tumour necrosis factor (TNF-α), a cytokine (also secreted in smaller quantities by stressed WAT) that promotes an inflammatory response, perpetuating the process26,27 . However TNF-α also promotes the phosphorylation of serine, an amino acid, in the protein ‘insulin receptor substrate 1’ (IRS-1), which inactivates it and hence impairs the insulin signalling pathway28–30. The culmination of this series of events is that the cells eventually become insulin resistant. As insulin promotes the maturation and proliferation of adipocytes by stimulating triglyceride synthesis and preventing lipolysis28, the desired effect of inducing insulin resistance may be, in the case of obesity, to try and prevent further accumulation of adipose tissue. However, as discussed earlier, the amount of visceral and ectopic (abnormal) adipose tissue is greatly increased in obese individuals, meaning that important insulin-sensitive tissues, such as muscle tissue, and organs, such as the liver, are exposed to the effects of this insulin resistance as well29. As a consequence, obese individuals are at a much higher

19TEM micrograph of a white adipocyte: N is the nucleus, M the mitochondria, and L the lipid droplet

risk of developing type 2 diabetes; in the period 1999-2002, 54.8% of US type 2 diabetics were also obese, and if you include those in the overweight weight category (a BMI greater than 2531), the number rises to 85.2%32 .

From this it can be seen that obesity is characterised by an increase in adipose tissue mass and volume, and that the risk factors associated with obesity stem from this proliferation. Obesogens function by promoting the differentiation and growth of the adipocytes that make up these tissues through a number of different mechanisms, so it is clear why they should be a matter of concern.

BIOLOGICAL MECHANISMS OF OBESOGENS

How exactly do obesogens, generally simple molecules, cause the body to produce abnormal numbers of adipocytes? As a loose definition, the term obesogen encompasses a huge class of different molecules which can function in a multitude of different ways. This sections aims to discuss a few of the most common/well-researched biological mechanisms by which an obesogen may work.

3a. Nanomolar Ligand Affinity for ‘Master Regulator’ Nuclear Receptors

In the late 1960s, tributyltion (TBT), a toxic biocide, was mixed into paints to function as an effective antifouling agent for ship hulls33. Discovered in 1954 by a research group from the Netherlands34, TBT was extremely effective at preventing barnacles and other marine organisms from infesting the bottom of wooden and steel plated boats. However, the TBT slowly leached from the paint into the surrounding marine environments. Due to its relatively long half-life, particularly in anoxic marine sediments, TBT is able to accumulate in ocean floor sediments, only to be released back into the seawater, re-contaminating the area33. (To be precise, TBT is an umbrella term given to a closely related family of organotion compounds with similar characteristics – the organotion commonly used in hull paints was primarily bis(tributyltion) oxide, TBTO)

Scientists first began to notice the effects of this TBT contamination in the 1970s, when numerous marine species were observed to have developed abnormal disorders. In particular, scientists noticed a severe decline in populations of the rocky shore sea snail, Nucella Lapillus, commonly known as the dog whelk. They found that this decline was due to the masculinisation of female dog whelks, a condition known as ‘imposex’, which was induced by TBT35.

37Photographs showing the development of a penis and vas deferens in imposex female dog whelks

46Structure of the PPAR-γ receptor bound to DNA

Imposex females are characterised by the growth of a penis and vas deferens (sperm duct) which, at high enough concentrations of TBT, grew large enough to block the vulva and prevent the release of egg capsules, hence rendering the female sterile36.

Further research was carried out on TBT to try and deduce the mechanism by which this could happen, and in the early 2000s Bruce Blumberg (mentioned earlier), whilst in a meeting in Japan, heard about a research group which had found that TBT also caused masculinisation in fish38,39. Hypothesising that the TBT functioned by activating a sex steroid receptor, he conducted his own research, testing to see if TBT activated any known nuclear receptors. Instead, he found that it activated the nuclear receptor ‘peroxisome proliferator-activated receptor gamma’ (PPAR-γ)40. Further study showed that TBT is also an agonist to retinoid X receptors, which is theorised to be the mechanism by which imposex is induced in gastropods such as the dog whelk41,42.

PPAR-γ is a nuclear receptor that is highly expressed in adipose tissue, and has been named the ‘master regulator’ - in other words, the apex, or most important, factor in a biological regulatory hierarchy - of adipogenesis, the process by which adipocytes mature from stem cells41,43. It forms a heterodimer with RXR, which when subsequently bound to a ligand, regulates the transcription of target genes. Through an increase in the mRNA expression of genes which promote fatty acid uptake and storage, and a decrease in the expression of

genes that induce lipolysis in WAT, the heterodimer stimulates the differentiation of multi potent stem cells and preadipocytes into adipocytes, in addition to increasing the size of existing adipocytes41,44,45. Therefore, obesogens such as TBT, which have a high ligand affinity for these particular nuclear receptors may be inducing obesity by acting as agonists to indirectly promote the expression of genes responsible for the differentiation and proliferation of white adipocytes. Animal studies have demonstrated that TBT induces adipogenesis in mice and frogs and that prenatal exposure has a particularly significant effect on adipose tissue accumulation, regardless of normal postpartum diet and exercise47 . Perhaps the most troubling fact to consider is that, as ligands, organotions such as TBT do not bind in a typical manner, particularly so in RXR activation. Typical RXR ligands have a carboxylic acid functional group, and mimic the structure of 9-cis-retionoic acid. TBT and its associated organotions, put simply, do not follow this trend in the slightest, but nonetheless are potent agonists to both the PPAR-γ and RXR receptors47. This strongly hints at the possibility that there are a wide range of substances and molecular structures that could have the potential to induce obesogenic effects through similar mechanisms to TBT and so, although TBT may have been the first obesogen to have been identified by Blumberg et al. in their seminal paper from 2006, it is certainly not the only one.

TOP: 48chemical structures of several typical RXR agonists; A – natural agonists; B – synthetic agonists. Note the carboxylic acid functional groups, CO2H, and broad structural similarity to 9-cis-retionoic acid (top left)

BOTTOM: the structure of bis(tributyltion) oxide, TBTO, also an RXR agonist, but atypical in terms of functional group and 3D structure

3b. Appetite and Satiety Dysregulation

The craving for a sugary snack in the middle of the night, or hunger after a long run, and conversely the feeling of fullness attained after a hefty meal stem from the same area of the brain: the arcuate nucleus (ARC). The ARC is a region of the brain responsible for Appetite and satiety49, and hence controls and maintains energy homeostasis. Neurons that stimulate Appetite, such as those that produce neuropeptide Y (NPY) and agouti-related peptide (AgRP) are said to be orexigenic; and those that inhibit satiety (i.e induce an Appetite), such as the neurons that produce pro-opiomelanocortion (POMC) and cocaine- and amphetamine-regulated transcript (CART), are anorexigenic50. These neurons develop prenatally in a process referred to as neurogenesis (although the occurrence of neurogenesis in adult human brains is hotly contested and is the subject of much research)51.

BPA was detected in the urine of over 92% of the participants53. Despite this, based on the current levels of BPA occurring in foods, the FDA considers BPA to be safe54 - regardless, a strong case is being made as to the endocrine-disrupting potential of BPA, and it is a strong contender on a growing list of potential obesogens.

Bisphenol A (BPA) is a synthetic compound widely used in plastic manufacturing, predominantly for the production of polycarbonate plastics and epoxy resins used to line food and beverage packaging, and has been in use since the 1960s52. Many may be familiar with it, as many household appliances and plastic products around today proudly declare themselves to be ‘BPA free’. This is due to various health concerns linked to BPA, although the presence of ‘BPA free’ products may be of little comfort - in a representative sample of the US population,

Different studies have postulated that BPA may be obesogenic through a variety of mechanisms, but one particular mechanism is through its interaction with neural progenitor cells (NPC), the precursors to neuronal cells, such as the NPY/AgRP orexigenic neurons. Studies conducted on mice suggest that prenatal BPA exposure leads to an increase in unregulated neurogenesis, altering brain structure and function55–57, and potentially disrupting the balance between Appetite stimulators and inhibitors, leading to, later in life, increased food intake and hence weight gain.

BPA is not the only compound theorised to have an impact on Appetite. For example, monosodium glutamate, or MSG for short, is a flavour enhancer that is particularly popular in Asia; recently studies have proposed that it may disrupt the secretion of glucagon-like peptide-1 (GLP-1), a hormone that, among other functions, acts in the hypothalamus to promote satiety58 - 60. Considering the socio-economic context of the obesity epidemic itself, in the majority of cases, decreased satiety and increased Appetite will lead to an increased consumption of cheap and readily available food, sacrificing the quality of the calories consumed for the volume of calories provided; and whilst this mode of action, an indirect exacerbation of the existing factors that have been driving the obesity epidemic, is much less well-researched than the mechanism discussed previously, it seems just as, if not more, worrying.

3c. The Search for New Mechanisms

The obesogen field is still a relatively novel area of research and study; around 50 obesogens have been identified so far, and the mechanisms for most of them are yet to be determined61. One growing field of interest is in a transgenerational approach to studying obesogens, as evidence suggests that certain obesogens, like TBT and BPA, can cause effects that are inherited across multiple generations62. However, gaining a better understanding of factors like the role of epigenetics - a relatively new field itself - in hereditary obesity is essential for further exploring the potential consequences of obesogens.

It is therefore important to keep in mind that, whilst much more is understood now about obesogens than when they were first discovered, there is still much to learn and so at this time, determining the relative impact of such chemicals on a global scale is, to put it mildly, difficult.

4. Epidemiological Evidence for Obesogens

ruin an epidemiological study - in the case of TBT, it was recently shown that plastic containers, widely used in studies on the prevalence of TBT in human specimens, strongly bind organotions themselves, meaning that the estimated organotion levels in the specimens from these studies was most likely significantly underestimated59,61

It may be unsurprising, therefore, that perhaps the strongest epidemiological evidence for the impact of obesogens itself predates the obesogen field.

Numerous studies have found that, whilst nicotine is an Appetite suppressor, there is a strong causal relationship between pregnant smokers and an abnormal rate of weight gain in their corresponding children once born, highlighting the consequences an EDC such as nicotine can have for the development of obesity16,63 .

“ 50 obesogens have been identified so far, and the mechanisms for most of them are yet to be determined.

Some of the mechanisms by which an obesogen may function have been discussed; however, is there sufficient epidemiological evidence to suggest they could be a leading factor in the expansion of the obesity epidemic? The global transition to a much more urbanised lifestyle for most of the developed and developing world, and the accompanying changes in diet and exercise that have followed clearly correlate with the rising trend in obesity, and so it would be perfectly reasonable to assume that, in comparison to such a huge societal shift, the effect of obesogenic chemicals would be minimal.

Unfortunately, even for TBT, the most well-researched obesogen, epidemiological studies of human TBT exposure are few and far between. This may be due to the difficulty of accurately determining gross exposure to any given chemical: the exposure may occur through multiple routes; the detected concentrations of chemicals with short half-lives will vary significantly over time; and sample contamination during the collection of data can

On the topic of childhood weight gain, trends in the prevalence of childhood overweight and obesity cannot be ignored. In the period 1980-2005, across the school-age population of 25 countries and pre-school populations of 42 countries, an increase in the prevalence of childhood overweight could be observed in almost all countries, with the only exceptions being among school-age children in Russia and Poland in the 1980s64. This is particularly significant, as the factors typically associated with the obesity crisis, particularly in the west, such as diet and a sedentary lifestyle, are not as predominant for children and infants. Generally, the extent to which one child may be more or less active than another is much less pronounced than between adults, and so this hints that there are external factors at play, especially when accounting for as sharp a rise in childhood obesity as has been observed in the past few years - in 2010, 43 million children under the age of 5 were estimated to be overweight or obese, an increase of 60% in only 20 years65,66.

Finally, although studies on obesogens in human populations are challenging to effectively carry out, model studies on animals can give researchers indicative and reliable results. Exposing mice,

rats and zebrafish to obesogens such as TBT have shown clear evidence of induced weight gain across multiple studies41,47. Whilst these animals are not perfect models, they are much closer to humans than, say, the molluscs claimed as victims by TBT. All three animals mentioned share roughly 70% of their DNA with humans67,68, and the mechanisms of adipogenesis and weight gain between these animals and humans are largely evolutionarily conserved69 Overall, it does not seem like a stretch to imply that obesogens could likely be playing a role in the accelerated epidemic of obesity.

CONCLUSION

The field of obesogens, and indeed that of EDCs in general, is still a novel area of research. Whilst it may not currently be possible to arrive at a definitive conclusion as to the magnitude of their impact, the evidence available indicates that they should be a serious cause for concern, and if we want to slow the current acceleration of the obesity epidemic, we need to carefully consider how, going forwards, the potential negative effects of such compounds can be mitigated.

Policy-making plays a crucial role in addressing the obesity epidemic and mitigating the impact of obesogens, and so governments and regulatory bodies need to prioritise the regulation and monitoring of potential obesogens in consumer products, although long-term human studies and further epidemiological research will be necessary to comprehensively assess the effects of these chemicals on human health and inform evidence-based policies.

Having said that, it also seems apparent that, if a world were to exist where all demographics had access to high quality food, frequent exercise and generally healthy lifestyles, the introduction of obesogenic chemicals at the levels present in our environment would not single-handedly fuel a global obesity epidemic. I do not believe that the evidence presented denotes that obesogens are the core drivers of the epidemic; rather, obesogens may exacerbate existing problems within urban societies, accelerating an epidemic that has its roots in a much less tangible web of factors.

For this reason, in addition to policy interventions, preventive measures should focus on promoting healthier lifestyles, including balanced diets and regular physical activity. Education and awareness campaigns can help individuals make informed choices and adopt healthier behaviours, and collaborative efforts among healthcare professionals, policymakers, food industries, and communities are essential for effective prevention and management of obesity.

In conclusion, it is clear that no single factor can be held solely culpable for the rapid global rise of obesity. Addressing the obesity epidemic requires a multi-faceted approach that considers the complex interplay of factors influencing weight gain. Understanding the role of obesogens and their mechanisms of action is pivotal in order to develop effective prevention strategies, and as more research becomes available, implementing evidence-based policies, in addition to promoting healthier lifestyles, will be crucial in order to secure a future with reduced obesity rates and improved public health.

REFERENCES

1. Haththotuwa RN, Wijeyaratne CN, Senarath U. Chapter 1 - Worldwide epidemic of obesity. In: Mahmood TA, Arulkumaran S, Chervenak FA, editors. Obesity and Obstetrics (Second Edition) [Internet]. Elsevier; 2020 [cited 2023 Jun 14]. p. 3–8. Available from: https://www.sciencedirect.com/science/article/pii/ B9780128179215000011

2. Flegal KM, Carroll MD, Kuczmarski RJ, Johnson CL. Overweight and obesity in the United States: prevalence and trends, 1960–1994. Int J Obes. 1998 Jan;22(1):39–47.

3. Flegal KM, Carroll MD, Ogden CL, Johnson CL. Prevalence and Trends in Obesity Among US Adults, 1999-2000. JAMA. 2002 Oct 9;288(14):1723–7.

4. Li M, Gong W, Wang S, Li Z. Trends in body mass index, overweight and obesity among adults in the USA, the NHANES from 2003 to 2018: A repeat cross-sectional survey. BMJ Open. 2022 Dec;12(12):e065425.

5. Bleich S, Cutler D, Murray C, Adams A. Why is the Developed World Obese?

6. Bhurosy T, Jeewon R. Overweight and Obesity Epidemic in Developing Countries: A Problem with Diet, Physical Activity, or Socioeconomic Status? Sci World J. 2014;2014:964236.

7. Caballero B. The Global Epidemic of Obesity: An Overview. Epidemiol Rev. 2007 Jan 1;29(1):1–5.

8. Cham B, Scholes S, Ng Fat L, Badjie O, Groce NE, Mindell JS. The silent epidemic of obesity in The Gambia: evidence from a nationwide, population-based, cross-sectional health examination survey. BMJ Open. 2020 Jun;10(6):e033882.

9. Kinlen D, Cody D, O’Shea D. Complications of obesity. QJM Mon J Assoc Physicians. 2018 Jul 1;111(7):437–43.

10. Tremmel M, Gerdtham UG, Nilsson PM, Saha S. Economic Burden of Obesity: A Systematic Literature Review. Int J Environ Res Public Health. 2017 Apr;14(4):435.

11. Swinburn BA, Kraak VI, Allender S, Atkins VJ, Baker PI, Bogard JR, et al. The Global Syndemic of Obesity, Undernutrition, and Climate Change: The Lancet Commission report. The Lancet. 2019 Feb;393(10173):791–846.

12. Endocrine disruptors - ECHA [Internet]. [cited 2023 Jun 18]. Available from: https://echa.europa.eu/hot-topics/endocrine-disruptors

13. National Institute of Environmental Health Sciences [Internet]. [cited 2023 Jun 18]. Endocrine Disruptors Available from: https://www.niehs.nih.gov/health/topics/ agents/endocrine/index.cfm

14. TEDX - The Endocrine Disruption Exchange [Internet]. [cited 2023 Jun 18]. Search the TEDX List. Available from: https:// endocrinedisruption.org/interactive-tools/tedx-list-of-potential-endocrine-disruptors/search-the-tedx-list

15. Baillie-Hamilton PF. Chemical Toxins: A Hypothesis to Explain the Global Obesity Epidemic. J Altern Complement Med. 2002 Apr;8(2):185–92.

16. Heindel JJ. History of the Obesogen Field: Looking Back to Look Forward. Front Endocrinol [Internet]. 2019 [cited 2023 Jun 18];10. Available from: https://www.frontiersin.org/ articles/10.3389/fendo.2019.00014

17. Grün F, Blumberg B. Environmental Obesogens: Organotins and Endocrine Disruption via Nuclear Receptor Signaling Endocrinology. 2006 Jun 1;147(6):s50–5.

18. Kim S, Moustaid-Moussa N. Secretory, Endocrine and Autocrine/Paracrine Function of the Adipocyte. J Nutr. 2000 Dec 1;130(12):3110S-3115S.

19. Giordano A, Smorlesi A, Frontini A, Barbatelli G, Cinti S. MECHANISMS IN ENDOCRINOLOGY: White, brown and pink adipocytes: the extraordinary plasticity of the adipose organ [Internet]. 2014 [cited 2023 Jun 18]. Available from: https:// academic.oup.com/ejendo/article/170/5/R159/6661591

20. Luong Q, Huang J, Lee KY. Deciphering White Adipose Tissue Heterogeneity. Biology. 2019 Apr 11;8(2):23.

21. Longo M, Zatterale F, Naderi J, Parrillo L, Formisano P, Raciti GA, et al. Adipose Tissue Dysfunction as Determinant of Obesity-Associated Metabolic Complications. Int J Mol Sci. 2019 May 13;20(9):2358.

22. Schelling JR. The Contribution of Lipotoxicity to Diabetic Kidney Disease. Cells. 2022 Oct 14;11(20):3236.

23. Xu H, Barnes GT, Yang Q, Tan G, Yang D, Chou CJ, et al. Chronic inflammation in fat plays a crucial role in the development of obesity-related insulin resistance. J Clin Invest. 2003 Dec 15;112(12):1821–30.

24. Zhang Y, Proenca R, Maffei M, Barone M, Leopold L, Friedman JM. Positional cloning of the mouse obese gene and its human homologue. Nature. 1994 Dec;372(6505):425–32.

25. Park YM, Myers M, Vieira-Potter VJ. Adipose Tissue Inflammation and Metabolic Dysfunction: Role of Exercise. Mo Med. 2014;111(1):65–72.

26. Bai Y, Sun Q. Macrophage recruitment in obese adipose tissue. Obes Rev Off J Int Assoc Study Obes. 2015 Feb;16(2):127–36.

27. Surmi BK, Hasty AH. Macrophage infiltration into adipose tissue. Future Lipidol. 2008;3(5):545–56.

28. Kahn BB, Flier JS. Obesity and insulin resistance. J Clin Invest. 2000 Aug 15;106(4):473–81.

29. Hotamisligil GS. The role of TNFα and TNF receptors in obesity and insulin resistance. J Intern Med. 1999;245(6):621–5.

30. Sethi JK, Hotamisligil GS. Metabolic Messengers: tumour necrosis factor. Nat Metab. 2021 Oct;3(10):1302–12.

31. Stein CJ, Colditz GA. The Epidemic of Obesity. J Clin Endocrinol Metab. 2004 Jun 1;89(6):2522–5.

32. Centers for Disease Control and Prevention (CDC). Prevalence of overweight and obesity among adults with diagnosed diabetes-United States, 1988-1994 and 1999-2002. MMWR Morb Mortal Wkly Rep. 2004 Nov 19;53(45):1066–8.

33. Beyer J, Song Y, Tollefsen KE, Berge JA, Tveiten L, Helland A, et al. The ecotoxicology of marine tributyltin (TBT) hotspots: A review. Mar Environ Res. 2022 Jul 1;179:105689.

34. Van Kerk GJMD, Luijten JGA. Investigations on organo-tin compounds. III. The biocidal properties of organo-tin compounds. J Appl Chem. 1954;4(6):314–9.

35. Novotny L, Sharaf L, Abdel-Hamid ME, Brtko J. Stability studies of endocrine disrupting tributyltin and triphenyltin compounds in an artificial sea water model. Gen Physiol Biophys. 2018;37(01):93–9.

36. Gibbs PE, Bryan GW. Reproductive Failure in Populations of the Dog-Whelk, Nucella Lapillus , Caused by Imposex Induced by Tributyltin from Antifouling Paints. J Mar Biol Assoc U K. 1986 Nov;66(4):767–77.

37. Oehlmann J, Stroben E, Fioroni P. The Morphological Expression of Imposex in Nucella Lapillus (Linnaeus) (Gastropoda: Muricidae) [Internet]. 1991 [cited 2023 Jun 24]. Available from: https://academic.oup.com/mollus/ article-lookup/doi/10.1093/mollus/57.3.375

38. Shimasaki Y, Kitano T, Oshima Y, Inoue S, Imada N, Honjo T. Tributyltin causes masculinization in fish. Environ Toxicol Chem. 2003;22(1):141–4.

39. McAllister BG, Kime DE. Early life exposure to environmental levels of the aromatase inhibitor tributyltin causes masculinisation and irreversible sperm damage in zebrafish (Danio rerio). Aquat Toxicol. 2003 Nov 19;65(3):309–16.

40. Holtcamp W. Obesogens: An Environmental Link to Obesity. Environ Health Perspect. 2012 Feb;120(2):a62–8.

41. Lyssimachou A, Santos JG, André A, Soares J, Lima D, Guimarães L, et al. The Mammalian Obesogen Tributyltin Targets Hepatic Triglyceride Accumulation and the Transcriptional Regulation of Lipid Metabolism in the Liver and Brain of Zebrafish. PLoS ONE. 2015 Dec 3;10(12):e0143911.

42. Nishikawa J ichi, Mamiya S, Kanayama T, Nishikawa T, Shiraishi F, Horiguchi T. Involvement of the Retinoid X Receptor in the Development of Imposex Caused by Organotins in Gastropods. Environ Sci Technol. 2004 Dec 1;38(23):6271–6.

43. Mohajer N, Du CY, Checkcinco C, Blumberg B. Obesogens: How They Are Identified and Molecular Mechanisms Underlying Their Action. Front Endocrinol. 2021 Nov 25;12:780888.

44. Li X, Ycaza J, Blumberg B. The environmental obesogen tributyltin chloride acts via peroxisome proliferator activated receptor gamma to induce adipogenesis in murine 3T3-L1 preadipocytes. J Steroid Biochem Mol Biol. 2011 Oct;127(1–2):9–15.

45. Griffin MD, Pereira SR, DeBari MK, Abbott RD. Mechanisms of action, chemical characteristics, and model systems of obesogens. BMC Biomed Eng. 2020 Apr 30;2:6.

46. A2-33. English: Structure of the PPAR-gamma receptor bound to DNA [Internet]. 2012 [cited 2023 Jun 25]. Available from: https://commons.wikimedia.org/wiki/File:PPARG.png

47. Grün F, Watanabe H, Zamanian Z, Maeda L, Arima K, Cubacha R, et al. Endocrine-Disrupting Organotin Compounds Are Potent Inducers of Adipogenesis in Vertebrates. Mol Endocrinol. 2006 Sep 1;20(9):2141–55.

48. Dawson MI, Xia Z. The retinoid X receptors and their ligands [Internet]. 2012 [cited 2023 Jun 25]. Available from: https:// linkinghub.elsevier.com/retrieve/pii/S1388198111001843

49. Na J, Park BS, Jang D, Kim D, Tu TH, Ryu Y, et al. Distinct Firing Activities of the Hypothalamic Arcuate Nucleus Neurons to Appetite Hormones. Int J Mol Sci. 2022 Jan;23(5):2609.

50. Desai M, Ferrini MG, Han G, Jellyman JK, Ross MG. In Vivo Maternal and In Vitro BPA Exposure Effects on Hypothalamic Neurogenesis and Appetite Regulators. Environ Res. 2018 Jul;164:45–52.

51. Kumar A, Pareek V, Faiq MA, Ghosh SK, Kumari C. ADULT NEUROGENESIS IN HUMANS: A Review of Basic Concepts, History, Current Research, and Clinical Implications. Innov Clin Neurosci. 2019 May 1;16(5–6):30–7.

52. Goodman JE, Peterson MK. Bisphenol A. In: Wexler P, editor. Encyclopedia of Toxicology (Third Edition) [Internet]. Oxford: Academic Press; 2014 [cited 2023 Jun 26]. p. 514–8. Available from: https://www.sciencedirect.com/science/ article/pii/B9780123864543003663

53. Rubin BS, Schaeberle CM, Soto AM. The Case for BPA as an Obesogen: Contributors to the Controversy. Front Endocrinol. 2019 Feb 6;10:30.

54. Nutrition C for FS and A. Bisphenol A (BPA): Use in Food Contact Application. FDA [Internet]. 2023 Apr 20 [cited 2023 Jun 26]; Available from: https://www.fda.gov/food/ food-additives-petitions/bisphenol-bpa-use-food-contactapplication

55. Nakamura K, Itoh K, Yaoi T, Fujiwara Y, Sugimoto T, Fushiki S. Murine neocortical histogenesis is perturbed by prenatal exposure to low doses of Bisphenol A. J Neurosci Res. 2006 Nov 1;84(6):1197–205.

56. Nakamura K, Itoh K, Sugimoto T, Fushiki S. Prenatal exposure to bisphenol A affects adult murine neocortical structure. Neurosci Lett. 2007 Jun 13;420(2):100–5.

57. Nakamura K, Itoh K, Dai H, Han L, Wang X, Kato S, et al. Prenatal and lactational exposure to low-doses of bisphenol A alters adult mice behavior. Brain Dev. 2012 Jan;34(1):57–63.

58. Baggio LL, Drucker DJ. Biology of Incretins: GLP-1 and GIP. Gastroenterology. 2007 May;132(6):2131–57.

59. Egusquiza RJ, Blumberg B. Environmental Obesogens and Their Impact on Susceptibility to Obesity: New Mechanisms and Chemicals. Endocrinology. 2020 Mar 1;161(3):bqaa024.

