Ioana Man

Page 1

The Durability of Engineered Prediction Formulas, recipes, specifications and the rogue checklist

Ioana Man Fourth Year 2017 Architecture’s Truths / Architecture’s Lies - Francesca Hughes


Predictive truth is better truth and long-lasting predictive truth is even better truth. 1

How long can truth last and how can we make it last longer? The desire to increase the lifespan of truths is apparent through the multitude of formats that have been devised to do so and through their relentless adoption for this purpose. We use tested techniques on a daily basis because we trust that they will help us avoid wrong outcomes. We also trust that they will keep doing so over time. Formulas, recipes, specifications and checklists alike claim to be durable means of engineering prediction that resist failure; their shelf-life is claimed to be long, but how long is it actually? When do they expire and how might one tamper with their expiration date? While truth creation relies on the precision of its logical syllogisms, durability is measured by different standards and creates different guidelines for truth production.

1 This premise is informed by the work of John von Neumann: “All processes that are stable we shall predict. All processes that are unstable we shall control.� and the writing of Karl Popper, specifically his essay Truth, Rationality and the Growth of Scientific Knowledge in Conjectures and Refutations.Routledge Classics. p.291-326


F1. The Formula Formulas are of central importance in scientific and pseudo-scientific contexts. Bruno Latour describes formulas as attempts to “isolate, characterize, synthesize and understand”2; formulas “are meant to explain”3. Formulas do not only represent the true nature of a theory, but they are also valuable tools for safeguarding against future falsehoods surrounding the law. In what follows I will regard the formula as the beacon of theoretical laws. Formulas collapse and compress centuries of experience. This is not only to be seen in the process of establishment and validation, but also in all subsequent deployments of a given formula. Any formula that we regard as true is the result of a theory being transformed into a scientific law, into a fact producing system and as Latour remarks this is a process that is itself highly constructed4. It is a conscious process of creating order out of disorder by means of establishing a methodology, and then by proving its validity. So, the dissemination of a formula relies on a certain level of consensus in the community it stems from, and according to Baigrie and Hattiangadi “Where there is a consensus, what we describe as ‘true’ is typically something to which we all give our consent, and it is true in the sense that we conventionally call it ‘true’.”5 Thus, a true formula is a formula that a scientific community regards as true. As opposed to an equation, a formula in a more narrow sense is “something that ‘gives you an output, whenever you give it an input”6. One uses a specific formula to solve a specific problem or to demonstrate a specific premise. This intrinsic character of the formula is set out by Wittgenstein in his Remarks on the Foundation of Mathematics We use the expression: “The steps are determined by the formula...”. How is it used?--We may perhaps refer to the fact that people are brought by their education (training) so to use the formula y = x2, that they all work out the same value for y when they substitute the same number for x. Or we may say: “These people are so trained that they all take the same step at the same point when they receive the order ‘add 3’”.7

The validity of a formula is then not only dependent on experience at its genesis, but also at its point of use. The choice of a formula and its deployment is highly dependent on the experience and past conditioning of the user. As explained by Wittgenstein, the statement “‘The formula determines . . .’ can be used as a description of the behaviour of people or a description of the formula”.8 This is to say that the formula determines a specific outcome only if the user has been trained to deploy the formula in order to achieve that specific outcome. For instance, if I were not to understand arithmetic multiplication, the formula c= a * b would not determine a true product. In this sense, the formula is easily susceptible to misuse should 2 Bruno Latour and Steve Woolgar . 1981. Laboratory Life. Princeton University Press. p.56 3 Nancy Cartwright. 1983. How the Laws of Physics Lie. Oxford University Press. p.3 4 Bruno Latour, and Woolgar Steve. 1981. Laboratory Life. Princeton University Press. 5 Brian Baigrie and J. N Hattiangadi. 1992. “On Consensus and Stability in Science.” The British Journal for the Philosophy of Science, 435-458., p. 475 6 Ramharter, Esther. 2014. “Wittgenstein on Formulae.” Grazer Philosophische Studien 79-91, p.79 7 Ludwig Wittgenstein. 2001. Remarks on the foundations of mathematics. Oxford: Wiley-Blackwell, p.34 8 Ibid, p.35


