Playing devil's advocate, it's possible that someone winds up getting a PhD or publishing crypto research after finding out that they really like doing math, after initially choosing it with the intent to teach.

I have absolutely no idea whether that matches reality for enough people to make any kind of a difference.