I've read that the curves on the newer preptests are less forgiving than the older ones. Is there any truth to this, and if so, is it b/c the test has gotten harder or b/c people have gotten better at taking it?
It is true that the raw points (# you answer correctly) to scaled score conversion charts have become much less forgiving in the last ~4-5 years than they were in the previous years.
Meaning that depending on which preptest one uses as a full timed practice test, 90 correct answer choices yielded a higher scaled score on tests from the '90s and pre 2005 era than on tests from 2005 to the present.
That is a big reason to save and use the majority of the most recent tests for full test timed practice in order to be able to get an accurate gauge of how you are performing and would likely score on a modern administration. Otherwise, if you base your likely potential score by taking an older test timed you will likely achieve a scaled score that is inflated compared to how you would ultimately score on an upcoming administration with your current skill/performance level.
Why is this the case? That is a matter of debate and is controlled by very complicated psychometrics (complex math and test development stuff).
I think the modern tests are much more standardized/predictable/stable than the tests from the past. The logic tested hasn't changed, but the ways in which it is tested have become much more consistent in many ways. One of the notable differences is that back in the '90s there were numerous test forms that contained really unusual oddball/non-conventional/one of a kind logic games. Odd-ball/non traditional logic game types pop-up much less frequently now than before.