Some people are great at teaching, others are not.
I believe this is a self-perpetuating myth. What the data shows is that new teachers in America improve rapidly over the course of about three years, after which they are about as good as they'll ever be. So it's certainly not the case that some people are just naturally teachers; great teachers have to learn the craft through practice, and that learning comes after they finish their official training.
But maybe what we're seeing is that it takes teachers three years to reach their inborn teaching potential, after which they no longer are able to learn anything more that might help them. My question is, how do we know that? How do we know that American teachers are actually completely incapable of becoming better teachers after three years of in-classroom experience?
We don't know. The remarkable thing is that until very, very recently, very few American school systems have actually attempted to systematically improve the performance of their teachers through observation of what they're doing in the classroom. They may have "professional development" where they get more of the same kind of theoretical training they got in education school, but they usually don't follow up to see how the teacher actually puts that to use, or even to identify bad habits the teacher may have developed over the years, or good habits he hasn't. In my kids' school system kids are sent home early on "professional development days" so that working with actual students won't get in the way of a teacher's skill development. It's like trying to make someone a better baseball hitter by banning bats and balls from training and simply talking to players about the theory of biomechanics.
Imagine you own a factory and your assembly line is turning out too many defective widgets. How would you address that problem? Would you send your engineers to a seminar every year on manufacturing theory and ask them to make design changes when they came back from that seminar? Or would you go over the assembly line with a fine tooth comb? While the seminar idea has it's merits, it's too slow and it'd take sheer luck for the seminar to hit on the particular problem that's affecting the line.
In America we have a simple model for improving the teaching at a bad school: fire the bad teachers and hire better ones. But imagine, just for a moment, it is possible to use empirical methodologies to improve the performance of any teacher. Imagine for a moment some bad teachers could be transformed into mediocre ones; some mediocre teachers into good ones; and some good teachers into great ones. In a world where that was possible there'd still be a place for the hire and fire strategy, but relying on that strategy exclusively leads to two unfortunate and unnecessary results: (1) Poor districts have to make do mostly with inadequate teaching and (2) teaching in rich districts tends to be adequate, but great teaching remains uncommon.