cahn: (Default)
[personal profile] cahn
The Mathematical Association of America (MAA) has a tiered level of competitions that, in the US, is the gateway to participating in the International Mathematical Olympiad (IMO). The first level is the American Mathematics Competition (AMC) 10/12 exams -- roughly speaking, you take 10 if you're in grade 10 or below, and 12 if you're in grade 11 or 12, though younger students can take the 12. This competition is multiple-choice and open to anyone who wants to take it; usually there are, idk, a couple dozen or more kids from E's school who take it, and I think most high schools around have it as a possibility. The second level, which you are invited to if you score above a certain threshold on the AMC, is the American Invitational Mathematics Examination (AIME). Usually at E's school there are a couple to a few kids who qualify for that. (These two contests are open to international students.) The third level, dictated by a threshold that is a function of your AMC and AIME score, is the USA Junior Mathematics Olympiad (henceforth JMO, for the route via AMC 10) or the USA Mathematics Olympiad (AMO, via AMC 12), which unlike the first two levels is a proof-based competition. (There are a couple more levels after this that lead to the six kids who are the US team for the IMO, but I have no experience with them and they are not relevant to this rant, so I won't talk about them here except to note that they exist.)

I have spent way too much time this winter being angry at the MAA (*), and it hasn't even directly affected my kid. It may have affected a couple of her friends. (I can't even tell you how incandescent I would be if it had directly affected my kid, who really loves math competitions and has put a lot of energy into them, and we talk all the time about how it's really OK if she doesn't do well, but it's one thing not to do well after having made an honest effort at an honorable goal, but not to do well because the system has screwed you over is another thing again!)

The issue here is that the MAA competitions have become these things that kids perceive as very important for college, etc. And what that means is that there is a very large incentive to cheat. And in the last few years there have been quite a few more widespread ways to easily cheat. (Ironically, because of all the rampant cheating, the MAA competitions are now somewhat less taken into consideration by colleges than they used to be.)

Since 2023 the level and widespread nature of the cheating has increased by a lot, with major leaks, some of which appeared on public websites. MAA's response has been veeeeeery slow. This year there have been some common-sense changes to the AMC/AIME, like, for paper copies, shortening the time between when the exam was available and when answer sheets had to be uploaded, although it's still more than 24 hours. (We do paper copies in our area.) Why has this taken so long to implement. Who knows. They have also moved the JMO/AMO administration to be in "central" testing centers to cut down on cheating. I think this is in general a good idea, though it's rather lucky that a) we live only a couple of hours from a center; b) we have the resources to easily be able to stay in a hotel that weekend so kiddo doesn't have to get up super early both days. It's not very equitable, but then everything about academic competitions has been getting less equitable since when I was a kid, so what else is new?

However, also this year, the additional wrinkle was introduced that AI became good enough that it can solve basically all of the AMC and AIME problems. Aaaaaaand MAA has an option to do the test online, but does not appear to have any measures in place like a locked browser to prevent a student from using AI while taking the exam (????).

Take a look at this score distribution for last year's 2024 AMC 12A exam. (All data is from https://maa.edvistas.com/eduview/report.aspx?self=&view=1561&mode=6, until they take it down, which at this point I don't put past them.)

The top score is 150, and as you might expect, the distribution is centered on the lower end and gets more and more sparse as it goes to the higher end, as this is a test that is supposed to have some questions doable for most strong math students and test up to the upper limits of the strongest students (with also time pressure as one of the constraints).

Now look at this year's distribution for the same test: 2025 AMC 12A.

Notice how much less sparse the distribution is on the right side in general, and in particular that huge hump on the very right side, at a perfect score. (In fact, the ENTIRE top 1 percent -- which is a particular category of distinction (DHR, for those of you who know what that is) -- was perfect scores, which has never been the case before.) You might say, "maybe the test was easier this year." I have reason to believe it was, if anything, harder (though to be fair I have not done the problems myself). You might say, "maybe all the top kids discovered an awesome way to study and so scores got higher on the top end." I... really... kind of think that if this awesome way had been discovered, E would know about it. (In the interest of full disclosure I should say that everyone was given full credit for one of the problems which was not written clearly. BUT this should STILL not have moved the distribution like that.)

Do you know what the actual difference is between this year and last year? AI got good enough to be able to do all the problems, is the difference.