the experience of the user not encompass the expected knowledge for correct use. Formulas are mathematical representations of theoretical laws that were constructed under specific circumstances and as such carry along the initial caveats of theoretical laws. According to Cartwright every theory that was proposed in physics, and as a consequence every formula that comes with it, was known to be deficient and rely on approximations. “in fact most of the high-level claims in science are ceteris paribus generalizations, which are false unless certain precise conditions obtain.”9 So, as soon as these conditions that were seen as fixed and necessary in the establishment of the formula are not fixed or present, the formula loses its power to safeguard against falsehoods. If “fundamental equations do not govern objects in reality; they govern only objects in models”, then formulas cannot explain situations that are not identical to their genesis model. The formula allows for the mathematisation of a theoretical law; it attaches a mathematical equation to an abstract concept which enables it to calculate further truths. As such, the formula does not only serve as a source of legitimacy for the more abstract law, but it also ensures both its corrective and predictive power. Taken as the true manifestation of a true theory, we immediately see the outcomes of a given formula as also necessarily true. In this sense, the formula is unable to accommodate aberrant data, as false results are seen as consequences of application errors. The formula predicts data. Facts that do not match the expected outcomes cannot be fit into the formula and must be eliminated. As such, formulas are highly intolerant to any transgressions and teleological. Their validity can only be judged through their end result and no unexpected end results can be allowed. While the formula has the appearance of stability and authority, as it embodies and reconfirms theoretical law, its shelf-life as technique for engineering truth is in fact fragile and highly dependant on volatile conditions: the experience of its users and a stable deployment environment.

R1. The Recipe According to the Oxford English Dictionary, the word recipe entered the language in the 1400s, written at the top of physicians’ prescriptions. The word derives from the Latin verb recipre, meaning to receive10 – the patient had to ‘receive’ the ingredients listed by the physician for his or her medical remedy; the cook has to ‘receive’ the ingredients and instructions for making a specific dish. The recipe is always based on a past instance of using the right ingredients in the right way to then achieve the right result. According to Gasking, the expression of a recipe is as follows “events of the B sort can be produced by means of producing events of the A sort”11. I believe that his definition demands additional specification. Recipes state that events of the B sort can be produced by means of producing events of the A sort, but only in the specific circumstances and using the specific ingredients. Recipes, just as formulas, rely on invariance of components, as “changing an ingredient might not only destroy the integrity, it might actually render the product inedible.”12In this sense, the recipe is a ceteris paribus instruction, but it is also inherently positivist. It is based on observed facts and seeks to formulate and disseminate the procedure that leads to an additional occurrence of a previously observed fact.13 9 Nancy Cartwright. 1983. How the Laws of Physics Lie. Oxford University Press. p.44 10 “recipe, n.”. OED Online. June 2017. Oxford University Press. http://www.oed.com/view/ Entry/159522?rskey=jLMnmg&result=1 (accessed November 30, 2017). 11 Gasking, Douglas. 1955. “Causation and Recipes.” Mind 479-489, p.481 12 Lisa Heldke. 1988. “Recipes for Theory Making.” Hypathia 15-29, p.24 13 I argue that the recipe is positivist in the sense that was put forward by Comte in The Positive Philosophy. My argument relies heavily on the translation and the preface to The Positive Philosophy by Harriet Martineau.


