The conversation about how to improve American education has often taken on an increasingly confrontational tone. The caricature often presented in the press depicts hard-driving, data-obsessed reformers — who believe the solution is getting rid of low-performing teachers — standing off against unions — who do not trust any teaching metric, and care more about their jobs than the children they are supposed to be educating. One promising area that can potentially serve as a middle ground is a greater focus on the developing classroom-tested, replicable, and scalable approaches to teacher training that make the educators we already have perform better.