[bromang]

This article has a pretty bad smell. The author makes a lot of fairly extreme claims but these are mostly based on his own anecdotes or non public data sets. The claims that

"you can separate all advanced math teachers easily into two categories: Okay with blacks in their classroom [...] Not okay with blacks in their classroom. Whites end up succeeding, blacks end up failing"

and

"the evidence shows that teacher recommendations have zero correlation with aptitude in a field"

are really quite dubious and almost certainly false. How did they measure whether they were "okay with blacks in their classroom"? As many others have said in the original thread, this is probably a hoax.

[Alex3917]

Actually those claims are basically true. If you look at the graph from my blog post here:

http://alexkrupp.typepad.com/sensemaking/2009/02/the-most-im...

What you see is that the data points from all three ability groups are basically overlapping, meaning that the students get sorted into ability groups based on factors like race rather than on actual ability.

That being said, while you might be able to model the bias of teachers and school systems by using an algorithm, I've never heard of schools actually using an algorithm to explicitly sort students based on race, which he implies that they are doing.

However, assuming he doesn't actually mean this, I think the post is actually legit but just poorly written, perhaps purposely to promote his startup.