A while back I looked at gaokao problems for fun. TL;DR something like gaokao seems like a nice thing to have, even if the china system puts too much weight on it.
But before I start, some thoughts I perhaps… forgot to add on the previous college apps blog.
Perhaps the most important takeaway is that applications measure what you say you did, and what you say you are, instead of what you really are. Like, it’s really hard to get an objective handle on what you did as X leadership position, or how good you are at Y, or how much of Z was your work and how much of it was your name being written on it.
Even objective skill, say on math, is hard to measure. And perhaps, it doesn’t need to be measured, because colleges don’t care that much, but surely it should be measured, to give an objective handle? Even if they end up being treated roughly the same way that SATs are treated?
Olympiads are kind of this way, but they see a significantly harsher difficulty curve, aren’t as widely adopted, and etc. After all, a first-stage selection test for the IMO team isn’t necessarily going to test an average student’s skill, and they are very high-variance. And most people will never take these tests. Olympiad tests are nice, but they are not designed to serve as a general test of knowledge and thus shouldn’t be used as such.
I want to preface this with the acknowledgement that gaokao is not a perfect system either. As a test, it definitely has problems, such as having considerable variance, plus time pressure, among other problems. As a system, it places a bit too much weight on pure academics, plus introduces a lot of stress and causes cheating problems, etc, etc, these are pretty well-known. It’s definitely not the best system. But there are definitely merits to this system.
The extent to which i’ve looked at gaokao is like around one test of math, physics, and skimmed the other sciences. I’ve solved through two physics tests for fun. From what I’ve seen, the tests range from reasonable difficulty to pretty hard. More specifically, it probably is like between a hard honors course test question and an intermediate olympiad exam question. This range is pretty nice imo, does a pretty good job of distinguishing the relevant range of students without losing resolution at the higher end (although I think it loses resolution at the lower end, idk tho).
The tests have a lot more breadth, too, compared to like the SAT. First, there are quite literally more subjects. Also, physics covers E&M, and some waves / thermo (qualitatively), chem covers organic, and bio is just more in depth. This is probably cuz china education on all of this starts in middle school, and is taught at least nontrivially well. as a result they have time to cover all of this. Meanwhile in the US this is like 2 years, and slowly.
These tests actually test how much you know, with real high-end resolution. It’s like measuring water boiling with a 120 C thermometer rather than a 40 C thermometer. From what I’ve seen, I’d probably do pretty well on physics, then like ok at chem and bio, but you’d see a difference. Unlike stuff like AP tests which honestly can’t distinguish anything, especially with 1-5 scoring.
One other thing I’ve noticed while looking at the tests, is that it’d be quite hard for the average US student. I don’t think most of the honors students at my school (no offense) could do well on it, because (a) a superficial understanding of content, (b) the courses just don’t go deep enough, (c) breadth, etc. Of course, they wouldn’t do too badly, either (that is, on the stuff that they have learned); they’d probably perform right at their skill level. This feels, perhaps, a bit concerning.
Of course, Goodhart’s law “When a measure becomes a target, it ceases to be a good measure” applies here. But then it applies everywhere, too. And overall, I think the gaokao does a pretty good job of what it’s intended to do. And it appears to be harder to optimize for gaokao testing, other than just getting better. It’s pretty easy to optimize for SAT without actually getting better.
Not that SAT doesn’t do what it’s intended to. It’s not intended to have higher resolution at the top. Or to test non-math/english subjects. The intended range of students is simply lower, and thus it has little to no resolution at the higher end. The problem with this, is that this is unsuitable for admissions to like top colleges, where you inherently need more resolution at the top. Well, perhaps the point to having little resolution at the top is to force students do to things other than studying, that we see in modern US college apps? But this makes little sense when the admissions can just choose to consider these other things too.
I’d definitely like to see something like gaokao implemented, just because it’s better. Of course, there would have to be some changes, and I wouldn’t entirely recommend 100% gaokao-based admissions, but it’s definitely a good thing to have an accurate measure of ability.
