I was mainly concerned with teaching Clojure in this project as everyone had little or no experience prior, so nobody was blocked by learners. Those who were fired all had one thing in common: They refused to adopt Clojure beyond its syntax in favor of patterns they learned to rely on in another language. This varied from porting Scala polymorphism to just banging out imperative while loops with atoms. I don't think they were incapable. They simply didn't want to learn it. Code reviews turned into arguments quickly, took endless time and lead nowhere.

Their code was never touched by those who stayed and were able to work together. Later it was rewritten, because in spite of tons of unit tests (which passed), it was full of bugs. The code was considered "write-only". The rewrites averaged at 5-10% LOC.

That’s still an argument in favor of Clojure. People spend loads of time and money trying to figure out how to do such filters. If it comes for free by choosing Clojure, that’s pretty great.

I doubt this filtering is in any way done by Clojure though. It's difficult to recognize top performers in interviews, but it's much easier to see that in day to day work. What is usually missing is the willingness to fire people for not being top performers.