> Oh come on. I scored higher than 135, and I am nowhere
> near a genius.
Depends on the test, including when the test was normalized. Some tests are crap. Some are better. And because of Flynn's data, we now know that it is easier to score higher on a test that has 20-year-old norms--regardless of the test. This means a high score is not an actual high score, because it's derived from the average of a stupider lot of people who took the test 20 years earlier. (Geniusness, like mental retardation, is relative.)
At a .33 IQ rate gain (the rate gain in the U.S. found by Flynn for the Wecshsler intelligence scales), a test normalized in 1980 that you took in 2007 would inflate your score by about 9 points. That is to say, if that test were re-normalized today, the average score would not be 100--like it was set to in 1980--but 109. So your 135 is more like a 126 (compared to your peers), which is smart to be sure, but not genius. Additionally that rate gain could be quite a bit higher, depending on the specific IQ test (for example, on culturally-reduced or nonverbal, "problem solving" type tests). That would mean you're even less smart than you think you are.
IQ scores only mean anything relatively. Absolutely, they are just an arbitrary number and describe nothing. It's how far you deviate from the mean that tells you anything, and, for that, you have to know what the mean is (which is why the Flynn effect is relevant). Most tests have a standard deviation of 15, which means 95% of the population is within two standard deviations (i.e., 70-130). If you score a true 135 (adjusted for the Flynn effect), then you are well in the top 2.5%, which is pretty damn smart.