60. Shannon M, Green B, Willars G, Wilson J, Matthews N, Lamb J, et al. The endocrine disrupting potential of monosodium glutamate (MSG) on secretion of the glucagon-like peptide-1 (GLP-1) gut hormone and GLP-1 receptor interaction. Toxicol Lett. 2017 Jan;265:97–105.

61. Heindel JJ, Blumberg B. Environmental Obesogens: Mechanisms and Controversies. Annu Rev Pharmacol Toxicol. 2019 Jan 6;59:89–106.

62. Chamorro-Garcia R, Blumberg B. Current Research Approaches and Challenges in the Obesogen Field. Front Endocrinol. 2019 Mar 22;10:167.

63. Behl M, Rao D, Aagaard K, Davidson TL, Levin ED, Slotkin TA, et al. Evaluation of the Association between Maternal Smoking, Childhood Obesity, and Metabolic Disorders: A National Toxicology Program Workshop Review. Environ Health Perspect. 2013 Feb;121(2):170–80.

64. Wang Y, Lobstein T. Worldwide trends in childhood overweight and obesity. Int J Pediatr Obes IJPO Off J Int Assoc Study Obes. 2006;1(1):11–25.

65. de Onis M, Blössner M, Borghi E. Global prevalence and trends of overweight and obesity among preschool children. Am J Clin Nutr. 2010 Nov;92(5):1257–64.

66. Avenue 677 Huntington, Boston, Ma 02115. Obesity Prevention Source. 2012 [cited 2023 Jun 26]. Child Obesity. Available from: https://www.hsph.harvard.edu/obesityprevention-source/obesity-trends-original/global-obesitytrends-in-children/

67. National Institutes of Health (NIH) [Internet]. 2015 [cited 2023 Jul 2]. New comprehensive view of the mouse genome finds many similarities and striking differences with human genome. Available from: https://www.nih.gov/news-events/ news-releases/new-comprehensive-view-mouse-genomefinds-many-similarities-striking-differences-human-genome

68. Howe K, Clark MD, Torroja CF, Torrance J, Berthelot C, Muffato M, et al. The zebrafish reference genome sequence and its relationship to the human genome. Nature. 2013 Apr 25;496(7446):498–503.

69. Lefterova MI, Haakonsson AK, Lazar MA, Mandrup S. PPARγ and the global map of adipogenesis and beyond. Trends Endocrinol Metab. 2014 Jun;25(6):293–302.

EXPLORING EMERGENT PROPERTIES OF COMPLEX SYSTEMS USING MACHINE LEARNING

STEM FINLAY SANDERS

This report was based on an Original Research in Science (ORIS) project in partnership with Imperial College London and was short-listed for the ILA/ORIS Presentation Evening.

INTRODUCTION

Many natural phenomena display properties or behaviours more than the mere aggregation of their parts. Humans, for instance, are capable of language, cog- nition and intricate social behaviours, none of which are properties of individual cells. Similarly, each cell’s functionality arises from the interactions be- tween molecules, even though none possess the cell’s capabilities independently. This pattern, where macroscopic properties arise from interactions between mi- croscopic components, termed

’emergence’, is a hallmark of complex systems. Emergence creates layers of abstraction within a system, where each behaves according to its own physical laws.

Formal theories of emergence have already been introduced using informa- tion theory, such as in5 The contribution of this paper is a novel method of identifying emergence using machine learning. By approximating the dynamics of a complex system at different spatiotemporal scales, I confirm numerically

that these layers of abstraction exist, and that the dynamics of each can be learned by a data-driven approach.

I evaluate this method using the Classical XY model, a lattice model of statistical mechanics relevant to phenomena such as the melting of crystals, magnetism and superconductivity, as an example. At the microscopic scale, the model consists of a collection of spins on a lattice that can point in any direction in the plane, which operate according to the dynamics of equation 4. At the macroscopic scale, the model is characterised by emergent structures termed ’vortices’ and ’anti-vortices’, which describe topological flaws where groups of spins make a 2π rotation either clockwise or anticlockwise, that follow Coulomb dynamics.

To this end, I propose a dual pathway approach to predicting the trajectories of spins and vortices using graph neural networks. First, I trained a model to predict spin dynamics, from which the vortices could be extracted. Second, I trained a model that bypasses spins, instead directly predicting vortex move- ments. By drawing parallels to commutativity diagrams, I demonstrate that both pathways converge to accurate vortex predictions, even over extended rollouts.

BACKGROUND

2.1 XY Model Simulation

the dynamics of spins and vortices can be learned. I chose to simulate the XY model with periodic boundary conditions (PBCs), meaning that spins on a given boundary are adjacent to spins at the opposite boundary. In essence, PBCs let a small system approximate an infinite one, as all spins have an equal amount of neighbours. Discussed below are two methods of simulating the XY model, the latter of which was ultimately chosen.

“ I propose a dual pathway

approach to predicting the trajectories of spins and vortices using graph neural networks.

Metropolis-Hastings Algorithm

The Metropolis-Hastings algorithm, a Markov Chain Monte Carlo method, is well suited for simulating the XY model. The term ‘Markov Chain’ indicates that the state of the XY model at time t + 1 is solely influenced by its state at time t. ‘Monte Carlo’ means that spins change according to a specified probability distribution. To this end, we assume the spins interact according to the energy given by the Hamiltonian (an equation for the sum of the system’s energy), as discussed in4:

where J is the coupling constant and j(i) denotes summation over each of the 4 nearest neighbours, j, of lattice site i. Intuitively, the interaction energy of a pair of spins i, j is minimised when θi = θj. The algorithm for updating a single spin proceeds as follows:

1. Pick a spin θi on the lattice

2. Change the angle of the selected spin by some small amount ∆θ to produce a new angle θi′

3. Calculate the energy change in the system due to this proposed change using the Hamiltonian:

4. Choose whether to accept the proposed change. If ∆E < 0 we accept the change because it lowers the energy. If ∆E > 0 we accept the change with probability:

By iterating over all spins on the lattice at time t, we reach a subsequent state of the system at time t + 1. Although reasonable, this method isn’t preferred due to the inconsistency of its spin updates over extended time periods.

Numerical Integration

Numerical integration, as detailed in2, produces predictable and therefore learnable, rollouts of the XY model by eliminating all randomness in spin up- dates. By altering the Hamiltonian above and

introducing a kinetic energy term, they derived the following equations of motion:

Intuitively, the right-hand side of this equation gives us the angular accelera- tion of each spin. To transition the state of an XY model from time t to t+1, we first update each spin’s angular velocity by adding the computed angular accel- erations. Then, we adjust each spin’s angle by adding their velocities. I found that slightly damping the added velocities led to more stable vortex movements and removed the possibility of vortices spontaneously appearing.

Final Intuitions

XY Model at time t XY model at time t + 10 Figure 1: An example of the XY model’sspins at times t and t + 10.

Figure 1 is an example of how the XY model evolves over time. Vortices and anti-vortices are marked with red and green dots respectively.

The most fundamental pattern of the system is that vortices of the same type repel and those of opposite type attract, although this is not always obvious. The first source of confusion is that the two vortices in the bottom right have disappeared, or ’annihilated’, in the t + 10 configuration. This happens when vortices inevitably meet, cancel out and allow the spins they previously distorted to become uniform. The next difficulty is that the vortices on the far left and right boundaries appear to have moved away from each other, despite my previous claim that they should meet and annihilate. This is caused by the previously discussed periodic boundary conditions, which allow the vortices to attract across boundaries.

2.2 Graph Neural Networks

Neural Networks (NNs), a pillar of machine learning, are proven to be capable of representing any continuous function given an appropriate

architecture1. In a trained network, input data is transformed according to internal parameters to produce useful output such as a prediction or classification. However, traditional feed-forward neural networks don’t inherently capture structured relationships between input data points, like the local interactions that are crucial to the XY model.

Graph Neural Networks (GNNs) are designed to work on data structured as graphs. These networks not only consider data points but also the relationships between them. Equipped with in-built assumptions that prioritise relationships (often termed as ’relational inductive biases’)3, GNNs are particularly adept at handling systems where interactions between individual components are paramount, making them especially suitable for approximating complex systems like the XY model.

Neural Networks

Neural networks are often visualised as interconnected layers of neurons. The type of traditional neural network I used to create graph neural networks is called a multi-layer perceptron, where all nodes in layer l are connected to those in layer l + 1. The first and last layers are the input and output layers and any intermediate layers are called hidden layers.

Figure 2: A multi-layer perceptron which takes 3 inputs, has 2 hidden layers with 4 neurons each and produces 2 outputs.

Each neuron has a set of parameters, specifically a weight for each of its inputs and a single bias, which transform its inputs from the previous layer into an output. We then apply a non-linear activation function which enables it to approximate non-linear functions. The operation can be summarised as:

The output of each neuron in layer l becomes one of the inputs to each neuron in layer l + 1. Repeating this until we get a result from the output layer is called forward propagation. To improve our network, we first need to quantify the difference between its outputs and the expected outputs using a loss function. For instance, the mean-squared error E, of neuron i’s output, yi, is:

Now we traverse backward through the network, using the chain rule to quantify each parameter’s contribution to the combined error, ET, with which they will be updated. For each neuron, if z is the input to its activation function and h is its activated output, then each weight, w’s, contribution is given by:

Once we have all the necessary derivatives, we update each neuron’s weights according to a learning rate, α. The learning rate determines the step size in the opposite direction of the gradient. Too large a value may cause the parameters to over-correct, missing their optimal values and too small a value might make the training process excessively slow. For example, we can update each weight, w, with this equation:

A similar process is repeated for computing the gradients for and updating biases. After many iterations of this 'training', the network will converge on ideal parameters for its task. In practice, we often 'mini-batch' inputs, or pass a group of them into the network simultaneously, to improve training stability and computational efficiency.

Graphs

A graph is a data structure defined by a set of nodes and edges. Nodes can be seen as individual entities, while edges depict interactions or relationships between these entities.

In Figure 3, nodes x1, x2 and x3 have associated features v1, v2 and v3, which define numerical or categorical properties. The edges also have features,

Figure 3: A graph with 3 nodes (x1,x 2, and x3) connected by an undirectededge e1 and a directed edge e2

e1 and e2, which might encompass the nature or strength of the relationshipbetween the nodes they connect. The arrowed or ’directed’ edge between v1 and v2 denotes a one-way influence. The edge between v2 and v1 is 'undirected', denoting mutual influence. Undirected edges are comprised of two directed edges in opposite directions, with the same feature. Graphs can also have global features,

representing holistic properties, but I have excluded them as they are not relevant to the methods in this paper. GNN Implementation Graph neural networks combine the concepts of traditional neural networks and graph theory. They take a graph as input and return an updated graph with the same topology, but different features, in a way that reflects the structure of the input graph. The method explained below is detailed in3.

We begin by updating the features of each edge. For each undirected edge with feature, ek, we use the update function ϕe to produce a new edge feature e ′ k.

Here, vrk is the receiver node’s feature, vsk is the sender node’s feature and u is the global feature. By sender and receiver, we mean the node from which the directed edge originates, and the node it points to. To update an undirected edge, we update each of its constituent directed edges and then average them to produce a single feature.

Next, we update the features of each node in a three-step process.First, for each node i, we collect into a group, E′ i, its own feature, ri, and the features of all of its incoming edges, sk, from each node, k.

We pass these groups into a neural network, giving us an updated embedding for each node that considers its incoming edges. Next, for each node i, we aggregate all of the embeddings of the nodes to which it is connected to produce a new embedding, ¯ e′ i, with the function ρe→v.

The aggregation function often computes the sum or mean of the features it takes as input. Finally, we update each node i’s feature, vi, with the update function ϕv using the embedding created in the previous equation.

Now we can reconstruct our graph. Each edge, k, now has the feature e ′ k,and each node i, now has the feature v ′ i. To create mini-batches of graphs we combine the constituent graphs into a larger, disconnected graph. This lets us use our original update and aggregation methods, no matter the mini-batch size.

METHOD

I approached the problem of predicting the trajectories of vortices using graph neural networks from 2 directions. First, I trained a GNN, the ’spin model’, that predicts the state of spins, S, at time t+1 from those at time t. Extracting vortices from these spins gives us the first set of predicted vortex trajectories. Second, I trained a GNN, the ’vortex model’, that predicts the state of vortices directly, V , at time t from those at time t, giving us another set of trajectories. The approach is summarised in Figure 4:

Figure 4: A commutative diagram where ψ is the function that extracts vortex positions from spins, and ϕs and ϕv are update functions approximated by the spin and node models respectively.

If both models’ predictions are accurate, all paths from St should yield the same result, Vt+1, and should be similar to the movements of vortices in the actual XY model simulation. I implemented both GNNs using the Pytorch Geometric (PyG) library for ease of use and efficient computation, and all of the accompanying code in Python.

3.1 Vortex Extraction

To find the positions of vortices in the XY model, we consider spins in square groups of four, termed ’plaquettes’ as shown in Figure 5.

5: An instance of the XY model, where a box outlines a plaquette.

We define the circulation around a plaquette as the sum of the phase differences along its perimeter. The circulation around a plaquette is +2π if there is a vortex and −2π if there is an anti-vortex.

3.2 Spin Model

The spin model takes as input the spins of the XY model at time t and outputs a prediction of their state at time t + 1.

Spin Graph Representation

In the XY model, spins only interact with their four nearest neighbours, so can be represented by a graph like in Figure 6.

In this graph, each spin is a node connected by undirected edges to its neighbours. Crucially, there are edges connecting nodes that are neighbours across boundaries, such as θ5 and θ9, even though they are not shown. Each spin, i, with angle, θi, and angular velocity, ωi, has a feature, vi, where:

Including both sin(θi) and cos(θi) accounts for the periodicity of angles. I found that no global or edge features were necessary for the model to make accurate predictions, so have discluded them.

Figure 6: A graph representation of the spins of an instance of the XY model of size 5x5. It is important to note that the XY model simulated for data generation was size 20x20, and that its graph representation follows the same pattern of connectivity.

Data Preparation

Data for the dataset was obtained by simulating many instances of the XY model, each initialized randomly, using numerical integration. For each instance of the XY model, I created a trajectory, or sequence of graphs like the one above, that captures the state of the XY model’s spins at consecutive time steps. If St is the graph representing the state of the spins at time t, then each trajectory, T, over n time steps can be expressed as:

Model Architecture

The spin model is a GNN, constructed using a single instance of the Metalayer class provided by PyG. The absence of edge and global features simplifies this model to a single node update. The forward pass of the spin model consists of the following steps:

1. Aggregate Node Features: For each node, we compute the sum of its neighbors’ features.

2. Update Node Features: We pass each node’s features and its neighbourhood feature into a multi-layer perceptron, to obtain new node features.

Figure

Model Training

I used an iterative refinement approach to train the spin model. Specifically, I compared small rollouts of length k produced by the spin model to real sequences of spin graphs from the XY model. To obtain a real sequence of spin graphs I took a subset, Treal, of a spin trajectory from the dataset of length k

Then, we produce a predicted spin sequence, Tpredicted, by applying the spin model k − 1 times to St

By summing the mean-squared error of each pair of predicted and real spins at each timestep, we obtain a total loss which we can propagate back through the model. This was done in mini-batches of size 5, using the Adam Optimiser with a learning rate of 0.001, and a k of 8.

3.3 Vortex Model

The vortex model takes as input the positions of vortices of the XY model at time t and outputs a prediction of their state at time t + 1.

Vortex Graph Representation

In the absence of clear relationships between vortices, I assumed that they all exert some influence on each other. To this end, I represented them as nodes in a fully connected graph. Figure 7 shows two vortices, V1 and V2, and two anti-vortices A1 and A2

Figure 7: A fully connected graph representing the vortices of an instance of the XY model.

Each vortex’s position (x, y) in an instance of the XY model of dimensions L by L, is first encoded to position (x′, y′) as follows:

This effectively converts the position into an angle, enabling the use of trigonometric functions to account for periodic boundary conditions. Then, a feature, vi, is constructed from the position (x′, y′) and the vortex’s type n (1 or −1 for vortices and anti-vortices respectively).

Data Preparation

The dataset for the vortex model shares many similarities with that of the spin model. Using numerical integration, I simulated instances of the XY model over various time steps. These simulations produced many vortex trajectories, T, represented as sequences of graphs, V.

this, I first paired any remaining vortices with their corresponding unpaired anti-vortices using linear sum assignment. Then, I found the midpoint of each pair and appended this to the trajectory of both vortices in the pair. Without this extra step, the GNN would have no data that captures vortex dynamics when very close together.

The final difficulty was tackling the discrete nature of vortex extraction, which led to erratic vortex trajectories. I chose to fit a bezier curve to each vortex’s set of points, and then sample it to obtain a smoother set of points. However, fitting a curve to vortex trajectories that include periodic boundary conditions requires some extra steps. I 'unwrapped' each trajectory, equivalent to expanding the area in which vortices can move, to ensure they didn’t traverse any boundaries.

I then fit a bezier curve to the unwrapped trajectory using the equation:

Given m such trajectories, our dataset becomes:

However, this process posed some unique challenges. The first challenge was tracking vortices through time. Extracting vortices from the XY model at a time t yields lists of both vortex and anti-vortex positions.

However, these lists are unordered. To track each vortex over time, we need to pair positions at time t with those at time t + 1. This is an ’assignment problem’, and requires us to pair positions in a way that minimises the sum of pairwise distances. I overcame this using the SciPy library’s linear sum assignment function.

The next nuance that needed to be handled was the annihilation of vortices. When vortices annihilate, the list of positions at time t outnumbers that at time t + 1, leading to unpaired vortices. To manage

Where B(t) is the position on the Bezier curve for parameter t in the range [0, 1], n is the degree of the Bezier curve (equal to the number of time steps in the trajectory we are smoothing) and Pi is the i-th control point (the positions of the vortex at each time step). The result of this is a smooth unwrapped trajectory, which can be re-wrapped to produce a smooth trajectory that considers periodic boundary conditions. The example below shows all these steps using an example trajectory of a single vortex, where the colour gradient reflects the vortex’s position over time.

Raw Trajectory

Smooth Unwrapped Trajectory

Unwrapped Trajectory

Final Trajectory

Figure 8: A visualisation of the steps taken in vortex trajectory smoothing. The top left shows the actual trajectory of the vortex. The top right shows the ’unwrapped’ trajectory. Next is the bottom left, where the unwrapped trajectory has been smoothed. Finally, the bottom right shows the smoothed trajectory of the vortex taking into account periodic boundary conditions.

Repeating this process for all trajectories creates a more suitable dataset 12 with which the vortex model can be trained.

Model Architecture

The vortex model is identical to the spin model, except in the aggregation step I chose to average the features of each node’s neighbours.

Model Training

Training an accurate vortex model proved more simple the the spin model, requiring only paired graphs, Vt, Vt+1, representing vortex states at consecutive time steps. This more straight forward approach is equivalent to the iterative refinement approach with k restricted to 2. The loss was calculated as the meansquared error between the model’s prediction and the real state of Vt+1. This was done in mini-batches of size 5, using the Adam Optimiser at a learning rate of 0.001.

RESULTS

Both models were able to produce physically plausible vortex trajectories over long rollouts, even with small network sizes. However, the spin model required an iterative refinement technique to have a similar performance to the vortex model. Without this assistance, its predictions quickly devolved into chaos. One example of this is shown below:

Real Spins

Predicted Spins

Figure 9: An example showing the instability of predicted spins (left) without the iterative refinement approach.

It is therefore important to consider that all of the following comparisons compare the spin model with an iterative refinement approach to the vortex model with its more simple approach. First, I conducted a brief hyperparameter search to identify the optimal network sizes, shown in 10, where darker colours denote a lower minimum loss over 50 epochs.

Grid searches for the spin and vortex models with smaller datasets to identify optimal network sizes.

I decided to use two hidden layers of size 16 for the training of both models with larger datasets to produce the following results.

10 epochs.

Model Predictions

Both models performed best when predicting the dynamics of more stable instances of the XY model, such as in Figure 13, where spins are most uniform. In each example below, the initial configuration is given to each model, and their predictions are compared to the real XY model.

When predicting the evolution of more chaotic instances of the XY model, or those that require longer time periods to reach equilibrium, the predictions become less accurate. The vortex model especially is sensitive to the number of vortices it has to predict, likely because the majority of its training data included 2 or 4 vortices. Even though no longer identical to the real model, the predictions of both the spin and vortex model still display physically plausible dynamics, with vortices attracting and annihilating, as shown in Figure 14.

Figure 10:

Figure 11: Spin Model Training Curve over

Figure 12: Vortex Model Training Curve over 10 epochs.

Spin Model Vortex Model

Initial Configuration

the

Real XYModel at t = 5

Spin Model at t = 5

Vortex Model at t = 5

Real XY Model at t = 15

Spin Model at t = 15

Vortex Model at t = 15

Real XY Model at t = 10

Spin Model at t = 10

Vortex Model at t = 10

Real XY Model at t = 20

Spin Model at t = 20

Vortex Model at t = 20

Figure 13: An example of

predictions of each model compared to the real XY model.

Initial Configuration

XY Model at t = 10

Real XY Model at t = 20

XY Model at t = 30

XY Model at t = 70

Spin Model at t = 10

Spin Model at t = 20

Spin Model at t = 30

Spin Model at t = 70

Figure 14: Another example of the predictions of each model compared to the real XY model. Here, both models produce very accurate vortex trajectories until around t = 15, after which they begin to deviate. Despite this, the trajectories remain physically

Model at t = 10

Vortex Model at t = 20

Vortex Model at t = 30

Vortex Model at t = 70

plausible and reach uniformity by t = 70.

Real

Vortex

Real

Real

Figure 15: A plot showing the mean spin error in Figure 13 between the real XY model, and the predictions of the spin model and an identity model.

CONCLUSION

Even though the error of both models is non-zero, they both predict physically plausible vortex trajectories, even over long rollouts. This confirms numerically that the spins and vortices of the classical XY model exist as different layers of abstraction, and that the dynamics both can be approximated in a data-driven way.

The ability of GNNs to approximate the dynamics of the XY model, even with small architectures, shows their promise as tools for studying emergence in more intricate systems. Future research could use similar techniques to investigate brain data to see if neurons follow different dynamics from thoughts, offering new insights into the emergence of consciousness.

16: A plot showing the mean spin error in Figure 14

between the real XY model, and the predictions of the spin model and an identity model.

“

Future research could use similar techniques to investigate brain data to see if neurons follow different dynamics from thoughts, offering new insights into the emergence of consciousness.

Figure

REFERENCES

1. Kurt Hornik, Maxwell Stionchcombe, and Halbert White. Multilayer feedforward networks are universal approximators. Neural Networks, 2(5):359–366, 1989. https://cognitivemedium.com/magic_paper/index.html

2. Xavier Leoncini, Alberto D. Verga, and Stefano Ruffo. Hamiltonian dynamics and the phase transition of the XY model. Phys. Rev. E, 57(6):6377–6389,1998. https://link.aps. org/doi/10.1103/PhysRevE.57.6377

3. Peter W. Battaglia, Jessica B. Hamrick, Victor Bapst et al. Relational inductive biases, deep learning, and graph networks. arXiv preprint arXiv:1806.01261, 2018. https:// arxiv.org/pdf/1806.01261.pdf

4. H. Jensen. Complexity Science: The Study of Emergence. Cambridge: Cambridge University Press, 2022, p. 144. https://www.cambridge.org/highereducation/books/ complexity-science/E0761D26BDAB25D75C6AB868AECE2F2 D#overview

5. Rosas FE, Mediano PAM, Jensen HJ, Seth AK, Barrett AB, Carhart-Harris RL, et al. Reconciling emergences: An information-theoretic approach to identify causal emergence in multivariate data. PLoS Comput Biol, 2020, 16(12): e1008289.https://journals.plos.org/ploscompbiol/ article?id=10.1371/journal.pcbi.1008289

IT’S A HIGH TIME FOR HYPERSONIC MISSILES

Their impact on geopolitics and the end of mutual assured destruction

STEM BEN TABBERNER

This Independent Learning Assignment (ILA) was short-listed for the ILA/ORIS Presentation Evening.

“It’s a high time for Hypersonic Missiles”1 is a line from one of Sam Fender’s best-selling songs. Looking at the line in the context of the song, it is a plea to society to do something meaningful with their lives because hypersonic missiles have the potential to destroy life as we know it. For example, a hypersonic missile launched from Moscow now could strike London with a nuclear warhead in under 10 minutes. Such missiles are disrupting the relative

security that the world has felt since the end of the Cold War. Since ex-United States Secretary of Defence Robert McNamara's suggestion of Mutual Assured Destruction (MAD) in the 1960s the world has been living in relative comfort in the belief that no superpower will strike another for fear of ‘Mutual Destruction’. Hypersonic missiles may allow any country who has them to make a nuclear strike with more success. Mutual Assured Destruction

relies on different countries’ military forces being equal - hypersonic missiles disrupt this balance because of the reduced response time and difficulty of defending against them. They have put the world in a precarious position; the ‘Destruction’ in Mutual Assured Destruction may not be ‘Mutual’ at all. Is it therefore time to look for a new way forward which will give the world a new era of relative peace?

To give a little more background, a hypersonic missile is a missile that travels at above Mach 5, or 1,700 metres in a single second. Using this definition, the existing Intercontinental Ballistic Missiles (ICBMs) are also hypersonic missiles because they can reach speeds of Mach 20. However, the hypersonic missiles this essay refers to are different because they have increased manoeuvrability and follow a different trajectory to ICBMs, making them much harder to track and ultimately defend against. They fly within the atmosphere manoeuvring to avoid detection and strike their target. The current detection systems that are in place are searching for the arc of an ICBM, not the complex trajectory of a new hypersonic missile. The new capabilities have provided fresh engineering challenges and so it is not known which countries possess hypersonic missiles.

There are many engineering challenges that need to be overcome to produce these missiles. For example, the method of propulsion must overcome incredible resistive forces of drag. The drag force is proportional to the square of the velocity and is governed by the equation:

Where ρ is the density of air, Cd is the coefficient of drag for a particular projectile and A is the cross-sectional area of that projectile. Using this formula, the drag force on a current supersonic missile can be calculated if it was travelling at hypersonic speeds. ρ was taken to be 1.204 kg/ m3, the value of the air density in the atmosphere where hypersonic missiles are likely to operate. The cross-sectional area of an MGM-140 ATACMS missile (a supersonic missile operated by the US military and their allies) was used. A graph was then plotted showing Drag Force in Newtons against Mach number (Figure 1).

This graph shows that as the Mach number increases, the drag force increases by the square of the velocity increase. If this missile was travelling at Mach 15 (which it cannot), the drag force would be 225 times higher than if it was travelling at Mach 1.

