Tao's argument is weird

1. it's slightly weird that 99% of math education uses already-solved problems. including all tests, the putnam, etc. if unsolved problems were so utterly valuable why do elite math tests still use solved ones? and nearly 100% of all education, up to high school and almost all of college and often beyond that level also exclusively uses only solved problems (not unsolved ones)
2. if the techniques for phd level math are only developed after one has moved into the high level student who allegedly "needs unsolved problems to improve," and also as a side effect, such work helps us learn how to solve those problems, then if we solve them all instantly, PhDs won't be learning how to solve those problems anymore - that is true.
3. if the solutions are valuable, then losing the ability to do so is bad; hold on though - we just gained the ability to solve them, and really fast, too, from AI. hmm, so AI is overall net valuable here*
4. if the solutions are not valuable, then what are we paying the math guys for?
5. so, Tao claims that the skills & techniques around solving these problems are what's valuable
6. this is so obviously a backfilled justification (since all prior math, during my entire life, nearly 100% rewarded being able to get solutions as the primary target; you 100% couldn't get a PhD for being very good at 3b1b type work. so it's BS.)
7. also, this is at risk of falling to nextgen AI: if they gain the ability to explain math to humans, to come up with new problems, and to generate new solution techniques. admittedly not proven yet. another win case for tao is if AIs stop improving now, and so never get the ability to do things which can only e done by humans which have been exposed to unsolved problems.

so basically, if AIs solving unsolved problems breaks the production chain for future unsolved problems, such as the ability to conjecture, or to care, etc. then we may be in a bind.

the only case Tao's proposal is right is when AIs are good at solving problems but not generating human-useful techniques or understanding, AND this understanding is still useful for humans in the future (for example: it feeds back to prove earlier education levels)

Is Shaq the best coach, or the best NBA administrator?

secondly, Tao relies on the unstated assumption that "the people best at a thing, should also guide a thing." i.e. the best eaters make the best cooks, the best movie critics make the best directors, the best basketball players make the best league sport regulators, or the best refs, or the best coaches. even in the NBA, the taller coaches (among explayers) have lower winrates; when meeting someone who's a former NBA-level player, people don't discount the height effect enough! (that is, people who are 6'6" and in the NBA are using up a lot of the selection power to be them (to be a 1 in a million type of person) on height; someone who was at that level but only 6'0" is probably way, way more special in other areas of life.) humans fail to fully discount "visible natural gifts" when trying to use achievement to evaluate ability later on. in fact, super fat sumo, super huge boxers, etc, are on average lower than their not-so-extremee peers within the remaining selection criteria.

so the idea that people who are good at a thing are also good at running the thing, is generally extremely untrue.

in math it might apply a bit more since the vector of "math" might lie close to the one for general IQ, or general IQ + management ability, but they're still not the same.

in fact, Tao is basically a random person who is extremely good at math; in the same way that Shaq (7' 1", 325lbs) was naturally good at basketball, and Akebono the sumo wrestler (6' 8", 530 lbs) was naturally great at tossing guys around.

in many respects, this kind of person should be listened to LESS than the average;

also, Tao has never been elected to anything like "president of the association of math lovers" and he has basically zero claim to represent them.

TL;DR: bite the bullet and fully accept elite math people go away; the cases where that's bad from a pragmatic pov are actually pretty thin.

the same way that the world's best at during-battle cavalry horsemanship today is probably lower, across the entire planet than it was 100 years ago. it's okay. forgetting is a thing

If aliens HAD found all our math results 10m years prior, would that harm Tao? What if we were told that they had, but they hadn't actually? etc. His claims of harm are very vague.

he's seeing the change and how bad it is for him and pretending its bad for everyone, and doing lots of scare-work at imagining all kinds of bad consequences, because he's the one set to lose his status. but he's doing no work looking for potential upsides, so he isn't seeing them.

Economics - It's not my job to make sure my legal actions have no economic consequences

i don't see how a world where one person's legal, good-directed actions can be blocked w/out any law in place when he did them, just because they change the economics of the world. if i discover a cheaper way to make, say, bread, the world could change a lot. if the method is legal, our whole system of law is set up to prevent special interest groups from being able to stop me. as is our patent system. so i say, I'm not responsible, because there is no moral system where I can be responsible; we have rights to do things and no ex post facto laws are allowed, and we have the right to do things generally w/out committees. can you imagine the world tao wants - an unelected, naturally gifted guy dominates a field; someone invents a better way to get results in that field; it would be absurd to listen to the guy and take his advice. AS IF!

imagine the headline: "US army agrees w/sword experts - Guns weaken character - army dumping artillery, broadly reintroducing swordplay"

nah, actually. sword guys, thanks for your service, learn to use guns now.

EU-style claims that we all have some kind of shared "responsibility for sustainable development" aren't right.

there's just no way to know what will happen. if your default position is to block things and let some expert come in and talk about how disrupting the status quo is bad, well, that's all tautological. Of course there will be an expert saying that. it's contentless to sample the world for people who make money and who don't want that to stop, and ask them "can you find me an expert to say you should keep making money, and indeed, make more?" yes they'll be able to do that.

there simply is no general practice we can follow, given the cost of sharing and understanding the world, which is anything other than "do good things and don't do bad". As much as people love to talk about this kind of weird situation where doing something normally bad has a good outcome, it's GOT to be rare, for very strong reasons; if it were generally true that doing bad were actually good, well, why do we define the states that way? you can try to be difficult and hedge and hem and haw if I ask you "hey, we invented a new planting method which produces more corn per acre;". I'll answer your questions about if it harms the soil, or the birds, etc and if it doesn't, i.e. it truly is just an improvement, and you have all the time you want to study and search and can't find any downside - you may not still stick to "well, I think we shouldn't do it"

ofc people CAN make actual mistakes; and experts are wrong sometimes. the enemy is not the one who disagrees with convention; it's just cases where people refuse to run tests or let others do so.

