The fact that there's a natural correlation doesn't mean the SAT's correlation is identical to the natural one. And if it's worse than it should certainly try and change that.
Of course - I never said otherwise. But the changes they're making don't seem to be simply about making the test more accessible and less elitist; they seem to invite ceiling effects, and I don't see how that helps anyone.
Rather than making test prep widely available, the College Board should strive to design a test for which the only effective prep is getting a solid well-rounded education, and they should pay no attention to the complaints that some of the questions are too hard for almost anyone to answer.
I agree that, if the SAT were ideal, there'd be no way to prepare for it in particular.
However, a nationally standardized test must have consistent scores to be useful, and currently the best way to do that is to keep the general format and content consistent. That means that a student can discover and prepare for the test's general format and content in advance, which is definitely advantageous.
A truly ideal SAT would have a different format and test wildly different skills every time it's issued in order to defeat the idea of test-specific preparation, but still produce meaningful scores. I'm skeptical that such a test can produce sufficiently consistent scores, though I'm very willing to someday be proven wrong by some clever test developer. That'd be pretty neat.
So, until we create an SAT for which test-specific preparation is meaningless, it's important that test prep resources are distributed evenly so that students with good educations won't have artificially lower scores due to unfamiliarity with the SAT's standard format and content. It's an unfortunate but necessary compromise :/