According to Haak, “We explain why the following of a recipe leads to certain results in terms of the causal connections that are utilised in following it.”14 He goes on to explain that our having and trusting recipes is dependent on a belief in causal connections between different sorts of events. When we set out to follow a recipe, we trust that the series of steps we need to take will cause the result that we expect.15 In that sense we trust not only the steps, but also their sequence. The validity of a recipe depends heavily on its temporal structure: the order and timing of the instructions. One needs to follow the steps stipulated by the recipe in the expected order, but also in the expected timespan. Recipes claim determinism in their nature and in their form. As seen above, given the same starting condition and initial ingredients, the recipe purports to always produce the same output. This is why we trust to follow them. But this also means that the recipe does not allow for variance in its form. The ingredients and environmental conditions must not change, but the instructions can’t change either. This has been acknowledged by Heldke in her formulation for a recipe for theory-making: “the recipe plan is not relativist: once you decide to make a certain food (…) some methods of proceeding will be closed to you “. The recipe then invalidates any other routes of proceeding that are not the ones it puts forward. This is not to say that the predicted outcome of a recipe cannot be achieved in alternative ways – we can’t say that one cherry pie is more cherry pie than another – but rather that each alternative is a new recipe. As such, recipes can’t be updated – recipes can only be overwritten. Whenever a recipe is deployed, it is done so in order to achieve a specific result – a specific dish, the cure of a specific illness or rather a specific chemical substance. Unlike other mechanisms for prediction, recipes always have a title and the title always hints at the outcome of the recipe. The recipe exists for the sole purpose of enforcing the predicted outcome and safeguarding against wrong products. “Recipes are only formulated in accordance with the purposes they may have.”16 So they too are intrinsically teleological and this binds their form to the outcome. A specific recipe is false as soon as it produces a false result. Given that it can’t be updated, the durability of a recipe is limited to the first false result.

S1. The Specification “The specification states what the contractor builds”17. The specification document serves to aid the translation of drawing into building. It safeguards against false interpretations of architects’ instructions by specifying the materials, techniques, components or standards to be used in the construction process.18 Looking back at the etymology of the word, specification is “the action of investing with some specific or determinate quality”19. The components or techniques mentioned in a specification document have the quality of being correct in that 14 R. J. Haak. 1967. “Recipes and Causes.” Mind 98-102. p. 99 15 “recipes can only be followed if one knows that this causes that, and this brings about or results in that.” (Haak, 1967) 16 R. J. Haak. 1967. “Recipes and Causes.” Mind 98-102. p.99 17 Wilson, Sarah, and Bal Manak. 2017. The legalities behind specifications. 14 March. Accessed December 3, 2017. https:// www.thenbs.com/knowledge/the-legalities-behind-specifications. 18 In what follows, I will use the traditional definition of construction specification, as opposed to the more flexible alternative described by the National Building Specification. Traditional or closed specifications are produced at the end of the design process and determine precisely the materials and components to be used in the construction process. Performance or open specifications gives suppliers and contractors more discretion to suggest alternative solutions given that the performance standard should be met. 19 “specification, n.”. OED Online. June 2017. Oxford University Press. http://www.oed.com/view/ Entry/186007?redirectedFrom=specification (accessed December 06, 2017).