According to Newton's Second Law:

The sum of all forces is equal to the mass of the object times the acceleration. So, the thrust must be higher than the drag force to get the object to accelerate. To allow the missile to travel at Mach 15 it must overcome forces of over 800kN. This is not possible with current methods of propulsion; a jet engine can only reach up to a maximum of Mach 3. One solution is a supersonic combustion ramjet engine (scramjet). A scramjet engine works by removing the moving parts of a jet engine and combusting the air at high pressure, temperature, and supersonic speeds. This is achieved by using a gradual diffuser to increase the pressure and temperature without slowing the speed down too much. This high energy combustion reaction then allows a high velocity stream of air to propel the missile forward. The development of scramjet technology has unlocked a new realm of speed for missiles, and due to these difficulties in manufacturing hypersonic missiles, we do not know who has these missiles and what stages of development they may be in. This causes more instability on the global stage.

Given the additional issues in defending hypersonic missiles it is inevitable that their emergence will have an impact on war and the geopolitical stage.

“ Assured Destruction may not be ‘Mutual’ at all. Is it therefore time to look for a new way forward which will give the world a new era of relative peace?

The situation can be likened to the emergence of nuclear weapons and the Cold War. When the United States dropped an atomic bomb on Japan in 1945, the world awoke to the devastating nature of them and a race to create the best warheads and missiles followed. This created a period of incredible tension between the East and West, the Cold War. By the 1960s both sides had missiles that were capable of carrying nuclear warheads at 7km/s. The adoption of McNamara’s proposal of 'Mutual Assured Destruction' provided world leaders with a way forward that led to a period of relative peace enabled by progress in the detection systems of ICBMs, albeit still punctuated with periods of huge geopolitical tension.

When thinking about the geopolitical landscape, it is interesting to note the correlation between political tensions and defence spending as a

percentage of GDP; hence the impact that ICBMs had and now the impact hypersonic missiles will have on the geopolitical landscape. For example, the Cold War period of high political tension is reflected in the high defence spending by the US as a proportion of GDP. The spending by the USSR would most likely have been equally high. The period after the year 2000 was more peaceful with nuclear engagement being less of a threat. Again, this is reflected by the decrease in military spending.

2

example of this is the UK’s pledge to invest £6.6bn into ‘novel military research’5. These statistics are a clear indication of the beginning of a new arms race and a period of heightened political tension.

The graph above stops in 2020 and therefore does not show the recent increased spending by miliary powers in relation to the new hypersonic missiles. However, research shows that the United States Department of Defence (DoD) spending in 2023 increased by $50 billion (bn) from 20192. In a statement from the DoD, within the 2024 fiscal year budget request, $29.8bn is set aside for ‘missile defeat and defence programs’3 and $11bn is set aside ‘to deliver a mix of hypersonic and long-range subsonic missiles’3, $1.2bn of this is a contract with Lockheed Martion to integrate hypersonic missiles into the US Navy’s new Zumwalt-class destroyer. Additionally, according to The Telegraph, NATO spending was up 2.2% in 2022 and yet the head of NATO, Jens Stoltenberg, has "urged member states to up their defence spending"4. It is therefore clear to see that not only is there a general increase in spending but as seen by the DoD’s statement, there is a specific focus on hypersonic missiles. Another

For example, war is brewing in the Far East, where China is flexing its muscles specifically in relation to Taiwan. The US has pledged support to Taiwan if China invades (Taiwan Relations Act of 1979) partly because of the Taiwanese domination of the microchip market: they produce 60% of all microchips and over 90% of the most advanced ones according to The Economist6. These microchips are in everything we use, from microwaves to computers to hypersonic missiles. They are the backbone of the technological age, and the US does not want China to gain control of their production. The recent military drills that China has been conducting in and around Taiwan’s airspace show that China is serious in its aim to take back Taiwan. This is no surprise as there is legislation in China called the 'Anti-secession Law'7, which states that Taiwan is a part of China7. Furthermore, the legislation states that China can employ non-peaceful means7 to reunify the two countries and it is the sacred duty of all Chinese people7 to do that. This is relevant to the topic of hypersonic missiles because China already possesses missiles that apparently have the speed and improved manoeuvrability of hypersonic missiles (the DF-17) and can travel above Mach 5; the DF-17 missile is a hypersonic glide vehicle with a range of up to 1,800 kilometres with manoeuvrability which therefore pose a huge threat to Taiwan and could destroy any defending US ships. This could lead to direct war between two superpowers as opposed to the proxy wars that have been fought in the middle East between the US and Russia. That said, as the exact capabilities of the DF-17 have not been openly demonstrated it is not possible to say if this missile is a game changer. This ‘unknown’ has been highlighted recently in Ukraine when Putin’s, in his words “undefeatable”8, Kinzhal hypersonic missile was shot down by the PAC3 MSE Patriot air defence system. Taiwan has these defence systems and, although the DF-17 is different and most likely more effective

Figure

than the Kinzhal, the Patriot systems are helpful in deterring China. Even so, China's production of the DF-17 and the threat this brings is one of the key reasons for the high tensions around the South China Sea.

Another potential flashpoint due to hypersonic missiles is the Middle East. Israeli and Iranian tensions have been rising dramatically over the last few years and in May 2023 Israel threatened Iran with a pre-emptive strike if Iran did not close down its nuclear programme9. It has also been reported recently that Iran has successfully tested an indigenous hypersonic missile10. According to Iranian General Amir Ali Hajizadeh10, this missile is capable of striking Israel and travels at speeds of Mach 12-13 whilst manoeuvring to beat their air defences. The Iranian production of hypersonic missiles has worsened already fragile relations with Israel and is a constant threat in this part of the world.

nuclear war but Vasili Arkhipov’s rational decision stopped a Russian submarine firing a nuclear torpedo. Today’s hypersonic missiles, with their superior technology, makes the threat of nuclear destruction ever more real as Mutual Assured Destruction is no longer guaranteed. Therefore, as the world enters a new era of increased tension a new geopolitical mechanism is needed to bring some stability to the world.

“ Today’s hypersonic missiles, with their superior technology, makes the threat of nuclear destruction ever more real as Mutual Assured Destruction is no longer guaranteed.

As has been mentioned, the acceptance of Mutual Assured Destruction led to a period of relative stability, however there were still moments of high tension within this period, and examples of moments where a single person’s decision stopped a nuclear war. For example, The Cuban Missile Crisis of 1962 which brought the US and USSR terrifyingly close to

The concept of Mutual Assured Security could provide this comfort. Mutual Assured Security is a doctrine based on the defence of missiles rather than Mutual Assured Destruction’s security created by the threat of offensive weapons. The concept of Mutual Assured Security is that no country will attack another if every country has effective defensive systems which can intercept hypersonic missiles. This is much safer than Mutual Assured Destruction for two reasons. Firstly, the defences of any nation should be able to defend against a nuclear attack with hypersonic missiles and secondly there would be less incentive to attack when every country is transparent about their military capabilities. Mutual Assured Security would rely on international treaties and inspections which would lead to the depletion of offensive systems stock and the instead

increased funding of defensive systems. However, the road to reaching Mutual Assured Security is very long and would require many international treaties and a completely new outlook on war and national defence, not to mention the technological progress that would need to be made. This new outlook would propose the idea that all countries are allowed to have a small number of offensive nuclear weapons, but they would be able to have much larger amounts of defensive systems. Lots of investment would be needed to produce these defensive systems for hypersonic missiles and other potential threats.

Hypersonic missiles could be defended in a variety of ways, notably a rail gun or lasers. However, the first technology that could be used is an invention that came at the end of the Second World War. In anti-aircraft gun rounds engineers places a radar transmitter and a receiver. When the transmitted wave is reflected by a target plane, hence being in close proximity, the round would explode and damage the aircraft. This could be extended to take down hypersonic missiles. A round like this could be fired towards an incoming missile until the received signal is strong enough to cause an explosion. This could be improved by firing this round out of a rail gun. A rail gun could accelerate a projectile up to speeds of Mach 20 using the electromagnetic force. A magnetic field is created inside the barrel and the magnetic round is propelled out of it at incredibly high speeds. If this round was equipped with a radar system that is described above, it could be effective at taking down missiles. The incredibly high speeds of the round would mean there is less position change

that would need to be accounted for in the aiming.

However, the most promising avenue for defending against hypersonic missiles is lasers. Lasers are beams of light and so travel at the speed of light and the distances involved are so negligible in this case that we can treat this as instantaneous. The laser transmits a huge amount of energy to the missile causing it to either completely vaporise or ablate, either way the missile is completely destroyed or disabled. Rafael, the state funded defence company of Israel, are already producing a laser defence system called the Iron Beam however lots more work is required to make the system reliable. All these systems do however require the detection and tracking of hypersonic missiles. This is not possible with the current infrastructure as the essay mentions before. This could be fixed with the increased funding that would come from the funding that is normally directed to offensive weapons. Mutual Assured Security would require no increase in funding, instead a change in view from the leaders of the world and with technologies like the ones I have discussed these leaders could definitely achieve it.

“The tensions of the world are rising higher”1 and the only way to stop the uncertainty is Mutual Assured Security. The Cold War near misses would be negated, the tensions and the killing capabilities of countries would be reduced. Mutual Assured Security is the way to secure the future of our planet; imagine if it allowed us to instead focus on the bigger issues and solve other potential existential crises.

ILA written September 2024

BIBLIOGRAPHY

1. Genius.com, 05 March 2019, Hypersonic Missiles, https:// genius.com/Sam-fender-hypersonic-missiles-lyrics, 20 June 2023.

2. United States of America Department of Defence, 13 March 2023, https://www.defense.gov/News/News-Stories/Article/ Article/3327644/lethal-resilient-agile-joint-force-basis-fordefense-budget-request/#:~:text=President%20Joe%20 Biden's%20national%20defense,the%20fiscal%202023%20enacted%20budget., 04 June 2023.

3. United States of America Department of Defence, 13 March 2023, https://www.defense.gov/News/News-Stories/Article/ Article/3327644/lethal-resilient-agile-joint-force-basis-fordefense-budget-request/tt:~:text=President%20Joe%20 Biden'sti20national%20defense,the%20fiscal%202023%20 enacted%20budget., 04 June 2023.

4. The Telegraph, 21 March 2023, https://www.telegraph. co.uk/world-news/2023/03/21/nato-members-urged-bolster-defence-spending-dangerous-world/, 04 June 2023.

5. Dominic Nicholls, 17 March 2021, Britain pours billions into race to develop world’s fastest missile, https://www. telegraph.co.uk/news/2021/03/17/exclusive-britain-poursbillions-race-develop-worlds-fastest/, 01 June 2023.

6. The Economist, 06 March 2023, https://www.economist. com/special-report/2023/03/06/taiwans-dominance-of-thechip-industry-makes-it-more-important, 02 June 2023.

7. 2023 Mainland Affairs Council, 29 March 2005, General Policy Archives (1994-2008), https://www.mac.gov.tw/en/ News_Content.aspx?n=8A319E37A32E01EA&sms=2413CF E1BCE87E0E&s=D1B0D66D5788F2DE#:~:text=1.,%5BArticl es%202%20and%204%5D., 01 June 2023.

8. Sean Bell, 20 May 2023, Vladimir Pution’s ‘undefeatable’ missile was a costly illusion – the West can’t make the same mistake, https://news.sky.com/story/vladimir-putions-undefeatable-missile-was-a-costly-illusion-the-west-cant-makethe-same-mistake-12884572, 01 June 2023.

9. Mohammad Al-Kassim, 25 May 2023, https://www.jpost. com/middle-east/article-744211, 04 June 2023.

10. Inder Singh Bisht, 1 June 2023, https://www.thedefensepost. com/2023/06/01/iran-tested-hypersonic-missile/, 04 June 2023 Figure 2 World Bank, https://data.worldbank.org/ indicator/MS.MIL.XPND.GD.ZS?locations=US, 21 June 2023.

“

As a globe, we are entering a dangerous period of not being covered by Mutual Assured Destruction. “The tensions of the world are rising higher” and the only way to stop the uncertainty is Mutual Assured Security. The Cold War near misses would be negated, the tensions and the killing capabilities of countries would be reduced. Mutual Assured Security is the way to secure the future of our planet and allow us to solve other potential existential crises.

HUMANITIES ARTS &

OLIVER GODKIN

JOSHUA INGLESFIELD HARIS MACNICOL Tom McClean Highly Commended Highly Commended Short-Listed Short-Listed

JASPER MORRIS XAVIER WALLIN Arts/Humanities Winner Short-Listed

SMART CITIES

The Future of Mankind?

ARTS HUMANITIES

OLIVER GODKIN

This Independent Learning Assignment (ILA) was highly commended at the ILA/ ORIS Presentation Evening.

INTRODUCTION

We’ve all heard the word smart added as a prefix to everything nowadays - smart speaker, smart businesses, smart plugs, smart air purifiers, and you can even buy a smart fridge.

There are also Smart Cities

Smart cities are a relatively recent breakthrough in urban planning and design, and for the past few decades, have seen proponents far and wide from individual planners to governmental organisations, global tech firms, and billionaires.

With the current global population sitting at 8,036,344,468 at the time of writing, and with 4.4 billion people (56% of global population) living in urban areas as of April 2023, cities are at the forefront of tackling global issues, as a majority of the world’s population lives in urban areas. With rural to urban migration showing no signs of slowing, and with widening inequality in cities worldwide, the city of tomorrow is becoming more and more of a concern, arguably an issue, with which we need to

deal with now, lest we forsake our descendants, and leave them a depleted, uninhabitable, planet.1,2

Smart cities are one of the foremost defences against modern global issues, however currently they remain largely a theoretical concept, often confused with other urban design principles, and hence their application in the real world has been varied in success and support.

(Credit: IBM)

DEFINITION

This is partly because there is no clear definition of what a Smart City actually is. Multiple definitions have been put forward over the years, all revolving mostly around the ideas of connecting the physical, digital, and business infrastructures of a city with the aim of improving the efficiency of city services3 .

IBM - a leading innovator in the field of the development of smart cities worldwide - defines a smart city as “one that uses technology to transform its core systems and optimise the return from largely finite resources. By using resources in a smarter way, it will also boost innovation, a key factor underpinning competitiveness and economic growth.”4

However, recently, the understanding of what a smart city is has been confused with the emergence of related terms - intelligent cities, digital cities, and green cities.

Green Cities

Green cities are becoming more and more widespread across the world, as the threat of climate change is prompting many to seek less ecologically damaging methods to achieve a harmony of the built and natural environment. Green cities are defined as “urban enclaves whose design, construction, and operation prioritises the preservation of the natural world alongside the economic, social, and physical health and wellness of the city’s inhabitants”.5

East London’s own Olympic Park, with its green roofed apartment blocks, vast open spaces, and eco-friendly bike infrastructure comes to mind when thinking of examples of green cities, along with Curitiba in Brazil - the birthplace of the BRT (Bus Rapid Transit) system, that allows for integrated bus travel across the city, thus greatly reducing congestion and emissions.

Many cities worldwide such as Dubai in the UAE brand themselves as 'green cities' using it as a buzzword to generate interest when in actuality, under the surface of fancy CGI renders and sleek architecture Dubai and many other cities fall short of their idealised vision of a green-utopia - the Burj Khalifa for example lacks a connection to the sewage system - instead relying on a fleet of polluting trucks to relocate the 15 tonnes of sewage a day.

Intelligent Cities

Intelligent cities are the most closely related term to smart cities, and from an outside perspective may appear to describe the same concept - however they differ in key aspects. Intelligent cities are “the technological computing and management network required for the smooth urbanization of physical urban components, residents, and other city participants.”6

Intelligent cities, when compared to the definitions laid out for smart cities, generally overlook citizen-centric aspects such as cultural features and citizen knowledge.

Digital Cities

In a similar vein, the term ‘digital city’ is often used to refer to smart cities too, due to it having almost the same criteria in its definition - “a developing city model of urban technology related to the creation of cultural, social, economic, and educational data related to citizens.” However, the digital city is remiss in placing sustainability as part of its criteria in the definition.6

My Definition

I have decided to lay down my own definition of a smart city, which combines both the digital city and the environmental policies of green cities along with sustainability - as defined by the United Nations Brundtland Commission as meeting the needs of the present without compromising the ability of future generations to meet their own needs.7

Thus, smart cities are cities that merge the physical (built) environment with the natural environment, to further the connection with the social and economic environments of the city, through transformative technologies such as IoT (Internet of Things), applied to all aspects of city life, with the aim of creating a more economically and environmentally sustainable fabric for urban human habitation and development.

So, What Does a Smart City Actually Look Like, and What Does it Mean for Residents?

Examples - Santander and Oslo

Currently, intelligent cities are the most common version of smart cities found worldwide. Santander, for example, has 12,000 sensors already installed that provide a range of data, fed to a central control hub, that can change variables within the city, to allow for an increase in the quality of living in the city centre.8

Sensors embedded in the tarmac of parking lots across the city detect available parking spaces, and thus, through IoT technology, this data is transmitted to large panels located by the city’s main intersections, and thus can direct motorists to the nearest available parking space. On average, 30% of driving time is spent looking for a place to park, and hence as cars produce 4.6 metric tons of CO₂ per year, if this 30% was reduced through these parking sensors, then emissions could be cut down to almost 3.3 metric tons of CO₂ per year.9, 10, 11

Furthermore, sensors are also placed in rubbish bins across the city that can notify waste disposal services when these bins are full, and hence can help to decrease littering, and prevent the piling up of rubbish on the streets.10

Figure 1 - A Map showing IoT Infrastructure around Santander(Credit: Smart Santander)

Sensors placed in streetlights also feed pedestrian and traffic volume data back to the central hub, which can then dim or brighten the streetlights depending on the need. This use of sensors to create a smart streetlight system was first pioneered in Oslo in 2004, and the systems were found to have the effect of reducing energy used by streetlights by 70%, the equivalent of 1440 tons of CO₂ emissions per year for 10,000 streetlights, or 144kg of CO₂ per year per lamp, if all the energy for the streetlights was generated through fossil fuel means.

As it costs between $0.05/kWh to over $0.15/ kWh to generate electricity from oil, and the energy savings are estimated at around 5 gigawatt hours per year (which is equal to 5,000,000 kWh per year), the cost saved for generating electricity ranges between $250,000 to $750,000 every year, for an implementation cost of $16 million in Oslo. This means that, aside from the obvious ecological advantages, the scheme repays in terms of cost saved in 32 years, with each individual streetlight saving city authorities between $25 and $75 every year.10, 12, 13

With calls from the UN to reduce 'Vampiric overconsumption' of water worldwide, overconsumption of water is another area that smart

cities can help to tackle. Santander has implemented 50 different sensors in several parks across the city to help monitor moisture in the soil, along with humidity and temperature of the air, and by sending this data to the central hub in Santander, algorithms can take this data and switch sprinklers in these parks on or off depending on the need, and thus water costs can be cut by 50%.14, 9, 10

These design choices adopted by Santander and Oslo embody the technological IoT aspects of intelligent cities, but also account for sustainability, as these technologies such as the smart streetlamps ensure that the needs of the present - lighting the streets - are met without compromising the ability of future generations to meet their own needs – in this case, the reduction in CO₂ emissions allows for future generations to still have a prosperous and habitable planet in the future - thus merging the built and natural environment, with an environmentally sustainable method, ergo achieving the criteria of a smart city.

So, are smart cities effective as a sustainable solution to the future of urban development? Certainly, the spread of the idea of a smart city worldwide appears to back this view in that if these

Figure 2: LED Streetlights in Oslo – (Credit: AEC and the partners of the city of Oslo)

ideals are being adopted worldwide, it must be for a valid reason. However, many cities have tried to adopt these smart city policies and even with large funding behind them have failed to achieve both the sustainable aspects of a smart city, and the merging of the physical and built environment.

Example - Toronto

In 2017, a subsidiary planning company of Google - called Sidewalk Labs - announced an initiative known as Sidewalk Toronto, which would construct an entirely new smart city from scratch - as opposed to the previously mentioned examples where the pre-existing built environment was retrofitted with IoT technology.15

This $1.3 billion investment by Sidewalk labs “aimed to advance a new model of inclusive urban development along Toronto's eastern waterfront, striving for the highest levels of sustainability, economic opportunity, housing affordability, and new mobility.”16

One of the key headlines of this project was the installation of sensors across the project to help improve the waste, transport and other associated systems that came with the 12-acre development. And herein lies one of the primary criticisms of smart cities - privacy of an individual.

These highly advanced smart cities require thousands of sensors - Santander has 12,000 alone - that collect data - not just about the environment (both built and natural) but also about people and their movements and habits - to help reduce energy or water consumption, for example, to help the cities run efficiently and thus deliver on their goals of sustainability.8, 17

The data collected by a large majority of these sensors is usually encrypted, or in the case of Toronto’s Sidewalk development, anonymised before being available for public access. However, there were major concerns that this data will not actually be private or anonymised, and instead sold on to third party bidders. Sidewalk Labs themselves said that it “will not sell personal information, not use personal information for advertising and not disclose personal information to third parties without explicit consent.”15

Figure 3: The City of the Future? Plans for the controversial Sidewalk Toronto – (Credit: Sidewalk Labs)

FAILINGS OF SMART CITIES

Despite this, the project was criticised by Roger McNamee - a founder of a leading global technology investment firm Silver Lake Partners - who called the project Surveillance capitalism, and stating that “It is a dystopian vision that has no place in a democratic society.” This followed the resignation of two members of the advisory panel to Sidewalk, one of whom, Ann Cavoukian - the former privacy commissioner for Ontario - called out the development, saying “I imagined us creating a Smart City of Privacy, as opposed to a Smart City of Surveillance.”18, 19

Eventually, the project ended in May 2020, with Sidewalk Labs citing the “unprecedented economic uncertainty brought on by the Covid pandemic”.15

2020 seemed to be the year that tech giants worldwide realised that smart cities are not necessarily the solution to all our current urban developmental problems. Cisco, who had attempted to digitise cities worldwide, by offering specialist software services for local authorities, abandoned this idea in 2020, again citing the pressure caused by the pandemic on the company’s bottom line.21

While IoT technologies such as sensors, apps for residents and specialist operating systems certainly could help to create a more sustainable and efficient framework for urban areas - as demonstrated by Oslo - questions will always inevitably be raised about how to go about creating this urban utopia in a democratic and fair way.17

With recent cybersecurity breaches of institutions in the UK such as British Airways and the BBC, where employee data including bank details may have been stolen, it is entirely understandable that potential smart city residents, ordinary people who lack access to as advanced cybersecurity as these companies, are worried about their personal data being stolen, and will be reluctant to allow IT giants like Cisco to collect data on their movements and behaviours.20

With all this in mind can we ever feel safe and secure enough to give away control of the places we live to the closed, empty-minded logical management of algorithms and computers?

And what of advanced AI? How long will it be until one of these tech giants decides that one of their new smart city developments should be run by an advanced AI that can control and monitor incoming data and adjust parameters in a sensible manner to help increase the efficiency of the city? Italy recently outlawed ChatGPT on data protection grounds, suggesting that if AI were to merge with smart city development, a potentially dystopian 1984-esque surveillance society may not be too far off.22

CONCLUDING STATEMENTS

In conclusion, although these smart cities may be an effective solution to solving many of the issues at the forefront of the global spectrum, there is still a long way to go until these policies may be seriously considered to be viable.

Many cities, seeing the buzz generated around the theoretical idea of a smart city have tried to implement their own take. Often, due to the conflicting and confusing definitions of what a smart city is, many authorities simply invest in thousands of sensors, and cease there - this in fact is what Santander, mentioned earlier, has done. Further investments into training people to maintain the sensors, or constructing a viable roadmap for the next 20 years is needed.23

The word smart has been used far too loosely, and as a result, cities are suffering in the long term due to the buzz around this miracle solution. What is even more worrying than the privacy and AI concerns behind the methods in which smart cities deliver on their goals of bettering and optimising urban development and habitation, is that some cities may fall behind on investing in what Mike Anderson calls sensible cities

“ The core idea of smart cities remains as a solid foundation for urban development for the cities of the future.

By measuring the smartness of a city based solely on sensors and IoT technology – due to the conflicting definitions – incentives are provided for cities to invest in these technologies as opposed to essential aspects of a city that are crucial in providing for the needs of the present and future tooeducation, healthcare, public spaces and housing - all of which fall under the umbrella of the sustainable aspect of my definition of smart cities.17

The core idea of smart cities remains as a solid foundation for urban development for the cities of the future - sustainable for residents and the environment - the definition needs to be clarified, and the way that IoT technology is applied needs reassessment before this idea can be fully explored on a global scale.

BIBLIOGRAPHY

1. Urban Development Overview. World Bank. [Online] World Bank Group, April 3, 2023. [Cited: June 1, 2023.] https://www.worldbank.org/en/topic/urbandevelopment/ overview#

2. World Population - Worldometers. Worldometers Web site. [Online] Dadax, June 1, 2023. [Cited: June 1, 2023.] https:// www.worldometers.info/world-population/

3. Yin C T, Xiong Z, Chen H, et al. A literature survey on smart cities. 2015, Vols. Sci China Inf Sci, 58: 100102(18), pg6, doi: 10.1007/s11432-015-5397-4 .

4. Dirks S, Keeling M et al. A Vision of Smarter Cities. s.l. : IBM Global Services, 2009. GBE03227-USEN-04.

5. Anderson, M. Onekeyresources. How to make a sustainable city. [Online] April 29th, 2022. [Cited: June 2, 2023.] https:// onekeyresources.milwaukeetool.com/en/green-city.

6. D. Çinar Umdu, E. Alakavuk. Understanding of Smart Cities, Digital Cities and Intelligent. Volume XLIV-4/W3-2020, 2020, Vols. The International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences. https:// doi.org/10.5194/isprs-archives-XLIV-4-W3-2020-173-2020

7. Brundtland_Commision. Our Common Future. s.l. : Oxford University Press, United Nations, 1987. ISBN: 019282080X.

8. SmartSantander. Home Page. [Online] [Cited: May 28, 2023.] https://www.smartsantander.eu/

9. Santander Facility. SmartSantander. [Online] [Cited: May 28, 2023.] https://www.smartsantander.eu/index.php/testbeds/ item/132-santander-summary

10. Wilson, B. Metropolis: A History of the City, Humankind's Greatest Invention. s.l. : Vintage, 2020. ISBN 9781784707521.

11. Tailpipe Greenhouse Gas Emissions from a Typical Passenger Vehicle | US EPA. U.S. Environmental Protection Agency | US EPA. [Online] United States Environmental Protection Agency, May 30, 2023. [Cited: June 2, 2023.] https://www. epa.gov/greenvehicles/tailpipe-greenhouse-gas-emissionstypical-passenger-vehicle

12. Lakhani, N. UN warns of ‘draining humanity’s lifeblood’ amid worsening water scarcity. The Guardian. https://www. theguardian.com/world/2023/mar/22/un-water-conference-warns-worsening-scarcity. March 22, 2023.

13. New York City Global Partners, Inc. Best Practice: Intelligent Streetlights. New York: New York City Global Partners, 2011. https://www.nyc.gov/assets/globalpartners/downloads/pdf/ Oslo_Climate%20Change_Streetlights.pdf

14. Sidewalk Labs. Sidewalk Toronto | Sidewalk Labs. Sidewalk Labs. [Online] Google LLC., 2023. [Cited: June 6, 2023.] https://www.sidewalklabs.com/toronto.