The test E took was the AMC 10A/B, not the 12A/B, and the distribution doesn't look nearly as obviously wanky for that one, but the thresholds are still higher than I would expect them to be. One of her good friends didn't qualify for AIME who I think totally should have (and who qualified last year). Another kid she knows would, I think, have qualified for the JMO this year if it had been the years before cheating got super bad, but the last several years the thresholds have been much higher than before.

Okay, so, remember how up top I mentioned that the way to qualify for JMO/AMO is by your "index" (a function of your AMC and AIME score that used to be AMC + 10*AIME but this year was announced to be AMC + 20*AIME) being above a certain "cutoff" threshold. (This threshold is often a little lower for girls so they can try to hit a ~10% JMO/AMO percentage of girls.) In all previous years they have released the index threshold cutoff and informed qualifiers at the same time.

This year they sent email to all qualifiers... and have not released the cutoff. AoPS Wiki last week showed that the cutoff will not be released (that verbiage was deleted when I checked it today, but there is still no cutoff listed), and kids were reporting that at least one MAA-associated adult said on Discord that they wouldn't be released at ALL. On both AoPS and the discord channels E is in, there are kids reporting an index who made it that's lower than kids who didn't, and even if you allow for the inevitable trolls, liars, etc. on AoPS, E knows the kids on Discord, and there does seem to be something going on here.

So basically MAA is doing SOMETHING, no one knows exactly what, and completely breaking any kind of transparency, AND basically changing how all this is done at the very last minute (there was NO communication beforehand that cutoffs would not be released this year, or that they would release JMO/AMO qualifiers before cutoffs, or anything). No communication that I am aware of of any kind from MAA about what's going on. (Maybe there will be something. Someday. Maybe that's why the AoPS verbiage changed, because they plan to release thresholds someday? Orrrr maybe they just asked them to take it down for even less transparency?? Who knows!) And as anyone could have told you, the absence of transparency has led to wild rumors about how they're doing the computation. The wildest ones have been getting refuted (it doesn't LOOK like there's any credence to the wild rumors it might be based on race or geographic location), but stiiiiiill.

E told me that someone on Discord said, tongue-in-cheek, that the index is now AMC + 20*AIME + 2d20 [two rolls of 20-sided dice, for those who aren't D&D enthusiasts] which was a joke but...

And if all this weren't enough, they seem to have made an error in the emails that went out telling people they had qualified, which meant a whole cohort of people didn't get the emails that day who should have. (For those of you who care, it seems to have been anyone who took the 10A and the 12B.) This does seem to have been fixed within a couple of days, but at the time it just added to the whole general aura of rage and chaos. Especially since no thresholds were released so no one knew whether it made sense for them to have qualified or not!!

E's and my best guess is that they are doing some sort of complicated cheating-detection and rejecting kids based on that, even if they are above the secret cutoff. (And I do think there is a secret cutoff number.) The problem is that any cheating detection you run on scores alone (and, idk, maybe it's possible it's not just the scores, there are rumors that although they don't have browser lockdown that they can detect tab changes, but again WHO KNOWS) cannot tell the difference between someone who cheated and someone who is really smart and decided to really study hard this year and also maybe had a cognitive jump. E had one of these cognitive jumps in sixth grade where her score jumped by a LOT (surprising all of us) on the AMC 8, and it kills me to think that someone might look at that and decide she was cheating. (There is also a chance that her score might jump a lot on the JMO this year. It might not! But it might -- she's been putting some work into it this year and I can tell she's more confident about it than she was last year (plus which last year there was that whole thing where she ought to have scored higher anyway, but never mind that now). I've been saying things to her like, "you should try not to go to the bathroom too much during the JMO in case someone thinks you're cheating," and I feel awful that I have to say things like that. Though of course the fact that cheaters exist is not the MAA's fault. But the fact that she might have no recourse if she does well and they assume she's cheating because she worked hard?? That WOULD be their fault!)

It is fortunate that A. doesn't seem to be super into math competitions like E has been, because the whole system is broken, as far as I can tell.

(*) ETA: I mean -- on rereading my post it's pretty clear that a great deal of it is misplaced anger at the whole cheating culture, which is obviously not at ALL the MAA's fault, and unfortunately for them they are a convenient punching bag (sorry MAA). I still don't agree with a lot they've done, but it is just a general big mess that probably has no really good solutions.