"akshually, we should produce LESS corn for the starving peasants because (maybe?) that will lead to locust population crises, meaning the crows die out, which prevents the spread of avian flu, which means our duck-raisers will raise 100x as many next year, which will in the end help us overwhelm the enemy with our copious stores of duck jerky and cross the uncrossable-due-to-no-food desert". Like, this CAN be the case sometimes, but if you know nothing at all except that in general "growing more corn is good", then until you prove otherwise, grow more corn. you can't, with no data, just contradict someone who's advocating "let's do well-known-to-be-good thing more". Ofc I do admit things like "adding more armor" which have natural pros and cons, were never pure gains in the first place!

So, this math thing is the same: overall, the entire setup was designed to find and promote and pay and give time to smart people, to solve math rpoblems. we now know how to solve them even faster; great, do it. I do not have to entertain the hypothetical fear-chain nightmare-visions of dethroned elites.

It also grants too much power to bad arguments

EVERY status quo is VERY GOOD at making up reasons not to change it. do you think the guy who got himself declared "head of the union of tasmanian window washers" is just screwing around and having fun? No; he's there because his constituency was selected partly for its ability to protect itself locally; so there is no doubt that they would oppose mechanization and replacement; they are pro-expansion of mandatory window-minimums for city buildings; pro-increasing the mandatory annual number of window cleanings; does the sun rise in the morning? do groups you observe (in contrast to ones which were destroyed, or which faded away) have a tendency to protect themselves? how could it be any other way? (it could be, obviously, with appropiate laws, culture, etc. but even in such a world, they'd be well-served to have an accurate understanding of how rare it is to be within such a naturally semi-stable system, and how hard it is to evolve things like that by chance)

So, these groups will have copious studies proving that clean windows actually leads to better public health, happiness, purity; in their hagiographies they credit the solitude of the window washer's high-altitude life with all sorts of spiritual benefits. overcoming the fear of heights isn't just a side effect of the job; all men (and women-washers!) really should go through it; history will be interpreted to find that even jesus was a window-washer of a kind by detailed textual analysis and vocabulary note. and for the teenagers out there: the glimpses of beautiful women seem scandalous, but at the end of the day, many couples owe their present non-lonely condition to this! so, change it at your peril. it's NOT hard to make this kind of thing up for nearly any status quo practice.

People are very good at making up stories about the SEEN, and very bad at thinking about things UNSEEN

Tao fails to distinguish himself from every other bad version of this.

when someone makes arguments that don't distinguish themselves from the worst, dumbest defense of any existing earth system ever: the fall of horse riding; the rise of chariots; the fall of chariots and the rise of carriages; like, literally every product we use is the 10th generation of iteration; every generation obsoleted the one before; and the guys who made the 9th version and earlier are at 1% or less of their former power; when they make this argument, they should realize the pattern they're in, and must try to distinguish themselves and explain how "why no, it's true that I do benefit a ton from the status quo, and yes I am arguing for absolute status quo protection, and I realize that everyone hearing arguments of that pattern should meta-bias very heavily against; I acknowledge this and do owe it to you to explain why, compared to other cases, this actually is a rare case where we really should think twice, for all our benefit, not just ours!"

tao had a DUTY, concretely, to have noticed the very suspicous arg pattern he was making, admit it, and distinguish himself from the fully-predictable horde of status-quo-defenders we should expect to, and actually DO, see everywhere.

in contrast, tao is using innuendo, and awful super negative connotation metaphors like "strip mining of open problems" which don't acknowledge the issue and are nothing more than low cheating.

Why status quo defense is so powerful and common

MANY systems, even horrible ones like medieval medicine based on the 4 humors, or the USA's obsolete and wasteful highly priced eyeglass system (100$+ a pair, when the marginal cost is 3$), accumulate socially beneficial customs around themselves; even if a system of medicine is actively harming patients, beneficial variants and traditions of activity will still grow around it since the participants are still human. i.e., there will be annual holiday parties; fun conferences you can hope to be invited to travel to and spend a week; addictive status hierarchies; ways to use power to gain status and wealth and access; chances to see old friends and scholars; memorials and gratitude ceremonies for innovators, stoking valid hopes, and inspiring the youth and leading to social unity; the associations, friendships, and class reunions held within the training schools will still lead to lifetime friendships, faith and trust in ones fellow man; the money those schools make, despite their content being false, will still be donated to important charities; their architecture may still be lovely and traditional. that the actual medical theories they spread are false is irrelevant to those good things.

this tendency applies to ALL beliefs and all status quos. so, a religion having beautiful buildings is not actually a defense of it - since nearly all religions have that trait.

so Tao may talk about the joyful deep relationships of prof & grad student, or the feeling of achievement one gets when solving a problem; but I claim that seminaries of false religions can produce these just as reliably! the only thing they can't do, is to be about actual universal truth, which math can. But: AI math and human math are STILL BOTH about this same truth.

so, Tao is acting out the classical status quo defense argument of a born elite. I have appreciated his work; him doing this has proven that elite math talent is NOT actually very aligned with good understanding of the human social condition.