15. Dudley, D. Renewable Energy Costs Take Another Tumble, Making Fossil Fuels Look More Expensive Than Ever. Forbes Magazine. 29 May, 2019, https://www.forbes.com/sites/ dominicdudley/2019/05/29/renewable-energy-coststumble/?sh=7a6c059be8ce

16. Musulin, K. 5 takeaways from Sidewalk Labs’ smart city master plan Smart Cities Dive. [Online] Industry Dive, June 25, 2019. [Cited: June 7, 2023.] https://www.smartcitiesdive. com/news/5-takeaways-from-sidewalk-labs-smart-citymaster-plan/557518/

17. Anderson, M. Smart Cities: Definition, Origin, Characteristics, & Examples. One Key Resources. [Online] Milwaukee Tool, April 29, 2022. [Cited: June 7th, 2023.] https:// onekeyresources.milwaukeetool.com/en/smart-cities.

18. Cecco, L. 'Surveillance capitalism': critic urges Toronto to abandon smart city project. The Guardian. https://www. theguardian.com/cities/2019/jun/06/toronto-smart-citygoogle-project-privacy-concerns June 6, 2019.

19. Canon, G. 'City of surveillance': privacy expert quits Toronto's smart-city project. The Guardian. https://www.theguardian. com/world/2018/oct/23/toronto-smart-city-surveillanceann-cavoukian-resigns-privacy. October 23, 2018

20. Gill, O. Russian hackers raid British Airways and BBC in cyber attack. The Telegraph. https://www.telegraph.co.uk/ business/2023/06/05/british-airways-and-boots-warn-staffdata-stolen-in-hack/. June 5, 2023

21. Tilley, A. Cisco Systems Pulls Back From Smart City Push. The Wall Street Journal. https://www.wsj.com/articles/ciscoturns-off-lights-on-smart-city-push-11609178895. December 28, 2020.

22. McCallum, S. ChatGPT banned in Italy over privacy concerns. BBC News. [Online] BBC, April 1, 2023. [Cited: June 8, 2023.] https://www.bbc.co.uk/news/technology-65139406

23. Euklidiadas, M, M. Smart Cities that Failed Along the Way. Tommorow.city. [Online] Fira de Barcelona, November 26, 2019. [Cited: May 30, 2023.] https://www.tomorrow.city/ smart-cities-that-failed-along-the-way/

UNPRECEDENTED

An in depth look into attitudes to China across 2020 alongside an investigation into how respondents make decisions at surveys

ARTS HUMANITIES

JOSHUA INGLESFIELD

This Independent Learning Assignment (ILA) was short-listed for the ILA/ORIS Presentation Evening.

SECTION 1 - ABSTRACT

The Covid Pandemic has been extensively shown to have contributed to declining attitudes to the PRC. Various organisations, including the Pew Research Institute and TUC of Wales, have reported how the Covid pandemic has led to increases in such unfavourable attitudes. However, no studies have specifically examined whether the trend is a result of conscious bias regarding the pandemic, whereby citizens actively view the PRC more negatively due to their handling of the pandemic, or subconscious bias, whereby citizens view the PRC more negatively due to a broader sentiment change influenced by undertones and subliminal messaging in society (caused by the PRCs handling of Covid) as opposed

to active views regarding the PRC and the pandemic. Thus, in this study I aim to not only examine how British attitudes to China have been impacted by the Covid pandemic, but also determine whether the trend in attitudes seen is a result of subconscious bias following the pandemic, or conscious bias regarding the pandemic. Through unique methods such as precise analysis of daily new Covid cases alongside attitude figures, as well as the designing of unique fieldwork to determine the biases prevalent, I have enabled the research into the pandemic’s impact on global attitudes to China to be breadthened, adding new dimensions to debate on the topic. Overall, I conclude that the United Kingdom saw, as with other

Western nations, a marked increase in negative sentiment to the PRC in 2020. Further investigation then reveals an undeniable link between daily new Covid cases and negative attitudes, with greater levels of negative attitudes to the PRC occurring in periods of higher infection rates. With regards to biases, I show that subconscious bias was the primary driver of negative attitudes to the PRC as a result of the pandemic, with individuals likely acting upon subliminal messages and undertones they are exposed to, such as speeches by political figures criticising the PRC, rather than them actively recognising the PRC’s actions in the pandemic and forming opinions from such recognition.

Literature Review

There has been a large variety of literature examining how American attitudes to the PRC have changed as a result of the pandemic, with studies such as Pew’s 2020 report1 breaking down the research into demographics, noting that older generations saw a greater jump in negative attitudes to the PRC - an interesting observation given Moonshot’s 2020 report2 suggesting that the social media space (used primarily by younger generations) had been rife with conspiracy theories, hate speech and incitements to violence… related to Covid-19.

Moonshot’s work represented a refreshing take on attitude research, work of a manner similar to that of Cook, Huang, and Xie (2021)3, who crucially mapped tweets mentioning China and Covid over attitudes, allowing for direct comparison of trends. The TUC of Wales4 also noted the prevalence of racist and violent hashtags against Chinese people during 2020 utilising Moonshot data - confirming the popularity of social media analysis for attitudes. The TUC of Wales report also offered a fascinating insight to developments closer to home, noting how some right wing politicians in the UK had adopted China-critical stances as a result of the pandemic, perhaps mirroring such sentiment changes of those they hope to represent. Further work regarding the UK comes from YouGov5, whose report asked questions focusing specifically on the pandemic, however there has been little in depth analysis which broadens the literature on this topic regarding attitudes in the UKa gap I aim to fill.

How it Contributes to Literature

As detailed above, existing studies are for the most part centered around attitudes of Americans to China - a justified focus, given the China-US relationship in the 21st Century is perhaps one of the most studies foreign policy areas. However,

1 https://www.pewresearch.org/global/wp-content/uploads/sites/2/2020/04/PG_2020.04.21_U.S.-Views-China_FINAL.pdf

2 https://moonshotteam.com/resource/covid-19-conspiracy-theories-hate-speech-and-incitements-to-violence-on-twitter/

3 https://www.tuc.org.uk/blogs/i-am-not-virus-anti-chinese-racism-and-coronavirus

4 https://www.tuc.org.uk/blogs/i-am-not-virus-anM-chinese-racism-and-Covid

5 https://yougov.co.uk/international/articles/31524-western-attitudes-china-are-hardening

the UK-China relationship is increasingly relevant, particularly economically, with governments having to address China at increasing frequency - thus, my research here helps extend studies of international attitudes and relations to the UK, adding greater depth to the ever-expanding field of international relations. Further to this, no studies of attitudes to China throughout 2020 and as a result of the Covid pandemic have gone beyond the standard theories of international relations - this study revolutionises this area by extending this research into sociological concepts of bias, thus inextricably linking the individual citizen to research - allowing for greater relevance and 'real world' application of such research for the average citizen. By exploring global attitudes in relation to subconscious and conscious biases, I understand better how attitudes are formed and may be changed. Overall, my research here both complements and builds upon existing literature to further and broaden this field - research I hope may lead to more informed decisions being taken by entities in the international space, as well as allow academics to observe the topic through a wider lens.

Hypotheses

H1: British attitudes to the PRC soured significantly in 2020.

H2: The Covid pandemic was the primary influencer of trends in attitudes to the PRC in 2020.

H3: British attitudes to the PRC were directly related to the pandemic’s prevalence.

H4: The trend in attitudes to the PRC as a result of the pandemic will be driven mainly by conscious bias.

“ By exploring global attitudes in relation to subconscious and conscious biases, I understand better how attitudes are formed and may be changed.

SECTION 2

To what extent have British attitudes to china been impacted by the Covid pandemic?

Brief Overview

Here I use data to determine whether any trend in attitudes across the period where the Covid pandemic was most prevalent6 (2020-2022) was a result of the Covid pandemic, and the extent to which the trend can be directly attributed to the pandemic. Further to this, I examine relationships between infection rates and attitudes to the PRC in more detailed analysis to help establish a greater certainty of whether trends can be attributed to the pandemic.

Attitudes to the PRC in 2020

Data

Somewhat surprisingly, with the resources at my disposal I could only source 2 surveys aimed at tracking British Attitudes to China7 which covered the pandemic years. The first was from the Pew Research Institute8, and the second from YouGov9

The initial data tracked from 2019 to 2020 can be observed below in Fig1 and Fig2:

6 Prevalent in the sense of infection rates, media coverage, and general public concern

7 Utilising similar questions asking of favourability to the PRC – the most common language used in surveys asking of attitudes to a nation.

8 https://www.pewresearch.org/global/wp-content/uploads/sites/2/2023/07/PG_2023.07.27_Vews-of-China_TOPLINE.pdf

9 https://yougov.co.uk/politics/articles/44064-quarter-britons-consider-china-be-enemy-uk

Fig 1 – Pew Research Institute

% decrease in total favourable attitudes from Spring 2019 to Summer 2020: 16%

% increase in total unfavourable attitudes from Spring 2019 to Summer 2020: 19%

Fig 2 - YouGov

% decrease in total favourable attitudes from 13/05/2019 to 11/08/202010: 19%

% increase in total unfavourable attitudes from 13/05/2019 to 11/08/2020: 22%

Average

In order to benefit from both of the above surveys, I have calculated the average change in both total favourable attitudes and total unfavourable attitudes across the two surveys, the figures for which are displayed below. I utilised both datapoints displayed above from the Pew data, and selected the datapoints which were closest to the Pew date descriptors of. 'Summer 2020' and 'Spring 2019' for the YouGov data11:

Average % decrease in total favourable attitudes to the PRC from 2019 to 2020 utilising data from the Pew Research Institute and YouGov: 17.5%

Average % increase in total unfavourable attitudes to the PRC from 2019 to 2020 utilising data from the Pew Research Institute and YouGov: 20.5%

Visualising the Data

Using the same process12 used to calculate the averages across the two datasets above, I created a graph (Fig 3) which plots the average % of respondents who answered unfavourable, somewhat unfavourable or very unfavourable across both surveys when asked for their attitudes to China. For this graph I extended the dataset I used back to 2017 and forwards to 2022 to allow for better understanding of how the drop in favourable attitudes to China in 2020 relates to longer term trends in the preceding and following years of the pandemic.

Analysis of Data

Initially it is clear that the year 2020 saw a jump in unfavourable attitudes to China - to the extent that when the Pew and YouGov surveys are averaged as shown above on the graph, 2020 saw the greatest jump in total unfavourable attitudes to China in the period shown above, as well as the highest point of unfavourable attitudes in the same period.

10 These dates were chosen as they are those that most closely resemble the rough time periods the Pew study covered, allowing for rough comparison to as accurate a degree as is possible with the provided data, thus allowing us to better build a picture of the attitudes in this period.

11 For convenience the datapoints used have been highlighted in the tables that constitute Fig 1 and Fig 2.

12 Here the data was taken from the Pew Research Institute as data by year, as only one datapoint per year was provided, and from YouGov it was taken from the datapoint nearest to the season quoted by Pew as when their survey took place - i.e. if for YouGov I had datapoints from May and August 2020, and the Pew datapoint was for Spring 2020, I would use the figure for May from YouGov.

Fig 3

“ Initially it is clear that the year 2020 saw a jump in unfavourable attitudes to China.

However, though there is a notable jump in unfavourable attitudes to the PRC in 2020, there also appears to be a longer term13 trend. This can be seen with the % unfavourable gradually rising between Spring 2018 and Spring 2019 prior to the 2020 spike.

With Regards to H1

This proves H1 as being true: there was indeed a souring in British attitudes to the PRC in 2020, with a 22% increase in unfavourable attitudes to the PRC between 13/05/2019 and 11/08/202014 significant in attitude research, and certainly more than grounds to satisfy H1’s wording of significant

TO WHAT EXTENT CAN THIS TREND BE ATTRIBUTED TO THE PANDEMIC?

Surveys provide some additional data as to how Britons link the pandemic (an event widely agreed to be negative) to the PRC

Britons actively attribute some blame for the pandemic to the PRC:

1. YouGov (04/06/2020 - 15/06/2020)15 conducted in Great Britain: 49% said “The Chinese Government” is “most responsible for the severity of the Covid pandemic”.

2. YouGov (04/06/2020 - 15/06/2020)16 conducted in Great Britain: a total of 86% said that they "don’t trust very much or don’t trust at all". The Chinese Government to tell the truth about the Covid pandemic.

3. BFPG (Poll1: Feb 2020, Poll2: Apr 2020)17 conducted in the UK: In their first 2020 poll on the matter (Poll1) conducted in February 2020 the BFPG found 21% of Britons held trust in China to Act Responsibly in the World, whereas in their second 2020 poll on the same matter (Poll2) conducted in April 2020 the figure had dropped to 17%. Given that the majority of focus on Chinas actions in a global context was on how it acted with regard to the pandemic, this can be considered as some evidence that Britons viewed how the PRC faced the pandemic negatively. The counterargument that this change seen can be attributed to other issues regarding China that Britons may be considering when answering the poll such as human rights issues can be disregarded due to not fulfilling the global actions the survey is focusing on - seen in the phrasing act responsibly in the world.

4. YouGov (04/06/2020 – 15/06/2020)18 conducted in Great Britain: 24% said that since the beginning of the Covid pandemic their opinion of the Chinese Government had slightly worsened, and 36% said their opinion had significantly worsened.

5. Sinophone Borderlands (Sept-Oct 2020)19 conducted in the UK: Utilised a word association question asking “What is the first thing that comes to your mind when China is mentioned?” - responses shown in Fig 4 below where Covid was clearly the issue at the forefront of individual’s minds regarding China, suggesting it would be the most prevalent issue when an individual is responding to an opinion poll, thus suggesting the pandemic was a significant contributor to the spike in attitudes to the PRC seen in 2020.

13 Relative longer-term trend - by which I mean it is not simply a matter of a short term spike in 2020, but instead a seemingly gradual process over 3 years (2018,2019,2020).

14 Taken from the YouGov dataset uttilised in this section.

15 https://docs.cdn.yougov.com/o90gsx3oja/TBI_CovidGlobalImpact_June2020_Topline_CLIENT.pdf

16 https://docs.cdn.yougov.com/o90gsx3oja/TBI_CovidGlobalImpact_June2020_Topline_CLIENT.pdf

17 https://bfpg.wpenginepowered.com/wp-content/uploads/2021/02/BFPG-Annual-Survey-2021.pdf

18 https://docs.cdn.yougov.com/o90gsx3oja/TBI_CovidGlobalImpact_June2020_Topline_CLIENT.pdf

19 https://sinofon.cz/wp-content/uploads/2021/01/UK-poll-report.pdf

Thus, upon first inspection it is clear that in 2020 Britons were critical of how the PRC reacted to the pandemic and did not hold trust in the PRC - two sentiments indicative of negative sentiment, thus here implying that the pandemic did considerably contribute to declining attitudes to the PRC in 2020. Further to this, Survey 3 implies that the pandemic was responsible for the trend in attitudes due to its singling out of the pandemic seemingly resulting in a large proportion of respondents noting their perceptions of the PRC had soured.

However, Britons may be concerned by the PRCs other actions, thus suggesting that other issues were to blame for the hardening of attitudes to the PRC:

1. Pew (March 2021)20 conducted in Australia: 59% said that China’s policies on human rights is a very serious problem for Australia. Though Australia is evidently not the UK, it is similar in terms of its links to the UK, and as the survey here was not distributed to the UK in 2021, it can be used for rough gauging of what public opinion may have been in the UK. Thus, here it is probable that UK residents held a similar scepticism and concern for the PRC’s approach to human rights. This however cannot be used as conclusive evidence, and strictly as a guide.

20 https://www.pewresearch.org/wp-content/uploads/sites/2/2014/03/Morality-Topline-FINAL.pdf

Fig 4

2. BFPG (6-7 Jan 2021)21 conducted in the UK: 40% supported the UK Challenging China on its human rights record.

3. Sinophone Borderlands (Sept-Oct 2020)22 conducted in the UK: Utilised a word association question asking What is the first thing that comes to your mind when China is mentioned? - responses shown in Fig 4, where while Covid was the word most associated with China, the appearance of Uyghurs, human rights issues, secretive, controlling, animal rights issues, authoritarianism, oppression, dictatorship, corruption, Hong Kong, and lack of freedom all suggest an underlying concern at a multitude of issues which are perceived to plague the PRC aside from the pandemic.

4. Sinophone Borderlands (Sept-Oct 2020)23 conducted in the UK: Asked respondents for their Perceptions of certain China-related issues (seen in Fig 5), of which key issues to consider here are: Chinese military power, China’s impact on global environment, and China’s influence on democracy in other countries - with the negative sentiment shown again reinforcing the idea that Britons were concerned by other issues regarding China to a reasonable degree aside from the pandemic in 2020.

5. Sinophone Borderlands (Sept-Oct 2020)24 conducted in the UK: Asked respondents if the UK should cooperate with China to advance 5G in the UK, with the results showing the complexity and depth of suspicion and concern towards China with negative sentiment extending beyond the issues of human rights and military power.

21 https://bfpg.wpenginepowered.com/wp-content/uploads/2021/02/BFPG-Annual-Survey-2021.pdf

22 https://sinofon.cz/wp-content/uploads/2021/01/UK-poll-report.pdf

23 https://sinofon.cz/wp-content/uploads/2021/01/UK-poll-report.pdf

24 https://sinofon.cz/wp-content/uploads/2021/01/UK-poll-report.pdf

Fig 5

Fig 6

6. Pew Research Institute (Poll1: Spring 2018, Poll2: Spring 2020) conducted in the US: In Spring 2018 (Poll1) 49% of US respondents stated that China’s policies on human rights were a very serious problem, whereas in Spring 2020 (Poll2) the figure had increased to 57% - thus giving greater validity to the argument that a portion of the spike in negative attitudes to the PRC (not just high levels of negative attitudes) was a result of non-pandemic concerns. I have shown this US-fielded survey as no such survey was available for the UK, and due to respondents in each nation holding roughly similar attitudes towards the PRC I felt it could be used for a guide to Western attitudes in general which may help the debate here - though it cannot be used as conclusive evidence.

Thus, here it seems that while the pandemic was likely the major cause of negative attitudes to the PRC in 2020, there were a number of underlying issues which catalysed such sentiment. However, it must be understood that many of these issues are not 'new' in 2020 in the way the pandemic was, thus suggesting that much of these other concerns Britons hold regarding China were simply party to the rough25 baseline from which a spike was observed in 2020.

Trend Matching

Though such numbers of surveys as referenced above are useful in providing a rough idea, I devised a method to attempt to determine a more conclusive link between the pandemic and negative attitudes to the PRC by comparing and overlaying infection rate data with attitude data provided by YouGov - possible due to

YouGov surveying on 4 occasions per year instead of the expected 1 for the years I examined.

It is clear that public opinion on the pandemic was influenced heavily by social media and news on the situation26. Given that the majority of reporting in the period was coverage of infection rates, it can be concluded that infection rate changes contributed to the nature of public opinion during the pandemic. Thus, overlaying specific data on attitudes to the PRC in 2020/21 in the UK over infection rate data presents itself as a compelling method to determine correlation and thus, through the logic outlined above, possible causation.

The Data Used

Attitude data: YouGov27

Infection data: Ourworldindata28

Note: Infection data provided by Ourworldindata is a 7-day rolling average of daily new confirmed Covid cases. Ourworldindata stresses that 'Due to limited testing, the number of confirmed cases is lower than the true number of infections.'

Initially, I will examine data from 2020 to see if immediate observations can be made.

Fig 7: Ourworldindata Daily new confirmed Covid cases (UK) between the dates 6/2/2020 and 10/11/202029:

25 Rough is used here in the sense that no scientific (no deviation) baseline was present, merely a period of no significant jump or fall in negative attitudes.

26 https://moonshotteam.com/resource/covid-19-conspiracy-theories-hate-speech-and-incitements-to-violence-on-twitter/

27 https://yougov.co.uk/politics/articles/44064-quarter-britons-consider-china-be-enemy-uk

28 https://ourworldindata.org/coronavirus/country/united-kingdom

8: YouGov % of respondents who expressed they had an unfavourable view of China (GB) (4 data points recorded, the first being on 5/2/2020 and the last on 10/11/2020)

Initially there would seem to be a reverse correlation to what one might expect between infections per day in the UK and attitudes to the PRC in the UK within this time period, with unfavourable attitudes seen in the 10/11/2020 YouGov poll dropping from the previous datapoint, despite this point marking the highest ever infections per day of the time period we have isolated. Further observations show that excluding the datapoint 05/02/2020 (when public awareness and commotion over the Covid was significantly reduced31) the higher the daily new confirmed Covid cases, the lower the negative sentiment shown towards the PRC. As previously stated, this is the reverse of what would be expected.

However, I felt that perhaps there was an incidence of delay - where the trend in negative attitudes did indeed correspond to infection rates, however the poll data seemed to be almost acting as if

respondents were looking at infection data a couple of months earlier.

Therefore, I re-evaluated the data by comparing infection rates in 2021-2022 to attitudes to the PRC in this period, as this would not only give me a greater dataset to compare, with a higher number of fluctuations to line up across the two graphs, but would also resolve fears that attitude data from 2020 - very early in the pandemic - would be skewed slightly by any panic and potential media manipulation of the event which may have resulted due to the unprecedented nature of Covid. Though of course panic and such reactions to an unprecedented event must be considered as valid when reflected in attitude data, the aim of our trend comparison here is simply to form an undeniable link between the pandemic and changing attitudes to the PRCan experiment for which we seek individuals able to rationally consider the pandemic without any interfering 'shock factor'.

29 Note: we selected the nearest possible dates as possible to the earliest and latest polls of the 4 YouGov polls we are comparing this data to – with the one discrepancy being the earliest YouGov poll was conducted on the 5th of February 2020, and the nearest date to this we could observe data for on the infection rate data was the 6th of February 2020. This made no difference to our conclusions due to the incredibly low levels of infection in the early periods of the data.

30 See footnote 6.

31 https://crisisnlp.qcri.org/covid19 - shows comparative low numbers of tweets regarding the Covid in February 2020 compared to later in 2020.

Fig

Fig 9: Side-by-side comparison of data

Fig 10: Ourworldindata Daily new confirmed Covid cases (UK) between the dates 17/02/2021 and 05/08/2022:

Fig 11: YouGov % of respondents who expressed they had an unfavourable view of China (GB) (8 data points recorded, the first being on 17/02/2021 and the last on 05/08/2022):

My immediate thought here was to overlay the two graphs, aligning dates, given the greater visibility of changes in public opinion and greater dataset. The result can be seen below, with the line for the total of unfavourable attitudes to the PRC from early 2020 to late 2021 shown in blue, and the line for the daily

new confirmed Covid cases shown in red. This at first glance presents a similar paradox as with the 2020 data comparison - with it seeming that at the lowest point of unfavourable sentiment towards the PRC there was a spike in daily infections.

However, this paradox is quickly solved here thanks in part to the greater dataset considered and greater variability of the public opinion compared to the data for 2020, which allows for greater visibility and understanding of trends. The solution is simple; when a respondent is considering how they will answer an opinion poll, they consider information at their disposal, and will often prefer basing a response from complete data rather than incomplete data as patterns are more visible - with pattern

recognition linked to intuition32 (the likely way by which respondents select an answer). Therefore, it can be assumed that one would consider33 complete past trends when answering such as opinion poll as utilised here - a reasonable assumption given the availability of infection data as graphs clearly showing peaks and troughs during the pandemic. As in the comparison of infection data to attitudes to the PRC from 2021 to 2022 the attitudes line seems slightly offset to what would be expected

32 https://www.harvardbusiness.org/data-and-intuiMon-good-decisions-need-both/ “at its best, intuiMon is a powerful form of pattern recognition”.

33 Either subconsciously or consciously.

Fig 12: Side by side comparison of data

Fig 13

(i.e. the trend is what one would expect should the line be shifted back a little - see Fig 14) it can be concluded that in fact the pandemic’s changes in daily new confirmed cases in the UK had a delayed impact on attitudes towards the PRC due to the consideration of past data, thus meaning there was a direct relationship between daily new confirmed Covid cases and negative attitudes to the PRC. The repeated occurrences of such a trend34 again evidence strengthening the claim. If further

34

evidence is needed, one might observe the YouGov polled figure of 74% respondents viewing the PRC as unfavourable on 17/02/2021, and though the data for daily new confirmed Covid cases prior to this date was not included in the comparison in Fig 13, we can see that there was indeed a spike in daily infections in Fig 15 which preceded the spike in negative attitudes, thus again reinforcing our conclusions here with a further example of the matching trends.

Fig 15

Fig 14 (for illustration purposes only)

Addressing the Potential Skewing of Data Due to the Invasion of Ukraine

One significant counterargument to the claim of daily new confirmed Covid cases being linked significantly to attitudes to the PRC is the claim that a final spike in negative sentiment towards the PRC in mid-late 2022 seen in Fig 13 and until here attributed to the double spike in daily new confirmed Covid cases occurring around early 2022 was in fact a result of Russia’s invasion of Ukraine (which also occurred early 2022). This is primarily due to the association often made between Russia and China due to their most glaring similarities for the average Western citizen of the two countries both being autocraciesan association only strengthened with regards to the invasion of Ukraine by Russia by the PRC’s abstention during the vote on UN General Assembly Resolution ES-11/1, being one of only 35 nations to do so, and the most significant nation of the 3535. With early 2022 marking a swivel from pandemic-related news to full focus on Russia’s invasion of Ukraine, as can be seen by the covers of the popular British satirical publication (selected for observation here due to its readily available archives of covers on the limited

Note how Issue 1566 is focused on the pandemic – and although it is focused on government scandals during the pandemic, it represents a continued focus on the pandemic in general and thus its position as being forefront in the public’s minds prior to the invasion of Ukraine.

35 Most significant nation in the UN of the 35 abstentions, based upon its status as a permanent member of the UN Security Council and thus one of the 5 generally accepted most prominent nations in the UN.

Fig 16: Private Eye Issue 1566, 04/02/2022 (20 days prior to invasion)

Note how suddenly the focus of Private Eye pivots from be enthralled by pandemic related news to fears of an invasion in Ukraine.

04/03/2022 (14 days after the

resources at our disposal, and it’s tendency to focus on the overall most popular/discussed news on its front pages) displayed in Fig 16,17, and 18 below, it is only natural that given both the pandemic and the invasion reflect in some form negatively on the PRC

for citizens of the UK that it be concluded at least some of the jump in negative attitudes to the PRC in later 2020 be attributed to the invasion of Ukraine, yet also some still be attributed to the pandemic. Given the pattern of a spike in negative attitudes to the PRC in the months following a spike in daily Covid rates, and the fact that the spike in negative attitudes that begins c.May 2022 follows a similar pattern it is imperative that the impact of Covid incidence rates here on attitudes is not discarded entirely.

