Conor Williams with some good points about Brooks' column – we're not talking about tests being able to distinguish between 55^th vs. 45^th percentile teachers; even if tests (maybe 50% weighting, plus many other measures) can only identify the stars (top 5%) and the horrific (bottom 5%) teachers, that would be a MASSIVE improvement from what we have today:

These tests aren't perfect, but they're simply not uncertain enough to produce those kinds of results. Not even close.

This is something often lost in education reform debates. When we're talking about using tests to measure accountability, we're very rarely talking about shades of gray. Reform opponents often argue that tests are inaccurate, and that they don't represent the work of students or teachers. To a degree, this is reasonable enough: we don't want teachers to be victimized by a bad day of testing, or a hard cap on "good enough" test results.

That's fine. The thing is, reformers aren't usually worried about test results that are just barely good enough vs almost good enough. They're not talking about results within the error bars of the test. They're talking about huge discrepancies, like the results cited above. They're asking: Why should we continue to pay for schools (public or charter) that are underperforming this dramatically—especially when there are other options?

---------------------