Date: 2026-03-03 09:56 pm (UTC)
crystalpyramid: (Default)
From: [personal profile] crystalpyramid
Oh god, I'm worried AAPT (the American Association of Physics Teachers) is not going to be able to handle this either. The AMC/AIME were decades ahead of the Physics Olympiad in terms of having any degree of test security when the last big cheating explosion happened on the Physics Olympiad, and AAPT's initial response was to stop permitting people to take the test internationally (somewhat reasonable) and stop letting international students and non-US citizens take the test (OMFG), and of course this was timed for right as US schools were trying to send their international students back to Wuhan because there was a pandemic starting and I was so angry. They've gotten better about not giving people days and days to take the test, and having people upload their results promptly, but their current lockdown browser strategy seems to be "make sure the proctors can see all the students' screens", and while that worked perfectly fine with my one student who wanted to take it on our conference day, it is not going to scale.

Thank you for writing about this. It is extremely relevant to my interests. And if the larger math organization is struggling with it, physics is going to be in trouble.

Also it is insanely cool to me that you have a child who can count on qualifying for the JMO. I was pretty proud of myself the year I got two points on the AIME!

Date: 2026-03-04 11:12 am (UTC)
crystalpyramid: (Default)
From: [personal profile] crystalpyramid
Oh interesting about the kids thinking Physics Olympiad locks down browsers. I know that for the digital PSAT, which really does lock down browsers, students had to download this whole separate app and make sure it worked before the PSAT and it was a nightmare at our "Bring Your Own Device" school. (And then their servers couldn't handle the bandwidth of everyone trying to log into said app at the same time and they had to reschedule the PSAT to another date.)

This year Physics Olympiad seems to be using the EdVistas platform, which looks very official, but it's just something you open in a browser so it doesn't seem like it could possibly be locking anything down.

Ha, it sounds like we had similar experiences of the AHSME/AIME in high school! I did have some students qualify for the Physics Olympiad semifinal at my previous job, and that was pretty fun.

Date: 2026-03-03 11:14 pm (UTC)
seekingferret: Two warning signs one above the other. 1) Falling Rocks. 2) Falling Rocs. (Default)
From: [personal profile] seekingferret
Aw, this all sounds like such a shame for something that should just be fun.

Date: 2026-03-04 01:23 am (UTC)
alias_sqbr: the symbol pi on a pretty background (Default)
From: [personal profile] alias_sqbr
Oh man I'm no statistician but even I know a normal-ish distribution probably shouldn't have a spike at the end... I hope this all gets fixed, its gotta be very stressful for the kids!

Date: 2026-03-04 01:29 am (UTC)
hamsterwoman: (find x)
From: [personal profile] hamsterwoman
Oh wow, those two distributions are so stark! Like, the main distributions are actually pretty damn consistent with each other, by eyeballing, but that spike of perfect scores in 2025...

Shocking that they did not try to do something about AI cheating considering AI cheating has been in the news a whole lot... (Or did they not realize in time that AI would be good enough to solve these kinds of math problems?)

So basically MAA is doing SOMETHING, no one knows exactly what, and completely breaking any kind of transparency, AND basically changing how all this is done at the very last minute (there was NO communication beforehand that cutoffs would not be released this year,

This is giving me Chengdu Worldcon Hugo nominations shenanigans flashbacks!

E's and my best guess is that they are doing some sort of complicated cheating-detection and rejecting kids based on that, even if they are above the secret cutoff.

That would be my guess as well... (And I do think it's very possible that they're looking at stuff like tab changes or something else other than scores, if that's what they're doing.)

I've been saying things to her like, "you should try not to go to the bathroom too much during the JMO in case someone thinks you're cheating," and I feel awful that I have to say things like that.

:((

Date: 2026-03-04 06:06 am (UTC)
hidden_variable: Penrose tiling (Default)
From: [personal profile] hidden_variable
Wow. That distribution definitely makes things very clear. (Like, if this year’s test had been easier, that should also have shifted the peak of the main distribution to a higher score, which clearly did not happen.) This doesn’t even directly affect me, but I’m so vicariously annoyed on behalf of everyone it does affect!

Side not: I don’t know if calculators are allowed for these exams (and I’m assuming this is the type of math for which a calculator isn’t very relevant). But apparently there now exists a modified graphing calculator that runs ChatGPT. I had to turn someone in last semester for using one of these.

Date: 2026-03-06 05:54 pm (UTC)
hidden_variable: Penrose tiling (Default)
From: [personal profile] hidden_variable
I must confess part of my brain is thinking that the ingenuity of that is pretty nifty.