However, due to the overwhelming media coverage of the invasion of Ukraine and it’s pivoting away from the pandemic, as well as the invasion’s dominating of social media seen in displays of support for Ukraine being displayed by Western social media pages (such as via the widespread addition of a Ukrainian flag with a motto expressing support for Ukraine being employed by such social media users36 - see Fg 19), it can be reasonably assumed that a large proportion of the spike in negative sentiment towards the PRC beginning c.May 2022 can be attributed to the invasion of Ukraine.

Fig 19

Despite this, it is important to consider that whether the final spike in negative sentiment towards the PRC is eventually attributed at all to the pandemic it does not discredit our main conclusion, as there are 3 other examples of a relationship between public opinion towards the PRC and Covid daily cases being delayed by similar time periods - with the average delay between the 3 peaks in daily new confirmed Covid cases and the 3 peaks in unfavourable sentiment identified below being 86.33 days.

36 https://www.leicestermercury.co.uk/news/leicester-news/how-show-your-support-ukraine-6730581

Fig 17: Private Eye Issue 1567, 18/02/2022 (6 days prior to invasion)

Fig 18: Private Eye Issue 1568,

invasion)

1. 17/02/2021 - 74% unfavourable - preceded by spike in daily Covid cases peaking at 62,520.14 on 06/01/2021 (Fig y)

2. 13/08/2021 - 70% unfavourable - preceded by drop in daily Covid cases troughing at 1,967.00 on 05/05/2021 (main comparison graph)

3. 17/11/2021 - 73% unfavourable - preceded by spike in daily Covid cases peaking at 48,923.29 on 21/07/2021 (main comparison graph).

With Regards to H2

H2 is proven correct here, as shown through both the analysis of external surveys asking of the pandemic itself, and the relationship between daily new confirmed Covid cases and unfavourable attitudes to the PRC.

With Regards to H3

“ A marked increase in negative attitudes to the PRC among Britons in 2020, a trend which can be largely attributed to the pandemic.

H3 is proven correct here, as shown through the overlaying of the graph showing daily new confirmed Covid cases and that showing total unfavourable attitudes to the PRC.

CONCLUSION

Overall, the Covid pandemic has undoubtedly resulted in a hardening of British attitudes towards the PRC, represented within the period of the pandemic as a period of peaks and troughs acting almost as 'steps' at times, building up anti-PRC sentiment throughout the pandemic. These 'steps' or peaks and troughs in negative sentiment can be asserted to have been following the general trend in daily new Covid cases in the UK. Some may argue correlation rather than causation, however due to such peaks and troughs being 'in tune' with the daily new confirmed Covid cases on multiple occasions it can be said to be beyond correlation, and thus 'beyond reasonable doubt' that this is a matter of causation.

There are however undoubtedly other factors beyond the pandemic alone impacting trends in public opinion we identified - hence why, for example, the 'delay' between a peak in daily Covid cases and the ensuing peak in negative attitudes to the PRC was in one case just 42 days, 44.33 days short of the average of 86.33 days. Despite this the general trends seem to show that the pandemic was indeed linked directly to attitudes to the PRC in the UK - with such precise analysis showing direct correspondence which is difficult to challenge. Therefore to conclude Section 1, hypotheses 1, 2, and 3 were proved correct, with there being a marked increase in negative attitudes to the PRC among Britons in 2020, a trend which can be largely attributed to the pandemic due to the close correlation between the daily Covid cases and negative attitudes, which show how the general prevalence of the pandemic was directly related on a month-by-month scale to attitudes to the PRC.

SECTION 3

Were the trends in attitudes to the PRC as a result of the pandemic a result of subconscious bias or conscious bias?

Brief Overview

As outlined above, it is clear that the pandemic resulted in an increase in negative attitudes to the PRC among Britons.

Here I aim to explore that conclusion to a greater extent, by looking at how respondents behave when polled. By this I mean I am aiming to determine whether in the scenario of the pandemic, those reporting their sentiment towards China did so with active consideration of the pandemic and the PRC’s role in it (which I term conscious bias), or conversely if they did so while subconsciously considering the pandemic (which I term conscious bias).

“ At any given moment, public opinion is a chaos of superstition, misinformation and prejudice

Gore Vidal

An Outlining of Subconscious and Conscious Bias in this Context

Subconscious Bias

• Where one subconsciously (not actively) links the Covid pandemic’s occurrence to their perception of the PRC

• By this we mean where one would simply think “I disliked the PRC between 2020- 2021 compared to 2019” - and despite Covid having an impact as we have proved earlier in this paper the impact is subconscious.

Conscious Bias

• Where one actively and consciously links the Covid pandemic’s occurrence to their perception of the PRC

• By this we mean where one would think “I disliked the PRC during the Covid pandemic (because of how they dealt with it etc.) compared to before it”- the impact of Covid shaping their opinion is conscious and actively introduced and entertained within their decision making of which response to give in the opinion poll.

How I Will Attempt to Answer this Question

Though this research is completely unique, some answers may be found in existing surveys through looking for respondent’s views on countries

responsible for the pandemic among other variations, and those which specifically mention attitudes with relation to the pandemic. A few such survey questions were found in my research, however due to a) the unusual question I am attempting to answer, and b) the focus country being the UK, a country I found to be less surveyed than the US the questions do not suffice to come to a reasoned judgement. Therefore, I designed my own fieldwork to specifically approach the research question of this section, which utilises a unique quality to allow for conclusions to ultimately be made. However, I have also included an additional useful survey found in order to provide some additional information or perhaps evidence of the difficulty in utilising existing surveys to approach this question.

Existing Survey of Use Question

“Tell me how much confidence you have in each leader to do the right thing regarding world affairs - a lot of confidence, some confidence, not too much confidence, or no confidence at all. B. Chinese President Xi Jinping.”

Why this Survey

Looking at how UK citizens perceive the way the PRC’s leader approaches world affairs before, during, and after the pandemic allows us to understand how

forefront the primary world affairs issue in 2020 (the pandemic) was in respondent’s minds. For example, a large drop from before 2020 to after 2020 in positive perceptions of the way Xi Jinping approaches world affairs would suggest that the increase in negative attitudes was a result of conscious/active bias as respondent’s focus is clearly on the pandemic.

Data

Conclusions

This data here would initially suggest conscious bias as the main driver behind the trend in attitudes to the PRC as a result of the pandemic, though as

previously stated this is a highly tenuous conclusion, and it is necessary to design unique fieldwork to answer this question.

My Fieldwork/Survey

Why I Designed My Own Survey

Unfortunately, given the ground-breaking nature of this research project, there were no pre-existing surveys conducted with the sole intention of determining if the impact of the Covid pandemic on attitudes to China was a result of subconscious or conscious bias. Therefore, I designed fieldwork of my own to answer this question, and distributed it to UK residents, accepting responses between the dates of 4/8/2023 and 12/8/2023 - thus allowing me to come to an accurate and well-informed conclusion rather than assumption based upon other surveys.

The Design of My Survey

My survey was designed with the aim of allowing for a clear and simple conclusion - with it’s principle purpose to answer the specific question at hand. The system I devised is as follows:

• Two separate surveys were distributed, where I would control response numbers

to ensure the two had similar numbers of respondents by locking one of the forms if necessary to allow the other to gather more responses.

• The two forms were identical aside from one detail. The forms both asked of the respondents attitudes to China prior to the pandemic, and then their attitudes to China today.

The forms varied in that one form explicitly used the words Covid pandemic, whereas the other used time language (i.e. 5 years ago) - which can be seen below under the Survey questions section.

This was done in order to distinguish between conscious bias and unconscious bias - by which I mean the survey where the pandemic was explicitly mentioned (Survey 1) was designed to trigger conscious bias by making the respondent actively recognise the pandemic as a potential influencer on their attitude to China, and thus make the respondent consider the pandemic actively before making their decision in the poll. On the contrary, the survey where the pandemic was not explicitly mentioned (Survey 2) was designed to not trigger any conscious bias by deliberately drawing the respondent away from actively considering the pandemic by planting the notion of binary dates - i.e. the respondent would be focused on dates, not events. Thus in Survey 2 the primary function I attempted to elicit was unconscious bias.

Thus, to come to a conclusion as to whether changes in attitudes to the PRC as a result of the pandemic were a consequence of subconscious or conscious bias I would simply need to observe the % changes in total unfavourable attitudes from prior to the pandemic to after the pandemic in each survey. If the survey designed to provoke conscious bias (Survey 1) saw an increase in unfavourable attitudes greater than the survey designed to provoke unconscious bias (Survey 2), conscious bias can be concluded to be the primary manner by which respondents considered the pandemic when responding to polls regarding attitudes to China - and vice versa.

Please note the following survey nomenclature:

Survey 1: The survey where the pandemic is explicitly mentioned, and the survey is aimed at provoking conscious bias.

Survey 2: The survey where the pandemic is not explicitly mentioned, and the survey is aimed at eliciting unconscious bias.

Respondent numbers

Survey 1 had 14.3% more respondents than Survey 2 survey questions: (Please note the sections of the questions which varied between the two forms are highlighted in yellow).

Survey 1: [Explicitly mentioned the Covid pandemic].

Please select the option which best describes your attitude to China before the Covid pandemic.

Options

- Very favourable towards.

- Somewhat favourable towards.

- Neutral.

- Somewhat unfavourable towards.

- Very unfavourable towards.

2. Please select the option which best describes your attitude to China today after the Covid pandemic.

Options

- Very favourable towards.

- Somewhat favourable towards.

- Neutral.

- Somewhat unfavourable towards.

- Very unfavourable towards.

Survey 2: [Did not explicitly mention the Covid pandemic].

1. Please select the option which best describes your attitude to China 5 years ago.

Options

- Very favourable towards.

- Somewhat favourable towards.

- Neutral.

- Somewhat unfavourable towards.

- Very unfavourable towards.

2. Please select the option which best. describes your attitude to China today.

Options

- Very favourable towards.

- Somewhat favourable towards.

- Neutral.

- Somewhat unfavourable towards.

- Very unfavourable towards.

Key Data to Observe

Key Figures

Survey 1: Increase in total unfavourable attitudes from pre to post pandemic = 25%.

Survey 2: Increase in total unfavourable attitudes from 5 years ago to today = 44%.

Additional Figures

Survey 1: increase in very unfavourable attitudes from pre to post pandemic = 16%.

Survey 2: increase in very unfavourable attitudes from 5 years ago to today = 11% .

Survey 1: increase in somewhat unfavourable attitudes from pre to post pandemic = 9%.

Survey 2: increase in somewhat unfavourable attitudes from 5 years ago to today = 33% .

CONCLUSIONS

Main Conclusion

From the data above it can be concluded that the trend in attitudes to the PRC as a result of the pandemic were driven primarily by subconscious bias, where individuals did not actively recognise the pandemic as their reasoning for their declining attitudes. This does not discredit the argument that the pandemic was the main reason for such trends, and however simply shows that respondents likely did not directly identify any factors at all for the drop in attitudes, and acted instead off of the subconscious decision making.

This conclusion can be reached due to the survey where the pandemic was not explicitly mentioned and the focus shifted to the binary timeline of years (Survey 2) seeing a 44% increase in total unfavourable attitudes from 5 years ago to today, and Survey 1 only seeing a 25% increase in a similar time period. This shows that even when the issue of the pandemic is not at the forefront of individuals’ minds, they still reported a large increase in unfavourable sentiment towards the PRC across the pandemic years.

Sub-conclusions

Upon closer inspection, further conclusions can be developed. Through looking at the increase in 'very unfavourable' attitudes across the pandemic years in

the surveys as separate to the increase in 'somewhat unfavourable' we can identify that the survey where the pandemic was explicitly mentioned (Survey 1) saw a greater jump in 'very unfavourable' attitudes than the survey which only referred to years (Survey 2). The reverse was true for the jump in 'somewhat unfavourable attitudes'.

Therefore, it can be concluded here that where the pandemic is actively considered regarding attitudes it catalyses more extreme sentiment, and when the pandemic is considered subconsciously it catalyses more mild sentiment.

Thus, the respondents who answered 'very unfavourable' to the second question of both surveys were likely driven more by conscious bias than subconscious bias, and the respondents who answered 'somewhat unfavourable' to the second question of both surveys were likely driven more by subconscious bias than conscious bias. However, overall subconscious bias was the more prevalent in caused changes in attitudes as a result of the pandemic.

Weaknesses in My Fieldwork and How to Improve It

Weaknesses

- Small sample size.

- Most likely concentrated demographic as shared via social platforms.

- Anonymous - additional useful data not collected e.g. political affiliation.

- Questions vary by 'real' timeframe - Survey 1 measured 2019 and 2023 attitudes, whereas Survey 2 measured 2018 and 2023 attitudes by virtue of wording.

The most obvious issue with my own survey is the small sample size, an inevitability resulting from a lack of resources and platform (e.g. The Pew Global Attitudes Panel is a large scale platform example) by which to distribute the survey more widely. Further to this, due to the survey being primarily distributed via social medias (in order to obtain a greater sample size), a narrower demographic can be inferred to have responded, as though the survey was anonymous social media users are likely to be younger, and with regards to this survey the respondents are likely those with whom I have some degree of connection, whether via friends or familythus they are likely to be of a similar demographic as myself.

To Improve

Aside from those two principle issues with the survey, a number of smaller concerns may be raised. One such concern is that the two surveys measured the change in attitudes across marginally different time periods. By this we mean Survey 1 asked for attitudes 'before the Covid pandemic' and 'after the Covid pandemic' - suggesting attitudes were measured between 2019 and an unspecified year the respondent deemed the pandemic to have 'finished' in (likely 2022 or 2023). On the contrary, Survey 2 asked for attitudes '5 years ago' and 'today', thus measuring attitudes between 2018 and 2023. Therefore, here it is clear that the two surveys may not be directly comparable.

One might be immediately tempted to offer the response of increased resources/funding - and though this would likely solve the majority of the issues above, as by allowing for the development of structures such as the Pew Global Attitudes Panel a greater number of respondents could be reached, and the demographic of respondents be widened. However, it is a largely unfeasible solution to such a project. Thus, I would propose the following: Firstly, that the survey be sent via school communications of the RGS Guildford, allowing for a greater variety and number of respondents - or, indeed, the survey could be publicised in local area noticeboards, whether on social media platforms or established local news outlets, which would allow for an even wider range of respondents, eliminating the potentially narrow demographic my study may have inadvertently surveyed. Regarding collecting additional useful data: It is perhaps an inevitability of small-scale fieldwork conducted by a student that for a poll to be effective it should be anonymous, due to likely lower levels of trust regarding storage of personal data in comparison to the trust placed in established organisations such as the Pew Research Institute. Therefore, though additional data such as political affiliation or age may be useful, it is largely impractical to collect as part of such a project, and nevertheless in this project would have added marginal benefit compared to the core conclusions drawn.

Finally, with regards to the slightly different time periods respondents may have considered between Survey 1 and Survey 2 I would perhaps reword the

question for Survey 2 to be '4 years ago' rather than '5 years ago'. This would result in Survey 2 having a time period over which respondents consider if their attitudes have changed of 4 years between 2019 and 2023, a likely similar time period as Survey 1. It must be considered however that in my fieldwork I utilised '5 years ago' for the first question of Survey 2 due to '5' years often being perceived as more 'well rounded' number than '4'. Nonetheless, changing the question wording is an incredibly simple change which I would implement should I conduct the fieldwork again.

With Regards to H4

H4 is proven false here, as it is clear that overall when an individual’s subconscious bias is elicited they report a greater increase in unfavourable attitudes as a result of the pandemic, demonstrating how the main driving force behind the trend in attitudes to the PRC as a result of the pandemic was subconscious bias.

SECTION 4

Summary

To summarise, my research here has utilised ground-breaking approaches to allow for a greater understanding of British public opinion as a result of the pandemic, with particular notable methodology being the overlaying of infection rate data over attitude data, and the creation of bespoke fieldwork utilising a novel approach in making use of 2 separate surveys to better understand the way opinion poll respondents make decisions when being polled.

Overall Conclusion

I shall start by outlining conclusions drawn beyond the testing of my initial hypotheses.

My research into the trend in attitudes to the PRC as a result of the pandemic in section 2 has shown not only a fascinating relationship between infection rates and attitudes, thus demonstrating how closely citizens follow unusual events how strongly this influences sentiments, but also the volatility of public opinion in times of crises. While the trend in attitudes to the PRC was overall shown as a souring throughout the period of the pandemic,

the presence of such distinct fluctuations in opinion linked to external stimuli emphasise this variability in sentiment individuals are susceptible to.

My later research into biases revealed the conclusion that when conscious bias is the main decision-making process for respondents considering their attitudes to the PRC, more extreme negative sentiments are shown, and when the reverse is true more mild negative sentiments are shown. With regards to my hypotheses.

H1: “British Attitudes to the PRC Soured Significantly in 2020.”

H1 was proven true, though to no great surprise: it is clear that in other Western nations more studied than the UK, such as the US, similar trends occurred. I proved this hypothesis through simple analysis of two datasets, one from the Pew Research Institute, and another from YouGov. With YouGov reporting a 22% jump in unfavourable attitudes to the PRC between 13/05/2019 and 11/08/2020 the provision of H1 of 'significantly' is satisfied.

Further to this, the evident trend may be visualised through my use of graphs utilising average figures from the Pew and YouGov surveys for simple observation of H1 being correct.

H2: “The Covid Pandemic was the Primary Influencer of Trends in Attitudes to the PRC in 2020.”

H2 was proven true through my analysis of surveys asking specifically regarding the period 2020, the pandemic, the PRC’s global stance, and attitudes with regard to the three. To further clarify my hypothesis, I developed my own methodology of overlaying the data for daily infections in the UK over attitude datapoints taken up to 4 times a year by YouGov. This revealed a correlation which undeniably proved the link between the pandemic and the increase in negative attitudes to the PRC. Though I also found that other factors resulting in attitude changes, such as human rights issues, should not be disregarded it is clear that the pandemic was the single most significant factor which was not as prevalent in 2019, and thus it can be held accountable for the attitude change.

H3: “British attitudes to the PRC were directly related to the pandemics prevalence.”

H3 was proven true through my overlaying of daily new Covid cases over attitude data provided by YouGov, which as outlined earlier initially showed the reverse to what would be expected: Though an instance of delay was soon recognised and accounted for - resulting in a more expected result, and one coherent with H3. I found direct relationship between daily new Covid cases and unfavourable attitudes to the PRC, showing clearly the validity of H3 and how individuals were evidently receiving and considering infection data to a great extent - or, alternatively, the attitudes of media publications may have been impacted by infection rates, attitudes then passed on to viewers/readers of such media.

H4: “The Trend in Attitudes to the PRC as a Result of the Pandemic Will Be Driven Mainly By Conscious Bias.”

H4 was proven false by my fieldwork. I wrote H4 on the basis that the widespread awareness of the pandemic would result in individuals actively considering it when answering polls, however this was not the case. Instead, it became clear that while the pandemic was widely recognised, a direct link to a strong dislike of the PRC was not actively made, a sensible conclusion given the lack of a prominent coordinated worldwide movement stoking anger against the PRC37

3. Moonshot 2020, From #CovidCoverUp to #NukeChina: An analysis of conspiracy theories, hate speech and incitements to violence across Twitter related to Covid, Moonshot, https://149736141.v2.pressablecdn.com/wp-content/ uploads/COVID-19-on-Twitter_Moonshot.pdf

4. Stanford Center on China’s Economy and Institutions 2022, The Impact of Covid on American Attitudes towards China, Stanford Center on China’s Economy and Institutions, https://sccei.fsi.stanford.edu/china-briefs/impact-covid19-american-attitudes-toward-china-0

5. Cook, G.G., Huang, J & Xie, Y 2021, How Covid has Impacted American Attitudes Toward China: A Study on Twitter, https://github.com/junminghuang/covid-impacts-attitudetoward-china

6. De Wall, J.R. 2020, Western attitudes to China are hardening, Pew Research Institute, https://yougov.co.uk/international/ articles/31524-western-attitudes-china-are-hardening

7. Tony Blair Institute for Global Change 2020, Resetting the West’s Relationship With China, Tony Blair Institute for Global Change, https://www.institute.global/insights/geopolitics-and-security/resetting-wests-relationship-china

8. School of the Art Institute of Chicago 2020, Lesson 6: Conscious and Unconscious Bias, School of the Art Institute of Chicago, https://libraryguides.saic.edu/learn_unlearn/ foundations6

9. Summers, T, Man Chan, H, Gries, P & Turcsanyi, R 2021, Worsening British views of China in 2020: evidence from public opinion, parliament, and the media, Asia Europe Journal, https://link.springer.com/article/10.1007/ s10308-021-00639-x

10. Farrell, M 2023, Data and Intuition: Good Decisions Need Both, Harvard Business Publishing, https://www. harvardbusiness.org/data-and-intuition-good-decisionsneed-both/

11. Summers, T, Turcsányi, R.Q., Šimalčík, M, Kironská, K & Sedláková, R 2020, British public opinion in the age of Covid: Worsening perceptions amidst the pandemic, Central European Institute of Asian Studies, https://sinofon.cz/ wp-content/uploads/2021/01/UK-poll-report.pdf

12. Gaston, S & Aspinall, E 2021, UK Public Opinion on Foreign Policy and Global Affairs, British Foreign Policy Group, https://bfpg.wpenginepowered.com/wp-content/ uploads/2021/02/BFPG-Annual-Survey-2021.pdf

Datasets:

BIBLIOGRAPHY

Works either referenced or considered for this report are included here. With thanks to the authors for such freely available academic resources.

Reports or other literature:

1. Silver, S, Devlin, K & Huang, C 2020, Americans Fault China for Its Role in the Spread of Covid, Pew Research Institute, https://www.pewresearch.org/global/wp-content/uploads/ sites/2/2020/04/PG_2020.04.21_U.S.-Views-China_FINAL. pdf

2. Detzler, M 2020, I am not a virus – anti-Chinese racism and Covid, Wales Trades Union Congress, https://www.tuc.org. uk/blogs/i-am-not-virus-anti-chinese-racism-and-coronavirus

1. YouGov 2020, https://docs.cdn.yougov.com/hdemoi825d/ Internal_ClimateChangeTracker_220720_GB_W.pdf

2. Pew Research Institute 2022, https://www.pewresearch. org/global/wp-content/uploads/sites/2/2022/06/ PG_2022.06.29_views-of-china_REPORT.pdf

3. Pew Research Institute 2023, https://www.pewresearch. org/global/wp-content/uploads/sites/2/2022/06/ PG_2022.06.29_views-of-china_REPORT.pdf

4. CRISISNLP 2020, https://crisisnlp.qcri.org/covid19

5. Our World in Data 2023, https://ourworldindata.org/ coronavirus/country/united-kingdom

37 Here the word 'coordinated' is key – there were of course groups however there was a distinct lack of publicly voiced worldwide dislike towards the PRC with direct reasoning being given as the pandemic.

BEYOND THE PÉRIPHÉRIQUE

Examining the failures of the built environment in the downfall of the Parisian suburbs

ARTS HUMANITIES

HARIS MACNICOL

This Independent Learning Assignment (ILA) was short-listed for the ILA/ORIS Presentation Evening.

INTRODUCTION

Paris is a city firmly detached from its surroundings. Beyond its ring road lies a ‘not Paris’, a parallel universe which exists in limbo, beyond the administrative boundaries of the city but part of its conurbation. Far from the idyllic American suburbia, the reality of the French Banlieues (or suburbs) is one of vast concrete housing complexes (known as cités) hosting poverty, crime and racial inequality. Although forcefully removed from the Parisian sphere of influence, the cités exist as scars of the failed modernist utopia, Paris would like to forget.

The architectural dogma of Le Corbusier paired with governmental constraints resulted in a built environment that failed and continues to fail to this day, the residents who find themselves in these beton-brut poverty traps. To examine the failings of the Banlieues, is to examine the lack of oversight that the architects and the government displayed in conjunction with inequalities that have persisted since the downfall of the banlieues some 50 years ago.

Still from film La Haine

Through decades of shockingly poor planning, design, management and policy, the banlieues have become areas of relegation, poverty and segregation. Categorised by a high percentage immigrant population along with low public spending and unemployment, the seeds of failure were sown some 60 years ago when the utopian promise of the hi rise began to fall through. After the ravages of World War Two, France was in deficit of 3 million residences. To address the dire housing situation, the government unleased beton-brut (raw concrete) modernism, a new radical modus operandi. This new style of building vast tours (hi-rises) and barres (long horizontal units) led to the creation of huge grouped clusters of buildings know as les grandes ensembles or habitation a at a rate of 300,000 habitation à loyer modéré units (social housing, HLM for short) per year from 1953-73 (Bergationi, 2013). The rate and scale of construction was unparalleled as the government experimented with this new type of architecture that went hand in hand with their desire to build big, quick and cheap. The shortcomings of these complexes were numerous, the architects were often blinded by a desire to determine a new Corbusierist future of uniformity and modernity whilst ignoring elements that were actually necessary for habitation. For example, in 1968 only 41% of HLM units had individual toilet facilities. Regardless, these vast complexes were beacons of modernity for the displaced middle classes, who flocked in droves to these concrete suburbs. The initial buzz of the grandes ensembles quickly faded as the reality of the impracticality of their design became evident to residents. Transport links to commercial centres were almost non-existent and public services such

as schools, medical centres and leisure facilities were severely lacking. In 1960, there were 29 lycées (secondary schools) in central Paris, yet the banlieues, with twice the population had less than half that. Furthermore, the number of day centres per child under 3 was and still is six times lower than the national average (Institute Montaigne, 2023).

Those who had the means to, left, leaving vacant units to be filled by first generation immigrants whom the government forcefully relocated from bidonvilles (shanty towns in and around Paris). Along with being in an unfamiliar environment, the new residents of the Grandes Ensembles inherited the problems that the first residents so quickly fled from, without the social support or economic means to do anything about it.

“ It’s hard to think of a worse physical context for the successful integration of millions of incomers. The banlieues may as well have been designed as ghettoes.

Peter Franklin

Construction of the 4000s housing estate, Courneuve Paris circa 1957 (Archives municipales de La Courneuve). The piles of prefabricated concrete panels facilitated the rapid construction of sprawling housing estates.

One of the famous wide angle shots from the film La Haine dir. Mathieu Kassowitz 1995. A gritty masterpiece that presented an introspective, unfiltered look into life in Parisian Banlieues, tackling the topics of racism, poverty and police brutality. The raw , visceral nature of the film peaked my interest in the French Banlieues and can be considered as the inspiration for this essay.

'We are products of our environment,' is none truer than in the cités of Paris. A clear distinction must be made between the French word ‘cité’ and its English homophone ‘city’; one refers to the large, dynamic urban settlements found globally, whereas the other is used to refer to the sprawling, drowsy, housing estates that were once beacons of modernity found around Paris and other major French cities. The building blocks, quite literally, of each cite are the aforementioned Grandes Ensembles. Initially the shape of each building was dictated by the axis movements of a crane (le chemin de grue), allowing for rapid assembly. However the rectangular form was abstracted over time to create shapes of varying obscurity. Some assume the typical Vertical Block form such as the Balzac tower block situated within the 4000 housing estate, whereas others assume unconventional shapes such as circular ‘Arenes de Picasso’ or Angular ‘Cité du Parc’.

