I hesitate to disparage the analysis of an anecdotal person I've never met, but I disagree with the claim that statistically teacher quality doesn't affect student outcome.
The current leading meta-statistical analysis of teachers and teaching methods and their effect on student learning is "Classroom Instruction that Works", by Marzano, et al. They have statistics on good teacher/bad school, good teacher/good school, bad teacher/good school and bad teacher/bad school and the effect on student performance. The average effect size when comparing multiple studies shows that good teachers do make a significant and measurable difference in student achievement.
I think leaving teaching to "people who can't do anything else" is a cop-out. I've taught the basics of programming to over 1000 students in my career thus far. I know of at least a dozen who now do it for a living, including some who say that a career in programming never crossed their mind before taking my class. Do you think I've had a net positive effect on the industry? I can write a lot of code, but teaching others to code eventually produces more. "Teach a man to fish" and all that.
I do agree with Prrometheus below that a lot of public education is "an inflexible, pathological bureaucracy."
I also agree that if you don't enjoy working with young people, don't bother. Further, if you can't set healthy boundaries, you'll eventually get burned out. Also, if you're a hacker at heart, you'll need to keep coding somehow, or you'll go mad.
Of course, I don't know anything about Teach for America either, but I do know that I really enjoy teaching, and I'm glad I went that route rather than industry or starting something of my own.
Just my two cents.