Yeah… that’s the paradox of LLM’s in general, I guess: a few people use a lot of ingenuity in order to ensure many other people never have to deploy the slightest ingenuity ever again. The student with the ChatGPT calculator was someone who seemed to outsourcing absolutely everything in their education to ChatGPT. Like, earlier in the semester I commented that something on their homework looked like chatbot output, and they responded with an email that was itself clearly a ChatGPT product.

Date: 2026-03-04 11:40 pm (UTC)
crystalpyramid: (Default)
From: [personal profile] crystalpyramid
Wow, ChatGPT on a graphing calculator. :(

(I guess I'm glad I bought TI-30xiis for my room but I guess I'll have to watch the kids with their own nicer calculators now. )

Date: 2026-03-06 05:42 pm (UTC)
hidden_variable: Penrose tiling (Default)
From: [personal profile] hidden_variable
Yeah, very disheartening. :(

Providing your own calculators is a good idea! I’m trying no-calculator-allowed exams this semester; we’ll see how that goes.

Date: 2026-03-04 02:06 pm (UTC)
landofnowhere: (Default)
From: [personal profile] landofnowhere
What you've said is basically what I'd gathered from my own sources. These days I no longer really have any emotional investment in the competitions (though I'm still amused by your describing it as a "fandom"), but it sounds miserable from a parent perspective.

It's not very equitable, but then everything about academic competitions has been getting less equitable since when I was a kid, so what else is new?

Mood. Though in my day it wasn't amazingly equitable, it just was inequitable among different axes that were probably less correlated with privilege. And some of these inequities have come from Pareto improvements (programs are around that didn't used to be), but the zero-sum aspects of competition means that it's necessarily worse for some people in some ways.

Running the MAA competitions is a tough job which I would hate to have. While it's never been done perfectly and I have been involved with them long enough both to have a long list of (mostly minor) complaints, I have a lot of respect for the staff for doing their best while being paid non-profit wages and having to put up with a lot of disrespect from students and parents. I mostly have experience from the era when they were operating out of Lincoln, Nebraska; around 2015 the Director of Competitions, who was a UNL professor, retired and the MAA decided to move the competitions division to its main headquarters in DC, which led to near-complete staff turnover and loss of institutional memory. My impression is that things have just been going downhill from there, particularly recently, with the DC office having trouble retaining staff. Hopefully the newly hired exective director of the MAA will be able to effect some positive changes, but she has a tough job!

Date: 2026-03-07 04:24 am (UTC)
landofnowhere: (Default)
From: [personal profile] landofnowhere
I'm glad you got things out of your system by writing this rant!

And yes, the information I've gathered does seem to generally support your rumors: it sounds like the recent level of tension is way beyond anything I experienced with the MAA in the past. And I partly went to the lengths to write what I did because I'm aware that my younger self would not have been so willing to extend charity to the MAA, but I've changed with age and experience. (But it's still the case that I'm not going to do any work for the MAA competitions or recommend it to colleagues in anything like the current situation.)

Incidentally: one of the inherent tensions in the MAA is that they are simultaneously the organization behind the major high school math competitions and the professional society for teaching-focused math professors; and while there is some synergy between these two, there are plenty of members and volunteers who support one side of the MAA but don't think much of the other, and recent culture-war polarization has only intensified some of these differences. On top of this, the non-competition side of the MAA gets a substantial amount of its budget from NSF grants of the sort that the current administration really wants to cut. I would guess that it would be require heroic efforts for the MAA not to be a mess under the current circumstances.

Date: 2026-03-04 03:02 pm (UTC)
zdenka: AO3 tag with text "Canon-typical levels of poor decision-making." (bad decisions)
From: [personal profile] zdenka
I'm sorry, that sounds so frustrating!

Date: 2026-03-07 10:22 pm (UTC)
schneefink: River walking among trees, from "Safe" (Default)
From: [personal profile] schneefink
Oof, that is blatant! A difficult problem to deal with but the way they're going about it is not it.

I'll have to ask my brother (who's somewhat connected to the national competitions here) if there's an online option here and how that's handled.

Profile

cahn: (Default)
cahn

March 2026

S M T W T F S
1 2 3 4567
891011121314
15161718192021
22232425262728
293031    

Most Popular Tags

Style Credit

Expand Cut Tags

No cut tags
Page generated Mar. 8th, 2026 01:42 am
Powered by Dreamwidth Studios