An aerial photograph of ‘Les Arenes de Picasso’ (540 units) designed by Spanish-French architect Manuel Nunez Yanowsky. (Nunez 1985)

The capacity and density requirements of units took priority, disregarding the scale of these buildings, leading to vast constructions that literally swallowed their inhabitants within labyrinth-like corridors and alleys. Once the posterchild for postwar modernism, these Parisian concretopias have fallen victim to the very thing that set them apart, their architecture. Imposing, unfriendly, soulless design has led the inhabitants to echo the buildings they found themselves in. Their failure came from the fundamental principles that informed their construction, the functionalist style. Although there is no unified doctrine as such, functionalism rose to prominence in the post-war period, the perfect anti-thesis to art deco and other pre-war decorative styles. It abandoned the cornices, columns, archways and other embellishments of yore to pursue simple, rational construction. But humans are inherently irrational creatures; they love beautiful, decorative things regardless of their practicality and so placing them in a sanitised environment is oppressive and saps the soul. By removing unnecessary aspects of construction, the Grande Ensembles feel like dormitories rather than homes, the uniformity of the units more akin to prisons or office buildings.

Constructions purely designed for utility, not for normal human habitation, ‘a machine for living’, as coined by architect Le Corbusier. Looking beyond the scale of each individual building, the blocks are arranged in introverted clusters around poor quality open spaces leading to a sense of entrapment. Perhaps as Le Corbusier desired, so called unnecessary human interactions were stifled by an absence of meaningful social space.

A lack of direct involvement from Le Corbusier did not prevent his radical thinking from informing the construction of many of the banlieues. The Franco Swiss architect has become synonymous with modernist architecture, primarily the subset of functionalism. Functionalism is a utilitarian approach to building, the embodiment of function over form, a building is simply a unit of habitation. Sanctified amongst the international community for his physical love letters to concrete, Le Corbusier’s unrealised concepts are often overlooked. Le Corbusier’s highly developed plans for urban renewal betray the disquieting undertones of his work. ‘Ville contemporaine’ could be considered a manifesto to authoritarian architecture, when state and the architect are one in the same as they are

An Aerial Photograph of the notorious La Grande Borgne housing estate designed by Emille Aillaud (©Alex Maclean 2010)

in the banlieues. The plan suggested the raising of the cultural centre of Paris, to be replaced by social housing for the proletariat, segregated from the 60 storey bourgeoisie inhabited skyscrapers by parkland. Randomness was Corbusier’s bête noire as he sought to homogenise all needs, facilities and services into single units, the precursor to the 15 minute city. The extent to which these plans were developed was troubling for many of Le Corbusier’s peers. He sought an architectural cleansing, rationalising the built environment and feudalizing society once more, under the roof of modern authoritarianism. Evidently, Corbusierist thinkings trickled down onto the pages of the banlieue’s authors. They were disciples of modernist principles, who materialised some of Corbusier’s dream in the 3 million council properties constructed during the Trente Glorieuses (A post war period of economic growth across France 1945-75).

The writtings of Le Corbusier in conjunction with his physical creations (such as the Unite d’Habitation Marseilles) were so prolific that, although not in direct involvement, the power of association was enough for his thinking to permeate through banlieue

design. The themes of car centric infrastructure, social stratification and uniformity present across the majority of banlieues can be accredited at least in part to Le Corbusier, in conjunction with the accompanying social problems. Le Corbusier’s dream came to fruition, not in Paris as intended but rather on its peripheries.

“

The capacity and density requirements of units took priority, disregarding the scale of these buildings, leading to vast constructions that literally swallowed their inhabitants within labyrinth-like corridors and alleys.

La Grande Borgne can be considered a case study of how the built environment has failed an entire community and created one of the most turbulent, disadvantaged neighbourhoods in France. Unemployment sits at a staggering 22%, nearly triple the national average. The serpentine cité is fenced by 3 major roads that both contain and ostracize those within. Aillaud intended for the city to move away from the monotonous slabs of the original Grandes Ensembles and be altogether more organic. Unfortunately, the problems of the hi rise were simply translated horizontally. Clusters are inward facing and claustrophobic, separated by ‘liberated ground’ that is no more than parched earth. Sculpture and art work attempt to lift the spirits of residents, but unfortunately paint does not fix deep rooted societal and spatial problems. A universal issue throughout the banlieues is that due to the uniformity of units, there is no semblance of a local property ladder, those who want an upgrade move out do so leaving those with no choice behind. Many of the cites, such as La Grande Borne fall victim to gigantism, that is to say they are of a scale that is both dehumanising and unsustainable. Many are small cities in their own right , density and population wise, but without any semblance of neighbourhood (La Grande Borne pushes 30,000 inhabitants). Perhaps the crux of the issue is that when the cités were constructed, the authorities forced physical neighbourhoods onto the environment in the form of clustered towers, however a successful neighbourhood is one that is both physically and socially coherent. One cannot dictate a neighbourhood, it is something organic that develops with time, but the authorities tried to quarter up commercial, educational and residential areas into separate sectors. The image provide of La Grande Borne demonstrates this, there is no

evidence of commercial activity, or anything other than the blank monotonous facades of its buildings. Consequently, the cités organs are detached. Due to the continuous length of units, La Grande Borne is more akin to a labyrinth than a housing project; it is closed off, secretive, and conceals the anti-social activities that occur within its confines. Its design means it is simply impossible to police effectively, residents are by themselves as there is no active police station within the cité. La Grande Borne is an archipelago of isolation, an architectural failure to be blunt.

The concept of the Hi-rise implies proximity to a city centre, however all the Parisian cités are firmly located en banlieue, that is to say, beyond the threshold that is considered Paris. The physical seperator is the Périphérique , an 8 lane ring road that encircles Paris, leaving ‘not Paris’ in its wake. Coincidentally opened in 1973, the year of the international oil crisis, by Georges Pompidou, the new ring road essentially sealed the fate of the banlieues as zones of relegation, alienating the suburbs to an even greater degree. Rather than facilitating connections for the disadvantaged, the Périphérique ruthlessly severed them. It is inherently inhospitable to pedestrians and often is often recessed into the ground like a modern day moat or elevated above ground. As journalist, Justin Tribillion, noted, the Périphérique plunges underground in the more affluent neighbourhoods of Vinciennes and Bolougne but remains omnipresent in the poorer suburbs of Seine St Denis. A constant visual reminder of the detachment from Paris the banlieuesards feel. As well as the metaphorical and physical division the road creates, the economic division is tangible too, best shown by housing prices. Cross the Périphérique from the 19e arrondisment into Auberbvillieurs (literally either side of the Périphérique ) and the

average cost per square metre of 8200 euros halves (www.meilleursagents.com).

Regardless of this concrete belt, the mere geographical distance from central Paris to even the most central cités is significant. It is crucial to note Paris is a highly centralised city. There is a distinctive lack of so called concentric towns that serve as localised centres orbiting the Parisian nucleus. Thus, there is a reliance on central Paris to provide employment. However, for residents inhabiting a banlieue beyond the Périphérique threshold, the commute into Paris can be a significant barrier to employment. Distances range from circa 25km for the most distant such as Les Ulis or Les Bergeries to 5km for the most central, but many lie between 10 to 15km away, certainly out of what ‘proper’ Parisians would call Paris. This geographical distance creates a sociopolitical disconnect that only serves to further ostracise the vulnerable communities that inhabit les Grandes Ensembles. Stuck in a no man's land, devoid of economic opportunities, a commute into Paris is necessary but the legacy of poor transport infrastructure has persisted.

Journalist Leopold Lampert (2015) , illustrated the spatial inequalities by way of the map above, showing transport stations with a 15 minute walkable radius around each, overlayed with the cités. Many cités are beyond the 15 minute threshold, such as Les Bosquets and La Butte Rouge pushing the public transport towards inaccessibility and thus routes to

“ What was once the

forefront of contemporary design theory is now regarded as dystopian and dysfunctional. The very physical landscape of the banlieues and indeed Paris has created an exclusion zone that divides and keeps separate the Parisian heart from its limbs.

employment become voyages rather than commutes. A plus-one-hour commute has become normalised, but when it is reliant upon at least 3 transfers the journey becomes arduous and susceptible to delays and strikes. A resident of the cité Les Bosquets who commutes to a restaurant in Nation would have to use 3 different methods of transport to cover the 1hr 10min commute, whereas a resident of Santeny (an affluent area) can reach Nation in under 50 minutes with 1 change despite being geographically twice the distance away.

The now defunct Congres International d’Architecture Mondiale, advocated for a built environment that fulfilled the four functions of human settlement: work, play, transport and housing. There is a distinct lack of the first three and an abundance of the fourth across the banlieues which has led to the establishment of communities filled with diversity in an environment that is distinctly authoritarian and dehumanising. The question arises, were the banlieues designed for an ideal case scenario that never was to manifest? A subservient, homogenous population of workers united under a concrete roof. What was once the forefront of contemporary design theory is now regarded as dystopian and dysfunctional. The very physical landscape of the banlieues and indeed Paris has created an exclusion zone that divides and keeps separate the Parisian heart from its limbs.

REFERENCES

1. Archives municipales de La Courneuve (n.d.). Le chantier de construction des 4000 à La Courneuve. Available at http:// books.openedition.org/psorbonne/docannexe/image/2425/ img-6.jpg (Accessed 20/06/23)

2. Baker.C, Mancunian Matters (n.d). Les Banlieues. [online] Available at: https://schoolofjournalism.shorthandstories. com/les-banlieues/index.html (Accessed: 13/05/2023)

3. Bertagnini, E. (2013) The French Banlieues between Appropriation and Demolition. Planum: The Journal of Urbanism, 27(2), pp. 10–16.

4. Carpenter, J. (n.d.) The French Banlieue: Renovating the Suburbs [from: The Routledge Companion to the Suburbs (ISBN: 9781138290235)] [online] Available at https:// radar.brookes.ac.uk/radar/file/0a668947-1e4c-4980-98399e6e7834292a/1/French%20banlieu%20-%202018%20-%20 Carpenter.pdf (Accessed: 13/05/2023)

5. Cupers, K. (2017). The Power of Association: Le Corbusier in the Banlieues — Critical Urbanisms, University of Basel. [online] Available at: https://criticalurbanisms.philhist. unibas.ch/library/the-power-of-association-le-corbusier-inthe-banlieues (Accessed: 21/05/2023)

6. Kassowitz, M. (1995) La Haine (frame). Available at https:// filmobsessive.com/wp-content/uploads/2020/02/MV5BZDhmNDFjMjEtOTUzZC00NTc4LWI3YmItMWVmM2IzMjk1MjI1L2ltYWdlXkEyXkFqcGdeQXVyNDAxOTExNTM@._V1_.jpg (accessed 21/06/23)

7. Lambert, L. (2015) 15-Minute Walking Distance From a Train Station: Spatial Inequality In Paris Banlieues (map). Available at: https://thefunambulist.net/app/uploads/2015/08/15min-Distance-Banlieues-Map-by-Leopold-Lambert-for-theFunambulist-20151.jpg (Accessed 20/06/23)

8. Lefebvre.I, Institut Montaigne (2022). Life in France’s Banlieues: Overview and Battle Plan. [online] Available at: https://www.institutmontaigne.org/en/expressions/ life-frances-banlieues-overview-and-battle-plan (Accessed: 10/06/2023)

9. Maclean, A (2010). Serpentione housing estate, France. Available at: https://i.redd.it/icn59od1p7r71.jpg (accessed 22/06/23)

10. Meilleur Agents (2023). Prix immobilier dans le 19ème arrondissement de Paris. [online]. Available at https://www. meilleursagents.com/prix-immobilier/paris-19eme-arrondissement-75019/ (Accessed: 11/06/2023)

11. Nunez, M. (1985). Place Picasso. Available at: http://www. nunez-yanowsky.com/images/projects/place_picasso_ public_and_residential/details/s/05.jpg (accessed 21/06/23)

12. Mollard.M , Architectural Review (2019). Tour de banlieue: the grands ensembles of Paris’s periphery. [online] Available at: https://www.architectural-review.com/essays/tourde-banlieue-the-grands-ensembles-of-pariss-periphery (Accessed: 20/05/2023)

13. Pearson.E (2023). Why the Paris périphérique is more than just a ringroad [online] The Local fr. Available at https:// www.thelocal.fr/20230228/why-the-paris-peripherique-ismore-than-just-a-ringroad (Accessed: 30/05/2023)

14. Says, H. bonval (2018). The brutal apartheid of the French banlieues. [online] UnHerd. Available at: https://unherd. com/2018/09/french-elite-ignores-apartheid-banlieues/ (Accessed: 20/05/2023)

15. Shreir, D. (2014). Utopia Deconstructed: Le Corbusier and the Banlieue [online] The Isis. Available at: https://isismagazine. org.uk/2014/04/utopia-unconstructed-le-corbusier-and-thebanlieu/ (Accessed: 21/05/2023)

16. The Associated Press (2015). Inside La Grande Borne: A lawless, concrete labyrinth where Paris terrorist Amedy Coulibaly grew up. Available at https://nationalpost.com/ news/inside-the-grande-borne-a-lawless-concrete-labyrinthwhere-paris-terrorist-amedy-coulibaly-grew-up (Accessed: 10/06/2023)

17. Tribillon, J. (2015). Dirty boulevard: why Paris’s ring road is a major block on the city’s grand plans. [online] The Guardian. Available at: https://www.theguardian.com/cities/2015/ jun/26/ring-road-paris-peripherique-suburbs-banlieue (Accessed: 30/05/2023)

18. Wikipedia Contributors (2023). Boulevard Périphérique. [online] Wikipedia. Available at: https://en.wikipedia.org/ wiki/Boulevard_P%C3%A9riph%C3%A9rique (Accessed: 30/05/2023)

19. Wikipedia. (2020). La Grande Borne. [online] Available at: https://en.wikipedia.org/wiki/La_Grande_Borne (Accessed: 10/06/2023).

REVIVING THE PAST

Unravelling the Magnetic Pull of Historically Informed Performance in the 20th Century

ARTS HUMANITIES

THOMAS MCLEAN

This Independent Learning Assignment (ILA) was short-listed for the ILA/ORIS Presentation Evening

HOW CAN WE DEFINE EARLY MUSIC?

To understand the impetus of the 20th Century early music revival and the significance of Historically Informed Performance, ‘early’ demands definition. If one utilises a literal lens, the definition of ‘early’ surely gets later as time progresses, as exemplified by the description of the 1776 ‘Consort of Ancient's repertory of music that was more than just 20 years old - clearly illustrating the slowing of the

rate that music has developed over time1. However, in the 21st Century world of early musicians, early music is generally regarded to be a collective term for the music from the Medieval to the Baroque, encompassing the period from the late 8th Century to 1750. This essay will examine the historically informed performance (HIP) of Renaissance and Baroque music in the 20th Century.

1 (Kelly, 2011)

HISTORICAL REVIVALS

History is sporadically punctuated with historical revivals of art and music: a cultures that seem distant to our own. One may observe instances in artistic and musical history when a certain culture has reflected upon previous civilisations with a certain intrigue. In architecture and literature, artistes have been inspired by the past, as manifested in movements like neo-classicism. Music has observed similar historical revivals. Although the 20th Century revival of early music is the most well-known and well-documented, it was by no means the only such revival. As early as 1585, in Vicenza, a production of the Ancient Greek play Oedipus Rex was performed in a theatre whose architecture mimicked that of Ancient Greece, accompanied by the music of the contemporary composer Andrea Gabrielli: an amalgamation of the contemporary and the ancient2. Furthermore, Mozart stated in a letter to his father in 1782:

“I go every Sunday at twelve o’clock to the Baron van Sweiten, where nothing is played but Handel and Bach.”3

adolescence, hence influencing his music taste for the rest of his life. However, the most critical revival of early music occurred after Felix Mendelssohn’s 1829 performance of JS Bach's St Matthew Passion, a work oft-played today, but that held little appeal in 19th Century European repertory outside Leipzig, highlighting the frequent neglect of certain musics that do not comply with contemporary trends throughout history.

Baron van Sweiten was one of Mozart’s patrons and employed his services to transcribe many of JS, CPE and WF Bach’s keyboard fugues for string quartets, as well as larger works like Handel’s Messiah for a contemporary scoring. 4 It can be observed that Mozart appeared a little frustrated by Van Sweiten’s rather regressive approach to music in the above quote, perhaps due to the fact that Bach was alive for the much of Van Sweiten’s

Mendelssohn’s 1829 performance of the Passion seemed to echo Mozart’s efforts in that it used a remarkably different orchestration to that which Bach would have heard: Mendelssohn famously conducted from a piano, not a harpsichord and his impressively large two choirs surmounted to 158 people.5 However, such large-scale performances were not a novelty, as one can observe in the performance of the Messiah mass in Handel’s own memorial service. Further, it can be observed that Mendelssohn condensed the Passion significantly as he believed that his audience would have too short an attention span for the whole three-hour work, choosing to remove ten arias and six chorales.6 Nevertheless, it must be appreciated that the work of Mendelssohn was critical in the revival of Bach’s larger choral works in a manner that cannot be observed again in music history. As Fanny Mendelssohn stated:

“[The choir sang] with a fire, a striking power and also with a touching delicacy and softness the like of which I have never heard”.7

2(Kelly, 2011)

3(Tomita, 2000)

4(Kelly, 2011)

(Blain, 2022)

6(Blain, 2022)

THE 20TH CENTURY EARLY MUSIC REVIVAL

As we have demonstrated, revivals of earlier cultures have occurred throughout history, but for a huge variety of reasons. For example, a sense of inquisitiveness about the exotic and novel, or an academic and historic investigation into the past, or to reflect the politics of the time. It certainly seems that Mendelssohn’s revival of Bach coincided with a surge of nationalism in continental Europe, in a manner that perhaps led to an increase in Germany’s interest in its cultural heritage: why did Mendelssohn choose to revive Bach rather than Palestrina whose counterpoint can be argued to be just as study-worthy as Bach’s? Such rises in nationalism can also be observed to coincide with the rise of conservatoire culture as educators sought out repertoire suitable (related to their country) for study, a concept that would eventually lead to the problematic ‘Western Classical Music Canon’.

The causative factor of the revival of early music in the 20th Century can be observed to be far more complex than a single factor, as evidenced by the divisions the subject has caused among critics. As Jonathon Shull (2006) asserts, 20th Century interest in early music seems to be similar to the 19th Century orientalising projects by composers such as Debussy and Ravel, and hence can also be observed to reflect the 20th Century rise of world music and ethnomusicology.8 Combined with this factor of excitement at novelty, it can be observed that many critics of the Early Music Revival speculate over the sincerity at the root of the movement; perhaps it was more egocentric or financially motivated, whilst some scholars like Kailan R Rubinoff suggesting that the principles of early music follow an almost Marxist agenda.9

7(Blain, 2022)

8(Shull, 2006)

9(Rubinoff, 2013)

Part of the beauty of the revival was the fact that it was not one dimensional, with many performers inspired by differing aspects of early music, leading to various, contrasting, and refreshing performances. Whilst many ensembles were simply ‘early music ensembles', one may observe how certain individuals and groups found their niche: Thomas Binkley’s Studio der frühen Musik specialised in 14th Century Italian songs, troubadour or French estampies; Michael Morrow's Musica Reservata became known for their uniquely focused and directed sound in relation to Renaissance music, with efforts focused on pronunciation, improvisation and dance rhythms; David Munrow brought early music to a wider audience through his TV presence; Anthony Rooley's Consort of Musicke became renowned for working with singers who sang with very little to no vibrato, with a language-forwarding style, like Emma Kirkby; and Benjamin Bagby’s Sequentia specialised in Medieval monophony.10

The revival began in earnest in the 1960s in The Netherlands due to their government’s generous funding to the arts sector and general support of novel and avant-garde ideologies.11 Whilst early

10(Kelly, 2011)

11(Rubinoff, 2013)

12(Kelly, 2011)

music is certainly not avant-garde, it was the openness of The Netherlands as a whole that allowed the revival to be centred so specifically there.

THE

PURPOSE OF THE EARLY MUSIC REVIVAL

“And it [the EMR] is organized in a sort of grass-roots, non-hierarchical network of training and playing opportunities that lie largely outside traditional concert and educational structures - or it did so in the 1970s and 1980s, when this aspect of the movement was at its height.” 12

The purpose of music is a neigh-impossible question to answer, but nonetheless, one can evaluate musicians’ motives for playing music and hence estimate by implication what the purpose of their work is. There is certainly a strong point to be made for the purpose of early music ensembles being increased connectivity with the past in a manner that is simply not achievable in a symphonic

setting where there exists relatively little connectivity between players, hence leading to a possible lack of connectivity with the music. However, Thomas Forrest Kelly argues that the early music revival itself ‘is self-consciously archaizing, in that it senses a gap rather than a continuity, with the past’, hence suggesting that at some point, early music simply ceased to be played, and hence that in order for it to be played today, it must be revived.13 In many ways Forrest Kelly’s statement is true, as the very nature of any form of art or music evolves at such a rate that often music that isn’t wholly compliant with contemporary social fashions is left obsolete: confined to the history books. Hence, it can certainly be argued that the early music revival acknowledges a gap in music history, simply through its very existence.

Furthermore, as previously alluded to, the Early Music Revival (EMR) acted as a method for some musicians to escape the confines of the symphony orchestra. As the musicologist Laurence Dreyfus highlights, the very nature of a symphony orchestra involves a hierarchical system, with the conductor at the top of a somewhat feudal system, with the principal player of each section being next in this pecking order, and lastly the ‘ordinary’ musicians.14 Often, such ‘ordinary’ musicians felt aggrieved by what they perceived to be the authoritarian nature that conductors’ indomitable interpretations had on the remainder of the ensemble:

the intellectual subordination of orchestral playing. Prominent ‘cellist and conductor Nikolaus Harnoncourt, as well as the conductor and harpsichordist Ton Koopman, both spoke out against such intellectual subordination: they later became influential in the growing Early Music world.15, 16

Early Music

1. The conductor is banished.

2. All members of the ensemble are equal.

3. Ensemble members play a number of instruments, sometimes sing, and commonly exchange roles.

4. Symptomatic grouping: like a consort – like-minded members of a harmonious family.

Musical Mainstream

1. The conductor is the symbol of authority, stature, and social difference.

2. The orchestra is organised in a hierarchy.

3. The “division of labour” is strictly defined, with one player per part.

4. Symptomatic grouping: the concerto-opposing forces struggling for control; later, the one against the many.

5. Virtuosity is not a set goal and is implicitly discouraged. 5. Virtuosity defines the professional.

6. Technical level of professionals is commonly mediocre.

7. The audience (often amateurs) may play the same repertory at home.

8. The audience identifies with the performers.

9. Programs are packed with homogenous works and are often dull.

10. Critics report on the instruments, the composers, pieces and that “a good time was had by all.”

6. Technical standards are high and competitive.

7. The audience marvels at the technical demands of the repertory.

8. The audience idealises the performers.

9. Programs contain contrasting items and are designed around a climax.

10. Critics comment in the performer and his interpretation.

13(Kelly, 2011)

14(Dreyfus, 1983)

15(Mertl, 1999)

16(Kozinn, 1987)

However, there were very many who criticised the infantile days of the 20th Century early music revival, mostly due to the often-poor quality of playing. Dreyfus draws his judgement in a table, perhaps more of a satirical observation than a critique:17

It must be observed that many critiques of the Early Music revival were by no means any way near as satirical and casual as Dreyfus. The Israeli violinist and conductor Pinchas Zukerman was a vehement and relentless critic of the early music stating:18

“The first time I heard that shit, I couldn’t believe it. It’s complete rubbish, and the people who play it… Maybe one or two or a half-dozen have wonderful musical minds. But I certainly don’t want to hear them perform.”

Whilst Zukerman’s judgement is certainly crass and non-discriminatory in his critique of the early music revival, he does raise a substantial point that many of the early music ensembles of the 1950s and ‘60s did not prioritise ‘good’ playing over historical sensitivity and the unity that players struggled to feel in larger ensembles. However, over time, the standard of playing in early music ensembles has increased to a highly virtuosic level that is surely now comparable to the level of any top symphony orchestras.

WHAT IS AN HISTORICALLY INFORMED PERFORMANCE (HIP)?

Historically informed performance is the performance of music using the instruments and techniques that a composer originally intended the music to be played with. Therefore, great efforts were taken in the 1960s and 1970s in researching obsolete instruments of the past and reconstructing them,

17(Dreyfus, 1983)

18(Everett-Green, 2000)

often from museum articles that remain in their original state. However, this method is flawed as one cannot ever know what, for example, a 17th Century harpsichord would have sounded when it was new; we only know what it sounds like 300 years later: should we guess how it would have sounded when it was new, or should we make new instruments that reflect the sound contemporary instruments make today?

However, HIP is not simply an appreciation of contemporary instruments’ sonorities, but is rather, in a perfect world, a complete recreation of historical performance practices and techniques, as well as tunings and temperaments. Often HIP has been mistaken as a performance on historically sensitive instruments, when in reality a true HIP is played with obsolete techniques and appropriate tuning systems. id est; A = 392-415Hz in Baroque music rather than the standard 440. Nevertheless, historically sensitive instruments are critical to a true historically informed performance.

It is also worth highlighting the fact that HIP is non-discriminatory in its period of history. Although the term is used mostly in relation to early music, it is possible to have an HIP of a work like Stravinsky’s Rite of Spring, which would utilise the now obsolete trumpet in F as well as unusual playing techniques, like the playing of the viola vertically rather than the usual horizontal. Such performances have indeed occurred, most notably at the 2008 Utrecht Music Festival, proving the timeless nature of HIP.19

HISTORICALLY INFORMED PERFORMANCE IN THE 20TH CENTURY EARLY MUSIC REVIVAL

“The concern for historical awareness, propriety and consequent responsibility, which initially spawned the much maligned authenticity movement in Early Music, has been tempered and mollified and now manifests itself in the guise of what is generally referred to as "historically informed performance."

(Glibly abbreviated as HIP). 20

The 20th Century Early Music Revival was unique in its emphasis on historically sensitive performance. As mentioned above, Mozart and Mendelssohn both drastically changed the orchestration that composers like Bach originally indicated. Therefore, it must be observed that the inclusion of historically accurate instruments and sensitive performance practices is a good thing for the fullness of time as it allows the past to be remembered and documented in a manner that has often been neglected in large periods of history.

Initially, the use of historically sensitive instruments had a mixed reception from critics like Richard Taruskin as he believed that many ensembles were ‘fetishizing some forms of historical information like metronome markings or vibrato-less sound [and the instruments themselves], whilst completely ignoring others like improvisation practices or tempo flexibility.’21 Such theories were not uncommon, and it does not take too much imagination to see the use of period-inspired instruments (perhaps cynically) as a way to increase the novelty value and perhaps expand the audience. However, the simple fact that historically informed orchestration still exists today perhaps disproves this argument as HIP has become thoroughly normalised in all the Early Music Ensembles of today.