particular context; the ingredients become the carrier of truth and the validity of the spec relies on their particular qualities. By pinning the truth of the specification to the intrinsic truth of the component, many avenues for “vicious errors”20 are opened. Most specifications lose their ability to engineer truth because of specified materials or components being substituted.21Once one substitution is made the specification is no longer intact. In order to trust the specification, one has to trust that the specified components themselves hide no errors; by specifying one faulty detail, the validity of the specification document crumbles. Specified materials also require both correct workmanship and a correct specification document. Additionally, the quality of the components must be invariant; when one component is specified, one should trust that it will be stable and uniform between the moment of specification and the moment of implementation. This makes the specification document highly time-sensitive; its longevity is dependent on the longevity of each individual component. The specification document is inherently linguistic; it relies on a code that has to be understood by all parties interacting with it. Specification writers should use standard classification systems in order to “remove any potential for confusion or ambiguity”22, however this opens up the possibility of mis-referencing and mis-translation. The codification of the spec document is supposed to ensure easy communication and consistency, but in fact it adds room for linguistic confusion and diagrammatic oversight. Here, Foucault’s remark applies: when we believe the truth of a specification document, we have to believe that “words had kept their meaning, that desires still pointed in a single direction, and that ideas retained their logic;”23 In this sense, the specification document can only be true if its content is described with correct terminology that is invariant in meaning and that can be translated between languages and disciplines without any inaccuracy. According to National Building Specification, poor specification writing and miscommunication between disciplines rank amongst the most common causes of specification difficulties. Additionally, a standardized description of a component implies stability and the validation of a body with authority; we’re much more likely to overlook an error if it hides behind a standard code. Specification documents are project- and context-specific. The specification document functions in conjunction with a set of drawings and is meant to communicate information that cannot be conveyed through those drawings. The immediate context of the specification document is determined by its relationship with the set of drawings; contradictions between drawn and specified information invalidate the specification document. In practice, most issues with specification documents stem from contradictions between them and drawings. In that sense, specs too are ceteris paribus means of engineered prediction: they hold only in the circumstances that are stipulated by the drawings. In addition, specification documents also depend on their wider context. Specification documents cannot (or should not, according to NBS) be directly transferred from project to project and need to be adapted whenever there is a change in the context of the project (eg. changing a detail or material specification when an unexpected soil condition is found on site). The validity of a specification document depends on the expertise and skills of the agents deploying it. The successful deployment of a specification document requires a context that contains the adequate level of knowledge to understand and carry out the instructions implicit in the component description. 20 The specification seeks to function in the same way as Quatremere de Quincy’s type; it functions as a “inflexible rule that redresses all depraved customs, all vicious errors that are the inevitable result of blind routine” Encyclopédie méthodique. (1788), vol.1 cited in Anthony Vidler, The Writing of the Wall: Architectural Theory in the late Enlightenment, p.151 21 National Building Specification. 2017. Specification report. RIBA Enterprises , p.19 22 https://www.thenbs.com/knowledge/what-is-a-specification-for-construction 23 Foucault, Michel. 1977. “Nietzsche, Genealogy, History.” In Language, Counter-Memory, Practice: Selected Essays and Interviews, by D. F. Bouchard, 139-164. Cornell University Press. p.141


Specifications are set out in order to ensure that the outcome of a construction process is the correct and unequivocal manifestation of the architects’ intention. In that sense, it is highly teleological – it exists and can be judged solely as a means of achieving the true goal of a project. Should the outcome not be the one intended by the architect, the specification loses its grounds and ceases to be a tool for prediction. The traditional specification document is also a highly intolerant system that does not allow for any transgression in material or means. Any failure to employ the correct unit in the correct way translates into a failure of the spec. In this sense it is highly deterministic and has a very brittle lifespan – its longevity is constantly under threat.

C1. The Checklist Generally, checklists are used to safeguard against failure in situations of high complexity that require either the coordination of extensive teams, have serious time constraints or both. Checklists are now routinely used in aviation, surgery, construction site-management as well as often provided in form of troubleshooting guides for household items. Given that they do not claim to achieve a specific outcome, but rather to respond to and avoid errors; checklists are forensic in nature and non-deterministic in form. Their formulation does not stem from the observation of past successful events, but rather of past failures that have been incorporated in the checklist. This allows for constant updates and improvements, and means that subsequently the checklist has a longer lifespan than the previously analysed forms of engineered prediction. The checklist does not claim determinism in form nor in content. The successful outcome of a checklist is defined by negation: any non-error is considered correct. The validity of a checklist is not bound to one specific outcome, so it cannot be invalidated by false outcomes. A plane crash does not mean that the checklist was flawed and needs replacing, it simply means that an additional step needs to be added. In this sense, the checklist bypasses Hume’s problem of induction by never claiming the truth of the current variant. The checklist’s claim to corrective prediction is not based on the validation of the fact that “instances of which we have had no experience resemble those of which we have had experience” 24, but rather that no past errors will be repeated in the same manner. Checklists give up precision in favour of longevity. They are not foolproof and do not contain all needed information to engineer predicted truth. Not everyone with access to a plane and an aviation checklist would be able to get it to lift-off, but my failure to fly a plane would not render the checklist invalid. Checklists do not predict data, but rather accommodate data and eliminate observed errors. After my failed attempt to fly a place, the checklists would have to include a step that ensures the pilot has had the appropriate training. The requisite of precision and prior knowledge is transferred onto the user. Errors don’t curb the lifespan of a checklist, but rather increase it by improving its corrective prediction power: more errors can be mitigated. As such, I believe that the tolerance and non-teleological nature of the checklists renders it a more durable form of safeguarding against falsehoods. So, the inherently unstable checklist is in fact a vehicle for more long-lasting truths than stable formulas, deterministic recipes and prescriptive specifications.

