yimmer
Our civilizations, for the most part, don't cultivate genius, unfortunately. And geniuses aren't radically less common in some groups, unless you mean groups like "people who suffered severe childhood malnutrition". Normal schools aren't really set up to deal with people with IQs more than a standard deviation, or perhaps two standard deviations, from the norm.
If you define genius to be an IQ of 160+, and model it as a Gaussian with a standard deviation of 15, most schools have no geniuses of any race. If you take a more-reasonable fat-tailed distribution, many schools still don't.
My personal, anecdotal bias: the best school I went to was quite small, and had several geniuses - including a black one. There weren't many black kids, but the ones who were there were exceptional; I wouldn't be surprised if they had the highest average IQ of any ethnic group at the school (and yes, there were plenty of Asian and Jewish students, from several countries).
Needless to say, a black Nobel laureate such as http://en.wikipedia.org/wiki/Sir_William_Arthur_Lewis is vastly more intelligent than the median member of any ethnic group...
Underprivileged students from minority backgrounds can do quite well at calculus - some teachers, like http://en.wikipedia.org/wiki/Jaime_Escalante , dedicated years to trying, with quite impressive results. This has been demonstrated again and again at schools with poor track records. Simply having a competent teacher and an environment that isn't entirely hostile to learning can do far more than many people tend to realize.
Teachers dedicate themselves to teaching for numerous reasons. They don't take on student loans for no reason whatsoever; the credentials they get are part of the price of going into a system with intentional barriers to entry.
Don't confuse a system of "reject some of the good, weed out most of the really bad, and usually get ok results" as something which is an entirely accurate indicator of competence. The shortcomings of the current system don't make it a lie, obviously - but putting decades into a process doesn't necessarily make it better, either.
Teachers are generally intelligent and well-meaning. Their paper qualifications and their ability to actually teach are not as strongly correlated as one would hope, though. While on average I'd expect a qualified teacher to be better at teaching than a member of the general public, I know many members of the general public who are not eligible for jobs as K-12 teachers, but nonetheless are excellent at teaching both children and adults. Amusingly, one qualified teacher I know was given a crash primer in the subject he now teaches professionally by a friend of mine... after he was officially qualified to teach it. On the other hand, I know a teacher in Switzerland who spent several years telling his students random anecdotes about his life, rather than actually teaching the students anything whatsoever about the subject he's paid to teach - and this is a country with much stricter requirements for teachers than the US.
Public schools have requirements for their teachers for a number of reasons: covering themselves, filtering out some unqualified people quickly, etc. The qualifications are neither necessary nor sufficient for finding good teachers, sadly. In programming, qualifications tend to be poorly, and sometimes even negatively, correlated with ability and productivity; I'm not convinced the situation is massively different in K-12 education.
There are no magic bullets for education. Not everyone teaching is even vaguely capable of doing so; there are some horrifically incapable people teaching in the public school system, and also some who homeschool. The majority of teachers in either system do a more-or-less tolerable job, and some excel.