19(Kelly, 2011)

20(Shull, 2006)

21(Taruskin, 1990)

Whilst many performers felt undermined and insulted by the harsh critique of figures like Taruskin, it can certainly be observed that such criticism led to a greater atmosphere of self-reflection in the early music revival, that perhaps made performers more conscious of the decisions they made surrounding historical music practices leading to a more informed and convincing performance.

However, HIP was in the revival’s ancestry when one researches the very earliest proponents of the 20th Century revival. Arnold Dolmetsch was a key figure in the commencing of the revival in the 1920s through his establishment of the Haslemere Music Festival where his Dolmetsch Consort performed.22 As he was an instrument maker, Dolmetsch’s focus was naturally on the instrument that early music was being performed upon, hence he began to focus his work upon the manufacturing of instruments like lutes, violas, recorders, clavichords, and harpsichords. Whilst the early music revival had certainly not started in earnest in the 1920s, it can be observed that many of the Dutch efforts in the ‘50s and ‘60s continued the ideas of HIP that Dolmetsch pioneered in the ‘20s.

It can generally be observed that the reason HIP remained such an important part of the 20th Century early music revival, despite its many criticisms by figures like Taruskin and Zukerman, was because of the involvement of what Thomas Forrest Kelly calls ‘scholar-performers’.23 Performers like Dolmetsch, Robert Donington, Thruston Dart as well as the distinguished composer Paul Hindemith sustained the intellectual aspect of HIP and allowed the academic and practical sides of early music to amalgamate for an ultimately historically sensitive and practically superior performance.

Whilst the more specialised music ensembles mentioned previously, were clearly instrumental in their development of HIP, there are, according to Thomas Forrest Kelly, several recordings that changed the course of the 20th Century Early Music Revival: These are Nikolaus Harnoncourt’s recording of Monteverdi’s l’Orfeo, Gustav Leondhardt's recordings of the Bach Brandenburg Concertos, and Christopher Hogwood’s recording of Handel’s Messiah. 24

22(Kelly, 2011)

23(Kelly, 2011)

24(Kelly, 2011)

Harnoncourt’s Concentus Musicus Wien ensemble was critical in re-establishing creative continuo playing, emphasising renaissance instruments’ sonorities, and appropriate singing, which can be observed quite specifically on their 1968 l’Orfeo recording which can be seen to have boosted significantly the popularity of Monteverdi for many years to come as well as setting the standard for historically sensitive techniques and performances.25

Gustav Leondhardt's Consort’s 1976 recording of all 6 of Bach’s Brandenburg Concertos was critical for much of the development of many historically informed techniques, with the importance of tuning and historically sensitive techniques being highlighted, along with metrical sensitivity, and the development of small crescendos on notes for shaping: techniques that are often taken for granted today.26

Finally, Christopher Hogwood’s 1984 recording of the incredibly popular Handel’s Messiah was highly important as it was the first time that a conductor had dared to apply HIP to a work that was and is exceptionally well-known to not only musicians but the world outside of musicology. Hogwood and his Academy of Ancient Music certainly paved the way for such projects to be taken up by other ensembles in the future.27

Such recordings were critical in highlighting the benefits of HIP to the Early Music world, as well as being relatively early examples of virtuosic playing of Early Music, another standard of today, pioneered in such recordings.

“ Several recordings that changed the course of the 20th Century Early Music Revival: These are Nikolaus Harnoncourt’s recording of 'Monteverdi’s l’Orfeo', Gustav Leondhardt's recordings of the 'Bach Brandenburg Concertos', and Christopher Hogwood’s recording of 'Handel’s Messiah'.

CONCLUSIONS

The 20th Century Early Music Revival’s apparent obsession with Historically Informed Performance was an inevitability when one examines previous historical revivals and their neglection of historically informed practices: it would seem simply a matter of time before musicians wondered how music would have sounded to the ears of its composers. The modern revival has had HIP in its bloodline from its earliest proprietors like Arnold Dolmetsch and hence, the socio-political widening of musical taste allowed early music to develop and grow in popularity with the work of performing scholars like Robert Donington and Paul Hindemith, and their investigative work in researching historical performance practices. Therefore, historically informed performance can be observed in today’s early music ensembles as standard, but only thanks to the work of the pioneers of such techniques and their ability to listen, and yet remain defiant in the face of criticism, creating one of the most popular and thoroughly revived sub-genres of classical music that exists in the 21st Century.

25(Kelly, 2011)

26(Kelly, 2011)

27(Kelly, 2011)

27(Kelly, 2011)

REFERENCES

1. Blain, T. (2022, April 6). How Mendelssohn helped bring Bach's St Matthew Passion back to life. BBC Music Magazine. Retrieved from https://www.classical-music. com/composers/how-mendelssohn-helped-bring-bachs-stmatthew-passion-back-to-life/

2. Dreyfus, L. (1983). Early Music defended against its devotees. A theory of historical performance in the twentieth century. The Musical Quarterly, 69(3), 297-322. Retrieved from https://www.jstor.org/stable/742175

3. Everett-Green, R. (2000, March 8). Does he have the right stuff? [interview with Pinchas Zuckerman]. The Globe and Mail.

4. Kelly, T. F. (2011). Early Music: A Very Short Introduction (Online ed.). New York: Oxford Academic. Retrieved from https://doi.org/10.1093/

“ Historically informed performance is the performance of music using the instruments and techniques that a composer originally intended the music to be played with.

THE INFLUENCE OF CLASSICS ON WHITE SUPREMACY

JASPER MORRIS

This Independent Learning Assignment (ILA) won the ILA/ORIS Award in the Arts/Humanities category.

INTRODUCTION

The Classics, the study of Ancient Greek and Roman civilisations, have long been considered the foundations of Western culture, with literature, art and architecture all adorned with marks of the Classics. Yet, since its renewed interest during the Renaissance the study of Ancient Greek and Roman literature has found itself amalgamated with pro-white notions of the time, riddled with misinterpretations and a prejudiced subject focus.

Continued to the present, the Classics are still often weaponised to bolster racist ideologies of the far right who admire its cultural potency. Despite the pursuit of knowledge often being noble, the legacy of classical education threatens to be tarnished by those abusing its promise of being the foundation of Western society for no good. In this paper I will examine both Classics’ influence on white supremacy and societal efforts striving to reverse such influences.

INTRODUCTION

The Classics, the study of Ancient Greek and Roman civilisations, have long been considered the foundations of Western culture, with literature, art and architecture all adorned with marks of the Classics. Yet, since its renewed interest during the Renaissance the study of Ancient Greek and Roman literature has found itself amalgamated with pro-white notions of the time, riddled with misinterpretations and a prejudiced subject focus. Continued to the present, the Classics are still often weaponised to bolster racist ideologies of the far right who admire its cultural potency. Despite the pursuit of knowledge often being noble, the legacy of classical education threatens to be tarnished by those abusing its promise of being the foundation of Western society for no good. In this paper I will examine both Classics’ influence on white supremacy and societal efforts striving to reverse such influences.

Historically, Classics has held a strong appeal for politicians. Thomas Jefferson, James Garfield, and Boris Johnson all studied Classics at university. Consequently, many public speeches made by politicians have employed classical motifs to reinforce their rationality. However, in light of the prejudiced past, classical ideas have often been used to justify, on the basis of their roots in our society, racist views.

In 1968 Enoch Powell’s Rivers of Blood speech was brandished with lines from Virgil’s Aeneid to incite hatred against racial groups, particularly immigrants. The most famous line - “like the Roman, I seem to see" “the River Tiber flowing with much blood” - is almost ironic how Powell, a Classics scholar, intensifies modern racial division through comparison to the Aeneid – a story of immigration. Additionally, the comparison grows stronger as he complains that immigrants take “spouses they have never seen” much like Aeneas and Livinia. Here, Powell is promoting a zero-sum game in that one race must dominate another. Despite a clear idea of domination in ancient Rome, this was impartial of race since the empire enveloped dark-skinned Syrians to the paler Germans, all of whom were subject to the empire’s benefits and takeaways. Also, credit to Roman success was the symbiotic relations between its provinces and their citizens' ability to climb the social ladder. One marching song even joked about how Caesar first triumphed over the Gauls, then packed the senate with them! There were mixed responses to Powell’s speech. Some paraded that “Enoch was right” whereas other groups such as the Beatles insulted him as “dirty Enoch Powell” in their song Get Back (1970).

Perpetuated into modern politics, the Capitol riots in January 2022 were flourishing with classical symbols. As rioters swooped upon the neoclassical senate house, they waved Spartan flags sporting Spartan helmets also, forcing thousands of bodies through the Capitol's gate, painting a picture much alike to the battle of Thermopylae. Their obsession with Spartan culture lends to its misrepresentation by Zack Snyder’s ‘300’. The film contorts the historically complex battle of Thermopylae into the straitjacket of a Hollywood action film. Like any Hollywood-style villain, the viewer grows to resent the villainous Persians while Snyder simultaneously illustrates their effeminate, dark-skinned, barbaric nature. Therefore, the juxtaposition promotes a white Spartan force above their eastern invaders. Portraying themselves – like the Spartans - as defenders of Western culture, rioters waved flags inscribed with 'Molon Labe' (translated as come and take it) – supposedly Spartan general Leonidas’ reply to demands he lay down his weapons. Whereas none of the symbols independently advocate for white supremacy, the Capitol riots provide a fitting example for how classical ideas have been twisted to align with their racist agendas. Such agendas have been enacted by groups such as the ‘Proud Boys’ who display a Roman eagle on their shirt declaring 'SMWE' (Six million wasn’t enough – talking of the Nazi genocide of Jews) or the group ‘Stormfront’ who include the acropolis as the banner for their neo-Nazi website. Groups like these declare and admire the Greeks and Romans as the founders of white civilisation and proclaim themselves as heirs of their tradition. Furthermore, by its assimilation with the education of the white elite in the Renaissance, individuals believe, through appropriating their take on Graeco-Roman ideologies, that their viewpoints are granted intellectual credibility.

appeared in the early nineteenth century at Oxford University. The subject contained a narrow focus on the imperial powers of Athens and Rome and even George Rawlinson, an early professor of ancient history at Oxford, admitted in a manual that “Chaldea, Assyria, Babylonia…” should be included in the subject - yet little of this was seen in his teaching. Consequently, the whitewashing of Classics ensued. Nineteenth-century scholars linked modern Europeans to Ancient Greeks by skin tone to claim an ancestral connection, a fact likely flawed by considering the Ancient Greeks’ more realistic olive skin tone and further ignoring dark-skinned people in antiquity. This led to the historical fascination over the pearly white marble sculptures excavated from classical sites. These were adorned as beautifully “pure” and, as Nathaniel Hawthorne writes, were “so pure and celestial a light”, in response to their white hue. Additionally, since fragments of paint were clearly apparent on statuary, one can only attribute their ignorance of the statues' originally coloured appearance to the admiration for the racial superiority the wholly white pigment provided. Some even made erasures to perpetuate this white domination. In Euripides’ tragedy, Helen, Helen exclaims she would “shed her beauty … the way you wipe colour off a statue” although some translations have erased any reference to colour or even a statue in this section.

Classics as a degree in the United Kingdom first

A similar idea is seen in translations of the Odyssey, where Athena enhances Odysseus’ beauty by turning his facial hair 'kuaneai' (blue) and his skin darker, yet many interpret this as merely meaning 'tanned'. Therefore, if colour was core to classical identity, why is this not widely accepted? By the appropriation of classical history to suit white ideals, the integrity of the subject is lost, and the door is opened to white supremacists to wield a whitewashed Classics as the hilt of their sword against social progression.

“ Hitler aims for the rebirth of the glories of classical Greece in Nazi Germany.

Still burdened by classical education’s troubled beginnings, universities have come under fire for racism rooted in the study of Classics. Two hundred and fifty-four alumni, of June 2020, have expressed their concerns in response to Oxford’s Faculty of Classics' statement on racism, in an open letter urging the university to take more urgent action on the situation. The faculty’s statement highlights the diverse nature of the subject - 'The first Greek poetry belongs in a tradition also seen in the epics of south Asia' - and the link between 'endorsing racial superiority' and 'the conditions of the slave-run silver mines of Athens'. As a result, the university has implemented modules representative of the subject’s

diversity, exampled in the recently implemented module 'Persian Achaemenid History'. In American universities also, a statue known as ‘Silent Sam’, outside the University of Carolina, has been torn down due to its racist origins, particularly for its abhorrent reveal speech in which there are many classical allusions. The speaker proclaims that all history written about “Andromache or Penelope, of Virginia or Lucretia … was for the women of the Confederacy” and in the same speech that he had “horse-whipped a negro wench until her skirts hung in shreds”, deplorably affiliates Classics to racism in the sphere of education.

Classical art has historically been used as a measure of white beauty. In 1938, Adolf Hitler set the precedent for classical art as a weapon for white supremacy. Purchasing the ‘Lancellotti Discobolus’ for five million lire, Hitler flaunted the statue in Munich’s Glyptothek. He professed at its opening to a Nazi audience that they may “speak of progress only when we have not only attained such beauty but even, if possible, when we have surpassed it (referring to the Discobolus’ beauty)”. Therefore, the statue is reframed as a standard for perverse Nazi ideologies, and, by these new standards, Hitler aims for the rebirth of the glories of classical Greece in Nazi Germany. Further elucidated by the 1938 two-part film Olympia documenting the Berlin Olympics, Leni Riefenstahl, the director, pans the camera across the Acropolis before finally focusing on the Discobolus. The scene transitions from the classical sculpture to an Aryan male holding the same pose exhibiting the statue as a parallel for pro-white ideals. Taking a similar approach, Josiah C. Nott, and George R. Gliddon’s 1854 book ‘Types of Mankind’ incorporated a racist illustration comparing the angle between the line from one’s forehead to their upper teeth and the horizontal running along the ear’s aperture.

With Dutch anatomist Pieter Camper claiming the formula for perfect beauty lay in facial angles, the image establishes a hierarchy pinning this white unachievable 90° angle as god-like, represented by Apollo (top), and compares an image of a black skull with an ape skull, racistly assimilating the skin tone with a wild, primal background. Looking forward, an alt-right group known as 'Identity Evropa' struck US campuses with posters reading 'Protect Your Heritage' and 'Our Future Belongs to Us' before the backdrop of the same bust, as before, of Apollo Belvedere and other statues. Coupled with other racist posters, they plague Greco-Roman art with bigotry and tarnish the respect for the discipline. Such beauty ideals have also affected the canon of Western art. Originating from Giorgio Vasari’s book Lives of the Most Excellent Painters, Sculptors and Architects, he promoted artists mainly of Italian, Florentine descent, white and male. This has sustained to the present as Michelangelo, Claude Monet, Pablo Picasso, J.MW Turner, Jackson Pollock, and Andy Warhol – all white, European/American, males – dominate the Western art canon. This is coherent through classical art. When searching 'famous statues in antiquity' in Google it returns the Discobolus, Venus de Milo, Aphrodite of Knidos, and the Farnese Hercules – all sculptures depicting white figures. What about the archaic art style or sculptures depicting black figures? Why are these not famed in the same way? While admittedly the

“ As a discipline frequently subject to the question 'What is the purpose of Classics today?', It is critical to confront the bigoted abuse of Classics, so that the discipline may continue to develop and prosper.

Pelops Kore is an exception to the argument, the overall picture of classical art represents white bodies. Reasons attributed to the disproportion could be the Eurocentric beauty standards they represent, providing further realism than the archaic art style, displaying symmetry, proportion, and balance. Above all, these statues portray white, idealised figures, reinforcing a narrow and exclusionary concept of beauty. This perpetuates a hierarchy that excludes and marginalises non-white cultures and strengthens racist notions of superiority. There have, however, been efforts to reverse the negative implications of Classics on our modern world as a weapon for white supremacy. Museums have begun to make the colouration of statuary clearer with colourful reimaginations of the originals occasionally coupled with the artefacts, such as the British Museum’s Luxury and Power: Persia to Greece which displays an estimated, coloured depiction of Persian temples. Furthermore, The Glyptotek held an exhibition by Michael Armitage, The Account of an Illiterate Man

The exhibition displayed classical artefacts, including ancient Egyptian creatures and marble sculptures depicting myths such as that of the god Apollo and King Juba II of Numidia, juxtaposed by Armitage’s own work, representative of his African heritage, painting on Lubugo, a Ugandan bark cloth. As a result, Armitage is helping to start a dialogue on how classical artworks have influenced and been influenced by African culture. Moreover, even music artists have helped to spark conversations on what

art is valued in our societies. Beyonce and Jay-Z in their music video Apeshit perform in Paris’ Louvre, famous for the Mona Lisa and Venus de Milo, reinterpreting Western paintings and sculptures to celebrate black bodies in an artistic canon plagued by histories of colonialism. They achieve this by contrasting the animated, coloured figures with the frozen white figures in the past in the Louvre’s paintings. Additionally, modern scholars have been striving to re-educate Classics and to diversify the study to combat the misinformed obsession over

particularly in areas with significant cultural potency such as politics and art. The idealised identity of the Classics, manifested in Western beauty standards, architectural principles, and cultural recognition, has led to a prolonged ideological fascination over Graeco-Roman history. This, coupled with a narrow frame of mind, demonstrated by the examples above, has been used historically to inflict hatred upon racial groups. Thus, it is impossible to dismiss the detrimental legacy of Classics, which has been wielded to buttress arguments attacking immigrants

the Eurocentrism the study has provided. Martin Bernal’s Black Athena, Donna Zuckerberg’s Not All Dead White Men and Mary Lefkowitz’s Not Out of Africa: How Afrocentrism Became an Excuse to Teach Myth as History are just a few examples of books written by scholars to combat white supremacy in Classics. Overall, there is a multifaceted endeavour to reverse the inimical impact inflicted upon Classics, highlighting a collective commitment to counteract racial prejudice caused by the small few. It is apparent that Classics’ influence on white supremacy has originated from narrow means of education and contemporary misconceptions. These can be attributed to the representation of the classical world in modern forms of media and, in history, by its appropriation to suit the times’ white ideals. Additionally, such misconceptions have been stretched to encompass all sectors of society,

and African Americans and to value white identity cardinally, whilst erasing and disregarding coloured history. Despite such idealisation being firmly entrenched in the subject, a novel and, in my view, immensely worthwhile effort to reverse such effects strives to recast Classics in a more constructive manner. Gravitas has been placed upon the re-education and diversification of the subject to provide an accurate and critical assessment, avoiding the promotion of ideas sweet to the lips of the white supremacist. As a discipline frequently subject to the question 'What is the purpose of Classics today?', it is critical to confront the bigoted abuse of Classics, so that the discipline may continue to develop and prosper.

BIBLIOGRAPHY

1. https://www.ensemblepublished.com/blog/gender-andpower-relations

2. https://www.frieze.com/article/crazy-louvre-how-beyonceand-jay-z-exploit-western-art-history-ask-who-controlsblack-bodies

3. https://library.artstor.org/#/asset/SS7731005_7731005_107 84869;prevRouteTS=1678320000154

4. https://www.bbc.com/culture/article/20150324-hitlersidea-of-the-perfect-body

5. Exposed – Caroline Vout

6. http://www.perseus.tufts.edu/hopper/text?doc=Perseusti3 Atext%3A1999.02.0132%3Alife%3Djul.%3Achapterti3D80

7. https://eidolon.pub/turning-the-tables-on-dominance-anddiversity-in-Classics-2467fe45734c

8. https://eidolon.pub/a-historian-annotates-the-horrificspeech-given-at-the-dedication-of-silent-sam-96acf5cea5af

9. https://pharos.vassarspaces.net/2020/02/20/not-just-hitlerand-mussolini-neo-nazis-love-neoclassical-architecture-too/

10. Capitol Terrorists Take Inspiration from Ancient World –Pharos (vassarspaces.net)

11. https://pharos.vassarspaces.net/2021/01/14/capitol-terrorists-take-inspiration-from-ancient-world/

12. https://classicalwisdom.com/people/top-11-most-famouspeople-who-studied-the-classics/

13. https://gradesfixer.com/free-essay-examples/an-analysisof-the-historical-inaccuracies-in-the-movie-300-by-zacksnyder/

14. https://www.cbr.com/zack-snyder-300-remains-intenselyproblematic/

15. https://www.newsweek.com/nazi-amazon-proud-boysholocaust-1555192

16. https://en.wikipedia.org/wiki/Stormfront_(website)

17. https://genius.com/The-beatles-get-back-no-pakistanisversion-lyrics

18. https://confluence.gallation.nyu.edu/context/interdisciplinary-seminar/whitewashing-antiquity

19. https://www.learningforjustice.org/magazine/spring-2021/ the-classical-roots-of-white-supremacy

20. https://www.glyptoteket.com/exhibition/michael-armitage/

21. https://arthive.com/exhibitions/5835

22. https://sites.lsa.umich.edu/learn-speak-act/2018/02/15/ classics-and-the-alt-right/

23. https://www.michigandaily.com/campus-life/over-200-students-and-administrators-protest-racist-posters-foundcampus/

24. https://www.bbc.com/culture/article/20150324-hitlersidea-of-the-perfect-body

25. https://www.loc.gov/exhibits/thousand-years-of-the-persian-book/classical-persian-poetry.html

26. https://www.classics.cam.ac.uk/prospective/undergraduate/ course

27. https://www.ox.ac.uk/admissions/undergraduate/courses/ course-listing/classics

28. https://www.thedailybeast.com/molon-labe-is-the-gunloving-rights-favorite-greek-taunt

29. https://www.royalacademy.org.uk/art-artists/name/ michael-armitage-ra

30. https://talesoftimesforgotten.com/2019/11/03/debunkingthe-movie-300/

31. https://www.frieze.com/article/crazy-louvre-how-beyonceand-jay-z-exploit-western-art-history-ask-who-controlsblack-bodies

32. https://www.ensemblepublished.com/blog/gender-andpower-relations

33. https://library.artstor.org/#/asset/SS7731005_7731005_107 84869;prevRouteTS=1678320000154

34. https://eidolon.pub/turning-the-tables-on-dominance-anddiversity-in-classics-2467fe45734c

35. https://www.salon.com/2013/04/14/no_pakistanis_the_ racial_satire_the_beatles_dont_want_you_to_hear/

36. https://pharos.vassarspaces.net/2020/10/01/proud-boysancient-western-civilization-trump-presidential-debates/

37. https://www.massolit.io/courses/homer-women/helen-inthe-iliad?autoplay=true

38. https://talesoftimesforgotten.com/2021/01/08/heres-whatthe-costumes-and-flags-on-display-at-the-pro-trump-insurrection-mean/

39. https://pharos.vassarspaces.net/onomasticon-of-classicalpseudonyms-and-avatars/

40. https://pharos.vassarspaces.net/2018/06/15/spqr-andwhite-nationalism/

41. https://www.adl.org/resources/blog/proud-boys-bigotryfull-display

42. https://www.justsecurity.org/72339/qanon-is-a-nazi-cultrebranded/

43. https://www.historyextra.com/period/modern/classicsracial-hatred-how-far-right-hijacked-ancient-greco-romanworld/

44. British Museum’s Luxury and Power: Persia to Greece Exhibition

45. The Louvre

46. Caroline Vout - Talks on classical art.

ARE ECONOMIC NETWORKS THE KEY TO SUCCESSFUL BUSINESS MANAGEMENT ?

A Case Study of Google and Amazon

ARTS HUMANITIES

XAVIER WALLIN

This Independent Learning Assignment (ILA) was short-listed for the ILA/ORIS Presentation Evening.

INTRODUCTION

The field of network economics has transformed business management over the last fifty years by highlighting the importance of networks, coordination and relationships within organisations in achieving efficiency, growth and innovation. A better understanding of the dynamics and principles of networks has enabled firms worldwide to navigate complex internal business ecosystems (that is, the management and utilisation of staff). In the case of Amazon.com Inc. ('Amazon'), its management’s

ability to embrace network theory in order to adapt to transitions in market dynamics has played a large role towards the company’s success. This paper will discuss the impact of network economics (network theory) on the management of firms and organisations generally with reference to the impact on Alphabet Inc (otherwise known as 'Google') and Amazon in particular.

“ This essay will show that the cultivation of strong networks within firms enables them to maximise the efficiency of their internal business ecosystems.

1. What is an Economic Network?

An economic network can be described as a combination of 'individuals, groups, or countries interacting to benefit the community as a whole'1 where the 'community' consists of the network members. The field of network economics focuses on how these networks impact economic outcomes such as production, innovation and exchange within the relevant community. A classic example of an economic network is a chamber of commerce or a department within a firm, in which departmental members are all working towards a specific goal or project outcome.

This essay will show that the cultivation of strong networks within firms enables them to maximise the efficiency of their internal business ecosystems and allow team members to adapt better to their roles and implement prescribed processes within the firm. Such benefits can be achieved due to: (i) networks’ impact on information flows in organisations, (ii) the advantages arising from decentralised management structures and decision making, and (iii) the advantages of increased social capital (being the potential benefits or resources that individuals can derive from relationships).

2. Information Flows

First, the connectivity of networks influences the development of information and communication flows within organisations. We can analyse the benefits of sparse networks versus dense networks by using a network diagram (Figure 1), containing

7 nodes (A, B, C et cetera) representing agents within an organisation (that is, employees) and edges (arrows) representing connections between different nodes that information can flow through. An example of a sparse economic network could be a collection of farmers in a rural/remote area where the farmers have few employees and fewer connections or means to exchange information (new farming techniques/crop yield/weather) between staff or other agricultural professionals. On the other hand, we can imagine a dense employee network in a tech firm, such as Google where team members work collaboratively with frequently used communication channels to come up with ideas and then implement them. Figure 1 below shows that denser networks allow information to flow more easily between the nodes than in sparser networks where information must first travel through other nodes to reach its target. It follows that the denser, more interconnected networks have more efficient communication flows than sparser, less developed networks and information bottlenecks2 are significantly reduced.

1Halton, Clay, Economic Network, October 2022, https://www.investopedia.com/terms/e/economic-network.asp

2Bottleneck: refers to a point within the network where the flow of information is restricted or constrained leading to an overall less efficient information exchange process.

2.1. Information Flow at Google and Amazon

In relation to organisations, greater network density is reflective of management structures where team members (employees) actively communicate and transmit information (simply put - teamwork) that inevitably results in greater information diffusion, analysis and evolution. Consequently, the beneficial effects of denser networks are:

1. Quicker and better-informed decision making as greater access to information allows firms to swiftly respond to market changes and capitalise on opportunities.

2. Resource allocation becomes more accurate as firms gain a better understanding of where resources are in high demand and how they can be utilised.

3. Collaboration is enhanced as better access to information promotes knowledge sharing among employees and an environment which fosters teamwork and collaboration; and

4. Innovation increases as greater diversity of thought, ideas and insights among employees and teams all contribute to creativity.