24

David Hume. 1888. Hume’s Treatise of Human Nature. Oxford: Clarendon Press. p. 89


F2. The Failed Formula Formulas should not be misused. Deploying the correct formula implies the necessary expertise for its correct use. Given that if I have the knowledge required to deploy the formula and am aware of the centuries of experience collapsed into it, I should not be able to misuse it and harm its longevity. The period T of a simple pendulum moving at small angles is related to its length l and gravitational acceleration g by the formula: g= 4lπ2 T

2

T = 2π l g

0≃ 40 °

l = 65cm

This simple experiment taking as a basis the formula above can be used to calculate the gravitational constant, g. The following table shows 5 sets of measurements of the period T, while l stays the same. According to an analysis by the British Geological Survey the value for the constant g in London, E2 is 9.8118759 m/s2. l (m)

0.65

10 T (s)

T(s)

g(m/s2)

17,42

1.742

8.456

17,21

1.721

8.663

17,06

1.706

8.816

16,59

1.659

9.323

17,38

1.738

8.495

The error is substantial so the formula must have been misused. In this case the error stems from the misunderstanding of the term small. Formulas are intolerant and inherently ceteris paribus, so their longevity depends on the invariance of their context - here, the formula expired because of its intolerance.


S2. The Failed Specification Specifications safeguard against failure. They clarify the information in the drawings and stipulate the components and construction method of a building. Should one employ the indicated ingredients, the specification would achieve its goal.

H2 Joint as specified

inside wall panel

outside wall

floor panel dry mortar pack

washer nut

bolt steel tie plate 1” bars

lifting rod

open joint fin w/ lime mortar

H2 Joint as built

inside wall panel outside wall

void filled with rubbish ties not attached void and rubbish

screw not tightened

blast angles added after collapse

In the case of Ronan Point misuse of the specification was manifold: the ingredients stipulated were not context-specific and as a ceteris paribus prediction, the specification failed; the use of the indicated ingredients by the indicated contractors did not imply correct workmanship; a specified detail had inherent flaws and this translated to a faulty specification251. 25 Pearson, Cynthia, and Norbert, Delatte. 2005. “Ronan Point Apartment Tower Collapse And Its Effect On Building Codes”. American Society of Civil Engineers. https://doi.org/10.1061/(ASCE)0887-3828(2005)19:2(172).


R2. The Failed Recipe Recipes should be fool-proof. They claim to offer all the ingredients and all the instructions to achieve the desired result, the true result. I should be able to simply receive the knowledge, follow the steps in the correct order and this should be sufficient to safeguard against falsehood and continue the lifespan of the recipe.

The recipe above failed because of a delay between step 5 and step 6.; the three hours that passed between the steps were not forbidden by the recipe but proved to lead to its failure. Recipes rely on time-constraints that are not explicitly formulated. In this sense, the recipe does not have all necessary instructions and implies that the steps are followed at a reasonable pace. Should unexpected delays occur, the recipe fails as it does not achieve its final goal


C2. The Faulty Checklist

The checklist gave up precision in favor of longevity. It is no teleological or deterministic, so a wrong outcome does not necessarily invalidate the format. It allows for failure and accommodates unexpected data. However, the checklist can fail to function in the intended way if, in its genesis, a recursive clause has been slipped.26

Close cabinet doors Make bed Empty dishwasher Sweep kitchen Make checklist Cook lunch Buy groceries Walk dog Take Rubbish out Clean up kitchen Focused 5 minute pickup Check bathrooms for clutter Hang up/put away clothes in bedroom Wipe Down Bathrooms Put out clothes Make lunches

The checklist cannot accommodate recursivity because it is a fallacy that cannot translate into a step to be added in the checklist; the only solution is to take it out. It is only in this case that the longevity of the checklist can be harmed. By deploying an endless loop, the almost endless lifespan of the checklist erodes.

