That would be 30 per category, so 60 points. Which can definitely be the difference between a top university considerign an applicant or throwing the app away.

Back in my day when SAT's were 2400, I heard the swing being a difference of 100 points. Since reverting back to 1600 a 60 point swings sounds proportionately similar to 13 years ago.

Anyways, I think when people mean "gaming the system", they mean test prep classes literally geared towards the SAT and little else. Not practicing calculus to overprepare for the SAT math portion (maybe even using it to validate answers on a lower level), but being told that X is on the test and using older copies of the SAT to practive with. That does start to feel a bit game-y in my eyes. You're not better understanding combinatorial or algebra, you're understanding the pattern that a test has and playing based on that knowledge.

Right. Growing up in Melbourne (Australia) there was a marked difference in the performance of state school and private school students at university. Theory (don't know about studies) was that the private schools were selective about their students, but also held a lot of their reputation (and their ability to justify tuition) on TER (Tertiary Entrance Ranking) scores, and as such, taught their students extremely well "how to get a very high TER", but not necessarily how to be an independent learner - information spoon-fed in classes as such meant that assignment and essay scores were quite a bit higher than those in state schools, where there wasn't the individual and personalized attention available.

Once those students got to university, while the state school students were used to conducting independent research, having no-one who particularly cared or was going to crack the whip on assignment turn-in ("You're adults - turn your work in or don't. Not my problem."), and such thrived, while many private students found themselves at a loss. I remember many questions about elementary "Well, how would I go about researching that/studying this/doing that" in my first year of university that were along the lines of this theory.

Don't you get the majority of the benefit of specific test prep by running through a couple of sample tests from a guidebook, which you can get from a public library? (I'd say most of the benefit, in my case, but probably shouldn't generalize that far.)

I can imagine a longer test prep class that was really more like general math tutoring could make a bigger difference, a la Bloom's 2-sigma problem.