These four benefits associated with increased information diffusion have been successfully leveraged by Google. The trillion-dollar corporation is renowned for its open communication culture5 as demonstrated by their CEO, Sundar Pichai, who said “I value teamwork quite a bit and setting up collaborative cultures is another big thing I’ve been trying to focus on”6 . This is reflected in Google’s fun, vibrant and stimulating offices which encourage a culture of transparency, open dialogue and innovation in the workplace. Equally Google has developed a suite of communication tools such as Gmail, Google drives, Google docs and Google chat which enable agents within Google’s internal network to easily share ideas, collaborate and make quick decisions. Additionally, these products have benefited Google customers to increase information flows within their own organisations. Amazon has also benefitted from the use of a successful information flow system. For example, the Amazon information flow system connecting the organisation to its customers is based on extensive customer data analysis and encouraging consumer feedback. Using these data and information tools, the Amazon group has managed to improve where and how resources should be allocated most effectively.

3The network density statistic is the proportion of possible edges that are actually present in the network. The sparse network has only 7 edges present out of a possible 42: 7/42 = 0.17. In the dense network on the other hand, all 42 edges are utilised.

4Network density, 04/03/2021, IBM, https://www.ibm.com/docs/en/spss-modeler/18.0.0?topic=networks-network-density

5Wickre, Karen, What Google's Open Communication Culture Is Really Like, August 2017, Wired.com, https://www.wired.com/story/ what-googles-open-communication-culture-is-really-like/

6Pichai Sundar, as cited by Sadun, Raffaella, Google’s Secret Formula for Management? Doing the Basics Well, August 2017, Harvard Business Review, https://hbr.org/2017/08/googles-secret-formula-for-management-doing-the-basics-well 2017, Harvard Business Review, https://hbr.org/2017/08/googles-secret-formula-for-management-doing-the-basics-well

Figure 1 : sparse versus dense networks4

The outcome is that Amazon can optimise inventory management and swiftly direct resources towards or away from the production of items following changes in consumer tastes or trends. Internally, Amazon employees will enjoy enhanced personal productivity as customer information flows translate into increased sales.

2.2. Measuring Information Flow and Employee Performance

The Google and Amazon examples demonstrate that networks of increased connectivity can lead to greater information diffusion and the four outcomes listed above which in turn have a significant impact on employee productivity and performance. However how can this impact be measured? Measuring information diffusion between agents can be challenging as studies tend to retrieve data through surveys and interviews. A study by Coleman, Katz and Menzel explored how information about the adoption of a new drug would flow between doctors in a region. In order to gauge diffusion information about the drug, various doctors were surveyed and asked how and when they found out about the new drug in order to establish how quickly information was transmitted.

and their objectives and key results framework. In measuring the performance of warehouse personnel, Amazon use measurement metrics such as the number of units packed or shipped. The resulting data in each case is specific and accurate (unlike surveys).

2.3. Downsides of Greater Information Flow

Whereas data from surveys is imprecise and sometimes unreliable, employee productivity and performance can be easily measured using a variety of quantitative metrics such as sales revenue, customer satisfaction and project completion rates. For example, Google measures employee performance through project-based assessments

However, excessive information diffusion and knowledge sharing can have disadvantages for employees and organisations. As communication within employee networks reach too high levels, network agents may suffer from information overload as employees struggle to filter and prioritise information most valuable to the network’s goal. Additionally, increased information flow in firms raises concerns of a lack of focus and productivity with excessive communication diverting employees’ attention away from primary tasks and responsibilities. Contrary to Google’s policy, a range of large corporations such as Goldman Sachs, JP Morgan Chase and IBM have taken steps to restrict social media and other communication tools in the workplace. Despite the benefits of these platforms in sharing key insights and knowledge with employees, recent studies (e.g., Ahmad 20228) have found that social media is responsible for losses of up to 9.5% in employee productivity. Given this data, it is difficult to justify the recommendation that managers should encourage the use of these platforms in the workplace. Moreover, increased information flow in employee networks also increases the risks of misinformation and rumours. As a consequence, a

7Coleman, J.S., Katz, E., and Menzel, H., Medical Innovation: A Diffusion Study, Social Forces, Volume 46, Issue 2, December 1967, Page 291.

8Ahmad, M.B., Hussain, A. & Ahmad, F. The use of social media at workplace and its influence on the productivity of the employees in the era of Covid. SN Bus Econ 2, 156 (2022). https://doi.org/10.1007/s43546-022-00335-x

9Peer to peer communication is communication that takes place at the same levels of an organisational hierarch.

fall in productivity could occur in certain cases due to confusion and misunderstandings but it is difficult to suggest that such instances would cause a net decrease in productivity, given the benefits outlined above.

3. Centralised Versus Decentralised Management Structures

Another factor to consider when observing the impact of management structures on information flows within firms is the degree of centralisation used by the relevant firm. A management structure is an economic network. Centralised structures have a clear hierarchal form and information primarily flows from the top down. In networks of this sort, decision making takes place far away from points of specific action. An example is McDonald’s where decision making power is concentrated at corporate headquarters and individual restaurants follow standardised systems and instructions. Conversely, decentralised management structures have a reduced hierarchal structure, and each department or /node interacts equally and autonomously within the network (organisation) using peer to peer or lateral communication.

within the organisation acts and communicates autonomously and is marked by the absence of teams. While distributed management structures are relatively rare and not yet used by any major corporations, Google is a well-known example of a successful decentralised management network which has incorporated elements of distributed networks. At Google, authority is dispersed among various departments and units, yet the company still provides its employees with autonomy, allowing them to make decisions within areas of their expertise.

The principal advantage of centralised management structures is that they can reduce administrative costs and they set out a clear chain of command. This is highlighted by McDonald’s management network, where clear, standardised operations have reduced the firm’s coordination costs, helping McDonald’s establish themselves as preeminent in cost leadership in its industry.

“ Decentralised management structures lack the rigidity of centralised networks as decision-making processes are delegated throughout the organisation at departmental level to allow for more localised decision making.

On the other hand, decentralised management structures lack the rigidity of centralised networks as decision-making processes are delegated throughout the organisation at departmental level to allow for more localised decision making. A distributed management structure is an extreme example or sub-category of a decentralised management structure. In a distributed network each individual

However, recent empirical work concerning long-term healthcare teams in Nova Scotia, Canada between 1971 and 201110 concluded that centralised networks can result in less efficient information diffusion. This is largely due to information bottlenecks because information must pass through the central hub where managers can be overwhelmed with decisions. In essence, managers can lack the capacity to process information streams leading to systemic computational weaknesses within the organisation. The consequence is that centralised organisations may not always be able to fully access the benefits of efficient information flows highlighted above.

10Scheuer, Cara-Lynn & Voltan, Annika & Chakraborty, Subhajit. (2021). Exploring the impact of decentralized leadership on knowledge sharing and work hindrance networks in healthcare teams. Journal of Management & Organization. 29. 10.1017/jmo.2020.37.

The centralised network shown in Figure 2 below is an extreme model where an absence of edges implies that colleagues do not speak at all and have no authority to make decisions. In contrast, the degree of interconnectedness in Figure 2(a) and Figure 2(b) highlight how agents/nodes in distributed and decentralised networks have quicker access to information coming from a range of sources.

Fig 2: centralised, decentralised and distributed networks 11

As a result, organisation structural reform is becoming increasingly common as “skeletal hierarchies are now overlaid with flexible networks and self-organizing project teams”12, rather than traditional, rigid, bureaucratic offices. An early adopter of decentralisation was General Electric Company ('GE'), an American multinational conglomerate. In the 1980’s, GE implemented a new decentralised business strategy, delegating more authority to specific business units, which resulted in “faster decision making and greater responsiveness to the external environment”13 and in turn increased competitiveness and financial performance. GE’s philosophy of a corporation without boundaries14 influenced other major firms such as IBM, Toyota and Unilever to follow suit in the 90s.

3.1. Amazon’s Decentralised Management Structure

Despite having a general framework resembling a traditional hierarchal corporate structure, Amazon now incorporates thousands of hybrid project groups and appears to contain many of the features associated with decentralised management networks. Former Senior Manager at Amazon, Eric Heller, commented on how Amazon’s highly decentralised structure, is the equivalent of “1000 independent businesses, all marching in the same direction”15. The company, which employs more than 1.5 million people worldwide16, encourages “communication between various elements within its ecosystem”17 and building strong networks among its employees. In doing so, Amazon has become a market leader in business diversification largely due to the flexibility and dynamism of its networks in response to market changes.

11Fichtner, Laura, Techno-Politics as Network(ed) Struggles, Sciforum conference ISIS Summit Vienna 2015, published on 30 June 2015 by MDPI https://doi.org/10.3390/isi-summit-vienna-2015-S3033

12Hendry, John, Management - A Very Short Introduction, Oxford University Press (2013), pp. 20 – 21.

13Grant, Robert M., Case Sixteen – General Electric: Life after Jack, Contemporary Strategy Analysis, John Wiley & Sons (2021) p 129.

14Gilmore, Thomas., and Hirschorn, Larry., The New Boundaries of the 'Boundaryless' Company Harvard Business Review, May – June 1992.

15Heller, Eric as cited in Stevens, Laura, For Amazon, Can two Headquarters still Equal One Culture?, The Wall Street Journal, 15 September 2017, https://www.wsj.com/articles/amazon-can-two-headquarters-still-equal-one-culture-1505467803

16Amazon.com quarterly financial report (December 2022), February 2023, https://s2.q4cdn.com/299287126/files/doc_ financials/2022/q4/Q4-2022-Amazon-Earnings-Release.pdf, p. 14

17Dudovskiy, John, Amazon.com Inc. Report 2022, Business Research Methodology, https://research-methodology.net/ amazon-report-2/

3.2. Amazon - Room To Improve?

However, Amazon’s management structure could be criticised for a lack of collaboration and communication between specific teams and departments. Despite Amazon’s efforts to decentralise and develop dense networks within departments, Forbes Magazine speculates that silos are killing Amazon’s potential18. The company incorporates thousands of teams and departments yet is significantly suffering from a silo effect19 as networks and information diffusion between teams are underdeveloped. As a result, cross-functional knowledge sharing, and collaboration is suboptimal. This is partially attributed to departments receiving too much autonomy, but it is also fuelled by competition and performance pressures between teams. Amazon should strive to develop greater connectivity between currently independent teams and reduce unnecessary competition between them in order to encourage the sharing of key insights and expertise to increase overall organisational performance.

3.3. Decentralise and Thrive?

Should firms seek to decentralise? Well, it depends. Given that denser, more flexible networks foster an environment of innovation and collaboration, for tech start-ups, companies or research and development facilities it would seem irrational for them not to adopt a decentralised approach. However, in other industries, such as manufacturing for example, a centralised structure may be more appropriate. In fields of work that require little innovation or dynamism, the discounted administrative costs, stability and coordination that arise from centralisation (standardisation, economies of scale, less duplicative errors) may be more beneficial. During the Covid global pandemic, the NHS’s centralised structure was essential in providing the UK with one of the quickest and most effective vaccine rollouts around the world. With increased coordination, stemming from centralisation and standardised procedures, the NHS was able to outperform similar economies such as France and the Netherlands in supplying and distributing the vaccine. Therefore, although there is no definite answer to the question of centralisation for organisations, its impact on the direction firms take is significant.

18Boss, Jeff, Silos Are Killing Amazon's Potential. Don't Let Them Kill Yours, Forbes Magazine, September 2017, Forbes.com, https:// www.forbes.com/sites/jeffboss/2017/09/18/silos-are-killing-amazons-potential-dont-let-them-kill-yours/ 19Silo effect or silo mentality is a reluctance to share information with other employees within the same firm resulting in reduced organisational efficiency.

4. Social Capital

Developing strong networks in organisations also contributes to the formation of improved social capital. Some academics describe social capital as the potential benefits or resources that individuals can derive from relationships20. Others have chosen a broader definition that includes trust and norms21 existing within social relationships. Organisations in which employees are encouraged to communicate with others, share ideas or perhaps take part in social activities outside the workplace should, in theory, accumulate greater social capital. When agents within organisations have strong ties22 or a sense of shared identity within the network (greater social capital), they are more likely to cooperate, support each other and work towards the common goal of the team. A study exploring the impact of social capital on organisations in Ghana23, found that social capital had a significant relationship with organisational performance and that Ghanaian firms with higher levels of social capital24 reported higher productivity at employee level and also at firm level.

4.1.'Strong Ties' Versus 'Weak Ties'

However, despite academic research showing the benefits of social capital in organisations such as improved team performance, increased information diffusion and greater innovation (published by Brass, Kilduff and Mehra in 200125 and Hanson and Krackhart in 199326), Mark Granovetter’s paper The Strength of Weak ties concludes that low levels of social capital with numerous agents can also be beneficial. His idea is that weak ties27 provide unique advantages by connecting employees to broader networks and diverse opportunities that cannot be accessed by stronger ties. The theory is based on the assumption that the number of stronger ties will necessarily be fewer than a large number of weaker ties. Granovetter reached this conclusion by interviewing people in Massachusetts to find out the degree to which people found employment via social contacts. He divided such contacts into three categories according to how frequently individuals interacted together. Strong equated to interactions at least twice a week, medium equated to interactions less than twice a week but more than once a year and weak equated to one interaction per year or less28. His research found that while only 16.7%29 found a job through strong ties, 27.6%30 had done so using weak ties (and 55.7% with medium ties), confirming his hypothesis. This study, backed by

20Bourdieu, Pierre and Wacquant, Loïc., An Invitation to Reflexive Sociology, University of Chicago Press, (1992) Chicago.

21Putnam, Robert D., Bowling Alone: The Collapse and Revival of American Community, 2001, New York, Simon & Schuster Paperbacks, (2001), New York, p. 19.

22Strong ties: term coined by sociologist Mark Granovetter in his paper The strength of weak ties (1973) to describe close and intimate relationships of frequent interaction.

23Ofori, Dan and Sackey, Jocelyn, Assessing Social Capital for Organisational Performance: Initial Exploratory Insights from Ghana, Organisations and Markets in Emerging Economies, Vol. 1(2), pp. 71-91.

24Social capital measured as a collection of variables (employee survey) including Reciprocity, Trust, Institutional ties, Diversity of Contacts, Density of Personal Contacts and social ties.

25Brass, Daniel J., Kilduff, Martion and Mehra, Ajay, The social networks of high and low self-monitors: Implications for workplace performance, Administrative Science Quarterly,46(1) (2001) pp. 121–146. https://doi.org/10.2307/2667127.

26Hanson, Jeffrey R. and Krackhart, David, Informal networks: The company behind the chart, Harvard Business Review, July - August 1993.

27A weak tie is a social relationship characterised by a lower frequency of interactions.

28Granovetter, Marc, The Strength of Weak ties, American Journal of Sociology, 1973, as reproduced in Jackson, Matthew O., Social and Economic Networks, February 2008, Princeton University Press, 101.

29Ibid.

30Ibid.

numerous follow-up studies (such as Friedkin in 198031) concludes the importance of weak ties in the diffusion of information and forming bridges between networks that have fewer connections.

However, Granovetter’s work primarily focuses on social networks rather than economic networks and the validity of weak ties theory as a case for organisational reform could be questioned. Yet Granovetter’s theory has been supported by Morton Hanson’s study which sourced data from a large electronics company with 120 new product development projects undertaken by 41 divisions. The study explored the importance of weak ties in knowledge sharing across organisation sub-units and concluded that weak cross-unit ties helped projects acquire useful knowledge from other units.32

4.2. Social Capital at Google and Amazon

Google’s employee networks demonstrate successful development of social capital. The company’s fun and relaxed work culture provides the environment needed for information diffusion to improve (as seen above) but also to build trust and confidence between employees. Over the last two decades, Google have adopted a range of successful policies to develop social capital such as their Thank God It’s Friday meetings (TGIF)33 and allowing employees to play ping-pong or even volleyball at work. Moreover, the company’s decision to opt

“ Google’s employee networks demonstrate successful development of social capital.

“ Although Amazon have claimed that they use a performance-based system, policies of reprimanding idle time lasting only a few minutes do not provide the environment for significant social capital development and its benefits.

for open office layouts means that weak ties form frequently and everywhere, whether it be waiting in line at one of Google’s many office café’s or while taking in part in a free workout class.

In contrast, it would not appear that Amazon has optimised social capital levels within its internal networks in the same way. The corporation has faced recent controversy regarding poor working conditions with workers commenting that their bathroom breaks are timed and that even robots “are treated better than us”34. Although Amazon have claimed that they use a performance-based system, policies of reprimanding idle time lasting only a few minutes do not provide the environment for significant social capital development and its benefits. However, it should be noted that the majority of such criticisms are directed towards Amazon’s distribution centre personnel, rather than those involved in fulfilment and operations management, where social capital is most beneficial. In such divisions, given limited data of Amazon’s internal work culture, it is difficult to gauge social capital levels within the company’s offices.

31Friedkin, N.E., A Test of Structural features of Granovetter’s Strength of Weak ties Theory, Social Networks, 2, 411-422.

32Hansen, Morten T. The Search-Transfer Problem: The Role of Weak ties in Sharing Knowledge across Organization Subunits Administrative Science Quarterly, vol. 44, no. 1, 1999, pp. 82–111. JSTOR, https://doi.org/10.2307/2667032. Accessed 18 June 2023.

33O’Callahan, Ted, What's the Google approach to human capital?, Management in Practice, Yale Insights, March 2011, https:// insights.som.yale.edu/insights/whats-the-google-approach-to-human-capital

34Conway, Zoe & Dearbail, Jordan, Amazon strikes: Workers claim their toilet breaks are timed.Published, BBC News Business, January 2023, https://www.bbc.co.uk/news/business-64384287

4.3. Impact of Social Media Platforms on Social Capital

Despite evidence highlighting the dangers of social media on employee productivity, it is difficult to deny its benefits for the development of social capital in organisations. A study taking data from 328 employees from the telecommunications company Tunisie Telecom35, revealed that social media usage in the workplace had positively influenced social capital levels and had also improved information flow, improving employee performance. Subsequently, managers should consider loosening social media policy in the workplace (except for where significant falls in productivity/focus occur), particularly given the rise of more business oriented social media platforms like LinkedIn.

4.4. What Should Managers Strive for (Social Capital)?

Considering the advantages of both weak ties and social capital, managers should seek to consider their own organisations’ unique needs and culture when implementing or altering a management structure. They should encourage collaboration and communication among individuals within teams and departments to build social capital in order to improve productivity and coordination within teams. However, for highly innovative industries, managers should also ensure that weak ties and external networks are expanded (with distant groups and individuals outside and inside the firm) to enable the flow of information that might not be immediately accessible to agents within the network. Ultimately,

35Hanen Louati & Slim Hadoussa (2021) Study of Social media impacts on social capital and employee performance - evidence from Tunisia Telecom, Journal of Decision Systems, 30:2-3, 118-149, DOI: 10.1080/12460125.2021.1872142

36Parsons, John and Weddle, Brooke, Social capital: Build back better relationships at work, October 2022, https://www.mckinsey.com/ capabilities/people-and-organizational-performance/our-insights/social-capital-build-back-better-relationships-at-work

37Consultancy.uk, The world's most innovative companies (according to BCG), September 2022, https://www.consultancy.uk/

following a 'dearth of social capital'36 stemming from the Covid pandemic, organisations should take internal action (such as organising employee social events, limiting working from home) and encourage external networking (with trade bodies, clients, customers and suppliers) to reverse sinking levels of social capital in order to achieve greater long-run employee and organisational performance.

CONCLUSION

I believe that the answer to the question 'Are Economic Networks the Key to Successful Business Management?', Simply put is yes. Although the field of network economics remains an evolving area of study, it is one that in the modern era all organisations should consider carefully in order to optimise their operational efficiency, innovation and continued development.

Google’s example demonstrates how the development of strong internal networks can contribute to significant advances in innovation. Its management strategy of developing information flows, social capital and highly decentralised decision-making has helped Alphabet Inc (Google’s parent company) sit amongst the top 5 most innovative companies of 202237

“ Emerging start-ups should endeavour to adopt strategies that encourage information flow and the development of social capital within their core business ecosystems (e.g., product development, fulfilment and operations management).

On the other hand, despite Amazon’s successes in decentralising, the company has not achieved levels of social capital and information diffusion comparable to those of Google. In particular, Amazon should seek to eliminate silo effects from their business model, allowing specific teams and departments to work more innovatively and productively. Alternatively however, Amazon’s absence of social capital and information sharing in its distribution centres highlights how low levels of social capital can be beneficial, when output is the principal concern. Therefore, emerging start-ups should endeavour to adopt strategies that encourage information flow and the development of social capital within their core business ecosystems (e.g., product development, fulfilment and operations management). Furthered by decentralised decision making, such policies will result in more creative, innovative and flexible work environment. When considering management strategies for manufacturing and distribution processes, centralised structures and limited social capital development may be more appropriate as standardised systems can be rolled out to maximise output.

36Parsons, John and Weddle, Brooke, Social capital: Build back better relationships at work, October 2022, https://www.mckinsey.com/ capabilities/people-and-organizational-performance/our-insights/social-capital-build-back-better-relationships-at-work

37Consultancy.uk, The world's most innovative companies (according to BCG), September 2022, https://www.consultancy.uk/ news/32425/the-worlds-most-innovative-companies-according-to-bcg#:~:text=Forti20the%20second%20consecutive%20 year,2021%2Deditionti20of%20the%20list

BIBLIOGRAPHY

1. Ahmad, M.B., Hussain, A. & Ahmad, F. The use of social media at workplace and its influence on the productivity of the employees in the era of Covid. SN Bus Econ 2, 156 (2022). https://doi.org/10.1007/s43546-022-00335-x

2. Amazon.com quarterly financial report (December 2022), February 2023, https://s2.q4cdn.com/299287126/files/ doc_financials/2022/q4/Q4-2022-Amazon-Earnings-Release. pdf, p. 14

3. Boss, Jeff, Silos Are Killing Amazon's Potential. Don't Let Them Kill Yours, Forbes Magazine, September 2017, Forbes. com, https://www.forbes.com/sites/jeffboss/2017/09/18/ silos-are-killing-amazons-potential-dont-let-them-kill-yours/

4. Bourdieu, Pierre and Wacquant, Loïc., An Invitation to Reflexive Sociology, University of Chicago Press, (1992) Chicago.

5. Brass, Daniel J., Kilduff, Martion and Mehra, Ajay, The social networks of high and low self-monitors: Implications for workplace performance, Administrative Science Quarterly,46(1) (2001) pp. 121–146. https://doi. org/10.2307/2667127

6. Campbell, Andrew., Kunisch, Sven and Müller-Stewens, Günter., To centralise or not to centralise, 2011, https:// www.mckinsey.com/capabilities/people-and-organizationalperformance/our-insights/to-centralize-or-not-to-centralize

7. Coleman, J.S., Katz, E., and Menzel, H., Medical Innovation: A Diffusion Study, Social Forces, Volume 46, Issue 2, December 1967, Page 291.

8. Consultancy.uk, The world's most innovative companies (according to BCG), September 2022, https://www. consultancy.uk/news/32425/the-worlds-most-innovativecompanies-according-to-bcg#:~:text=Forti20the%20 second%20consecutive%20year,2021%2Deditionti20of%20 the%20list

9. Conway, Zoe & Dearbail, Jordan, Amazon strikes: Workers claim their toilet breaks are timed.

10. Published, BBC News Business, January 2023, https://www. bbc.co.uk/news/business-64384287

11. Dudovskiy, John, Amazon.com Inc. Report 2022, Business Research Methodology, https://research-methodology.net/ amazon-report-2/

12. Fichtner, Laura, Techno-Politics as Network(ed) Struggles, Sciforum conference ISIS Summit Vienna 2015, published on 30 June 2015 by MDPI https://doi.org/10.3390/isi-summitvienna-2015-S3033

13. Friedkin, N.E., A Test of Structural features of Granovetter’s Strength of Weak ties Theory, Social Networks, 2, 411-422.

14. Gilmore, Thomas., and Hirschorn, Larry., The New Boundaries of the 'Boundaryless' Company Harvard Business Review, May – June 1992.

15. Granovetter, Marc, The Strength of Weak Ties, American Journal of Sociology, 1973, as reproduced in Jackson, Matthew O., Social and Economic Networks, February 2008, Princeton University Press, 101.

16. Grant, Robert M., Case Sixteen – General Electric: Life after Jack, Contemporary Strategy Analysis, John Wiley & Sons (2021) p 129.

17. Halton, Clay, Economic Network, October 2022, https:// www.investopedia.com/terms/e/economic-network.asp

18. Hanen Louati & Slim Hadoussa (2021) Study of Social media impacts on social capital and employee performanceevidence from Tunisia Telecom, Journal of Decision Systems, 30:2-3, 118-149, DOI: 10.1080/12460125.2021.1872142

19. Hansen, Morten T. The Search-Transfer Problem: The Role of Weak Ties in Sharing Knowledge across Organization Subunits Administrative Science Quarterly, vol. 44, no. 1, 1999, pp. 82–111. JSTOR, https://doi.org/10.2307/2667032. Accessed 18 June 2023.

20. Hanson, Jeffrey R. and Krackhart, David, Informal networks: The company behind the chart, Harvard Business Review, July - August 1993.

21. Heller, Eric as cited in Stevens, Laura, For Amazon, Can two Headquarters still Equal One Culture?, The Wall Street Journal, 15 September 2017, https://www.wsj.com/ articles/amazon-can-two-headquarters-still-equal-oneculture-1505467803.

22. Hendry, John, Management - A Very Short Introduction, Oxford University Press (2013), pp. 20 – 21.

23. Network density, 04/03/2021, IBM, https://www.ibm.com/ docs/en/spss-modeler/18.0.0?topic=networks-networkdensity

24. O’Callahan, Ted, What's the Google approach to human capital?, Management in Practice, Yale Insights, March 2011, https://insights.som.yale.edu/insights/whats-the-googleapproach-to-human-capital

25. Ofori, Dan and Sackey, Jocelyn, Assessing Social Capital for Organisational Performance: Initial Exploratory Insights from Ghana, Organisations and Markets in Emerging Economies, Vol. 1(2), pp. 71-91.

26. Parsons, John and Weddle, Brooke, Social capital: Build back better relationships at work, October 2022, https:// www.mckinsey.com/capabilities/people-and-organizationalperformance/our-insights/social-capital-build-back-betterrelationships-at-work

27. Pichai Sundar, as cited by Sadun, Raffaella, Google’s Secret Formula for Management? Doing the Basics Well, August 2017, Harvard Business Review, https://hbr.org/2017/08/ googles-secret-formula-for-management-doing-the-basicswell

28. Putnam, Robert D., Bowling Alone: The Collapse and Revival of American Community, 2001, New York, Simon & Schuster Paperbacks, (2001), New York, p. 19.

29. Scheuer, Cara-Lynn & Voltan, Annika & Chakraborty, Subhajit. (2021). Exploring the impact of decentralized leadership on knowledge sharing and work hindrance networks in healthcare teams. Journal of Management & Organization. 29. 10.1017/jmo.2020.37.

30. Wickre, Karen, What Google's Open Communication Culture Is Really Like, August 2017, Wired.com, https://www.wired. com/story/what-googles-open-communication-culture-isreally-like/

Royal Grammar School

High Street

Guildford

GU1 3BB