>>1Given that you can see different levels of skill among programmers, you can then arbitrarily take the top 2% of them (that number is also arbitrary, but the precedent is that it's what mensa uses for what that's worth) and say "these are geniuses (genii?)." Of course, you can also argue about semantics or specifics forever if you prefer, but the same way you can designate a rare positive event as a miracle, you can designate an unusually good programmer as a genius.
Of course if you want to evaluate a specific individual as genius or non-genius you have a harder problem. I would say the ability to solve problems others cannot is a good indicator; speed might be as well, although that's less clear. If you solve some problem previously thought impossible, and it takes you a really long time to do it (or to do anything), most people will probably call you a genius anyway. But not a rock star.