Thank you, I know basic statistics. If you look at the number of people who take the SAT and who are competitive applicants for top schools, then 0.6% makes a difference.
Applying to Princeton, you're really not going to see anyone (excluding the 25% special cases, and perhaps not even then) in the 2100s getting in. Remember that medians are reported with special cases in mind.
http://www.princeton.edu/admission/applyingforadmission/admission_statistics/
Percentage of applicants by SAT range.
2100-2290 - 7.5%
overall acceptance rate - 7.4%
(of course this accounts for the people in the 2200s as well as the 2100s, but - 1900 - 2090 - 4.9%. we can infer that people with SAT scores in the 2100s are probably looking at around a 6.5-7.0% acceptance rate). The point is, getting a score of "only" 2100 isn't nearly as damning as you're making it out to be, and actually seems to be only a little bit below the average of people getting in.
However, I would like to point out another stat on this page that on the surface seems to give merit to your argument.
2300-2400 - 16.5%
Yes. This small group of people does seem to get significantly better odds of getting in. However, it's important not to look at this and immediately assume that this comes as a result of the SAT. As you said, people in this score bracket tend to be extremely high-achieving. They have multiple e-cs, volunteer, classes, gpa, etcetera. They may be also more likely to spend large amounts of time and money taking the SAT several times, taking prep courses, achieving perfection through trial and error and a playing into the system that most colleges go to (looking only at highest score, superscore, etc). People like to have SAT scores similar to those they consider their intellectual peers, and so they'll push themselves through whatever available means to get an inflated score. This doesn't completely account for it, of course, and it does seem to point that a higher SAT at these levels does give you a better chance. (the scale here is significantly bigger than 100 points)
tl;dr compromise, but I do have a couple more things to say that are relevant
If you look at stats like these and see that 2300-2400 people (The top .4%!) still only have a 16% chance of being accepted, that certainly raises some questions about the validity of the admissions process - if these nearly "perfect" children are being rejected on a massive scale. Part of this can be attributed to luck/chance/whatever, but there's another factor that people often overlook that ties in with the SAT-craze I mentioned earlier.
Everything has to fit.
What I mean by this is that a 2350 SAT with a shitty GPA and little to no E-Cs isn't going to impress any top-tier college. Because the SAT by its very nature is so variable and easily influenced through disadvantaging methods like prep courses, review books, prep classes in school, it's inherently discriminatory and not a good predictor of anything by itself. As such, admissions officers don't give it nearly as much weight as you would think upon first glance.
The comparison between a student with a 2200 SAT and a 2300+ SAT was valid, but a more likely scenario is one in which the 2200 student has a higher GPA, shows more ambition, and seems more talented than the 2300 SAT student. Because these two numbers are mathematically very close, the fact that the 2300 is higher than the 2200 is nigh irrelevant in the face of the superior other factors.
My point was never "you have the same chance if you're a randomly selected person with 2100 as 2200". My point was more that for a high-achieving student who excels in other regions than the SAT, the stress that they often feel to score within the "golden" zone may be unwarranted, and is at the least overblown. As seen by the very significant amount of people who get into prestigious universities like Princeton with "low" SAT scores, the SAT is more of a factor that tends to click with everything else than it is a deciding factor between two otherwise similar candidates.