I'm not sure there is necessarily a productivity hit. The tradeoff is between a smaller number of better programmers on a new technology stack and a larger number of worse programmers on some boring old technology stack.
Also, don't forget that these people have to invent the product as well, not just code it. Their productivity as inventors may be a lot higher on their preferred platform.
Obviously, it's not the same for every kind of product, but Fog Creek doesn't make medical devices.
It is not clear if there are better programmers on a very new cutting edge technology stack.
I heard there are more works to clean up Rails codebases nowadays.
As of recently, I'm primarily using Java and I can say with confident that while it is old and boring, there are plenty knowledge sharing out there that elevates the brain of many developers (even by a small amount). Not to mention that the libraries have evolved in a much greater pace to support recent best practices (MVC, DB migration, unit-test, DI, mock, BDD, TDD, you name it, the Java community have them).
Maybe so, but it's not like Joel Spolsky chose some arbitrary new technology and is now desperately looking for good people to use it. It seems to be literally the other way around.
Also, don't forget that these people have to invent the product as well, not just code it. Their productivity as inventors may be a lot higher on their preferred platform.
Obviously, it's not the same for every kind of product, but Fog Creek doesn't make medical devices.