26 White, Dana. 2017. “Printable Household Cleaning Checklists | A Slob Comes Clean�. A Slob Comes Clean. https://www. aslobcomesclean.com/2012/01/printable-household-cleaning-checklists/.


C3. The long-lasting checklist If long-lasting truth is better truth, then the checklist wins. It provides an alternative to the determinism and teleological nature of formulas, recipes and specification. By relinquishing precision, it delegates responsibility to the user and is able to respond to failures, rather than being killed by them. As responsive tools, checklists are unstable in content but stable in their modular form and more pervasive than the previously mentioned forms of engineered prediction. Being constantly updated in a forensic manner, their predictive power becomes distributed and networked. The checklist is then a marker for the shift in the construction of facts from linear causality to a cybernetic feedback model. And just as cybernetic truth, the truth of the checklist rides on the back of the illusion of being participatory, up-to-date and never to forgo.


Bibliography Baigrie, Brian, and J. N Hattiangadi. 1992. “On Consensus and Stability in Science.” The British Journal for the Philosophy of Science, 435-458. Cartwright, Nancy. 1983. How the Laws of Physics Lie. Oxford University Press. Comte, Auguste, and Harriet Martineau. 2000. The Positive Philosophy. Batoche Books. Feyerabend, Paul. 1962. “Explanation, Reduction and Empiricism.” Minnesota Studies in the Philosophy of Science 29-97. Foucault, Michel. 1977. “Nietzsche, Genealogy, History.” In Language, Counter-Memory, Practice: Selected Essays and Interviews, by D. F. Bouchard, 139-164. Cornell University Press. Gasking, Douglas. 1955. “Causation and Recipes.” Mind 479-489. Gawande, Atul. 2009. The Checklist Manifesto: How to Get Things Right. 2009. Henry Holt and Company. Haak, R. J. 1967. “Recipes and Causes.” Mind 98-102. Heldke, Lisa. 1988. “Recipes for Theory Making.” Hypathia 15-29. Hume, David. 1888. Hume’s Treatise of Human Nature. Oxford: Clarendon Press. Hughes, Francesca and Kovacs, Gergely. 2016. “Fear is in the Detail.” Harvard Design Magazine. 86-93 Latour, Bruno, and Woolgar Steve. 1981. Laboratory Life. Princeton University Press. Levy, Matthys, and Salvadori, Mario. 1987. Why Buildings Fall Down. London: Norton and Company McPartland, Richard. 2017. What is a specification for construction? 2 May. Accessed December 3, 2017. https://www.thenbs.com/knowledge/what-is-a-specification-for-construction. National Building Specification. 2017. Specification report. RIBA Enterprises. Pearson, Cynthia, and Norbert, Delatte. 2005. “Ronan Point Apartment Tower Collapse And Its Effect On Building Codes”. American Society of Civil Engineers. https://doi.org/10.1061/ (ASCE)0887-3828(2005)19:2(172). Ramharter, Esther. 2014. “Wittgenstein on Formulae.” Grazer Philosophische Studien 79-91. White, Dana. 2017. “Printable Household Cleaning Checklists | A Slob Comes Clean”. A Slob Comes Clean. https://www.aslobcomesclean.com/2012/01/printable-household-cleaning-checklists/. Wilson, Sarah, and Bal Manak. 2017. The legalities behind specifications. 14 March. Accessed December 3, 2017. https://www.thenbs.com/knowledge/the-legalities-behind-specifications. Wittgenstein, Ludwig. 1976. Lectures on the Foundations of Mathematics, 1939. Cornell University Press. —. 2001. Remarks on the foundations of mathematics. Oxford: Wiley-Blackwell.


Turn static files into dynamic content formats.

Create a flipbook
Issuu converts static files into: digital portfolios, online yearbooks, online catalogs, digital photo albums and more. Sign up and create your flipbook.