But he is slow! He did a bunch of cognitive tests where he was testing 3rd std deviation on most tasks. Except processing speed, where he was in the 2nd percentile. You'd present him with his times tables, and he would take 15 seconds to get "8x7"... but he didn't slow down as as the psychiatrist progressed through the test. He'd get to some 6th grade problem, and it'd be the same 15 seconds and blank look before the solution came out. He was an O(1) machine with a huge constant.
And school utterly killed his interest in math. From grade 2 onwards it was speed competitions and public humiliation in front of his class. After grade 3 he refused to do anything related to math outside of what's required because he was "way to stupid for that".
My school had this thing called mental math, learning to do arithmetic really fast as a large portion of my 4th or 5th grade material. I absolutely sucked at it and I lost interest in any kind of math. My dad who teaches college math for a living was very disappointed when that happened.
Around 8th grade, I started doing some plane geometry and it sort of opened up my mind in a way that's hard to explain. I remember spending a lot of evenings working my way through several geometry books and problem sets. I sort of branched out into other areas and developed enough skill to make it to national olympiads and other such competitions. I don't do math now, but I think it had a reasonable amount of influence of my eventual path through life.
Who knows, maybe he'll pick it back up again in a few years.
"My specificity today among French mathematicians is not to know more mathematics than others, I can even admit that I know less mathematics than most French and foreign mathematicians. It's not about being 'great' either, I'm not great at all, on the contrary I have a rather slow mind. When I listen to a seminar, I'm probably one of those who understand the least what is being said: an idea in an hour, for me, that's already a lot. No, the only explanation for the successes that I have been able to obtain in mathematics is my being imbued by the reading of the great classics."
https://www.biennale-lf.org/b22/index.php?p=https://www.bien...
As for your experience, you are indeed very far from alone.
"In my parents' house, there were a lot of books: all the great authors of the 17th, 18th and 19th centuries, translated foreign authors. I spent my youth, and still today I spend most of my time, reading literary works, and particularly those of French literature."
If this doesn't make you think about math, nothing will.
When I started my undergrad in maths (in the UK) they gave us a day 1 test to check we could recall certain pre-requisites from school. I did poorly because I had forgotten all the tricks for those particular exam questions and was put in an additional "catch up" class. It was a real gut punch and I questioned whether I was cut out for maths at all. Fast-forward six years and I ended up with a PhD in pure maths from the same university. Turned out that when given the time to read and think at my own pace and allowed reference to books and my own notes I was pretty good at abstract math.
Fast-forward another ~15 years and I'm a machine learning researcher for a SV company. The same issue rears it's head now as an inability to contribute usefully in meetings. I can't think fast enough to keep up and don't have anything useful to say when put on the spot. I much prefer to read written communication and take the time to formulate my responses. This is not the company culture though, so I do suffer problems with effective collaboration.
In school I never had a problem with speed or understanding, nor at university. But what hit me was the realization that "actual math" is probably not a load of little riddles. I often wondered how I'd ever solve something that someone hadn't contrived to have an elegant solution, something completely new and insightful that nobody had proven before. There's quite a leap from your tricky olympiad problem that you can find a good video for, and proofs of things in higher math.
The other thing that struck me was that our current system destroys all appreciation for beauty. There's so many beautiful things in math, but we're forced to calculate, quickly. Especially in the engineering side I went into, it ended up becoming very mechanical.
Currently I'm doing an ML research internship at a SV company and this problem of "slowness" at meetings is really resonating with me. I have a hard time following quick verbal communication, I just need the time to think in order to contribute. I don't know it's like I just phase out or forget what we were talking about.
Have you found any useful techniques to do better in this type of environment? Is there any advice you would give your younger self?
I think I know some good strategies, but I often fail to live by them still. Firstly, encourage written/async communication by setting a good example:
- Keep up-to-date written descriptions of projects and progress in a wiki that is available to all colleagues.
- Deliver high quality presentations on your research periodically. These don't have to be reserved for big public conferences.
- Send regular but concise status update e-mails to a list of likely interested colleagues. Advertise your wiki here.
- Send out edited meeting notes within a day or two of working meetings.
General advice I'd give my younger self is pretty cliche, but to trust in my abilities and the value I can provide by working in ways that are comfortable to me. It's more important to take care of my long term mental health than it is to be in the spotlight or maximizing output all the time. Quality work speaks for itself and doesn't need to be continually delivered. Don't get sucked in to false urgency. Delivering a couple of genuinely useful things per year (and publicizing those appropriately) is enough to keep you in people's minds but give you the space to think deeply about the real work.
I say this from a place of privilege though where, somehow, I've managed to accrue a large amount of experience and ended up in a secure research position. The company (and team) I work for give me a lot of autonomy and I have a very supportive manager.
Applying computational algebraic geometry to robotics sounds fascinating. What are some highlight papers in that area?
The bar for SATs and similar college admissions tests is set remarkably low, though, such that one may never run into this problem without going into a mathy field. But it does seem like a slow-and-steady approach taken by a greater number of people could encourage more interest and deep thinking, and possibly a greater contribution to the field and society overall.
There is a known (but mostly ignored) educational method called https://en.wikipedia.org/wiki/Mastery_learning -- you should teach kids until they actually get it, and only then move to the next lesson. This method may be slower at the start, but later gains speed because kids with solid background understand the new concepts faster, and do not need to relearn the old concepts all the time.
If you think about the problem, discuss it, and then implement the solution properly, it's only one ticket. But if you make a half-assed solution, change it twice, and fix seven related bugs, that is ten tickets, with a potential to make even more tickets in the future.
In practice, if I write my own ticket, some manager will give it a priority 2. And the tickets with priority smaller than 1 never get done, because there are always enough tickets with priority 1, i.e. those written by the management. A few years later, when the project is decommissioned, the tickets with priority lower than 1 are still in the backlog. Of course, if having your ticket in the backlog makes you happy, just go ahead and make one.
It is not how it should be done, in theory. Knowing that does not help much.
There's tickets I know we'll never get to, but things we do actually have time for get real attention, with enough focus and resources to do whatever solution is actually correct. There's often tradeoffs for time, but they're ones we have real input in.
He proceeds just as deliberately when doing mathematics. Wang was shocked when he first witnessed it. “I have this math competition experience, that as a mathematician you have to be clever, you have to be fast,” he said. “But June is the opposite. … If you talk to him for five minutes about some calculus problem, you’d think this guy wouldn’t pass a qualifying exam. He’s very slow.” So slow, in fact, that at first Wang thought they were wasting a lot of time on easy problems they already understood. But then he realized that Huh was learning even seemingly simple concepts in a much deeper way — and in precisely the way that would later prove useful.
Having gone through the American education system, I felt like I was in a factory being evaluated as a potential end product. Steamroll through as much material as quickly as possible. Optimize behavioral patterns for the next performance metric. Rinse and repeat. Is this really the best system we can come up with? Can the pace of mathematical and scientific advancement be optimized by steamrollers? I wonder how much potential we've missed as a society because the education system has been trying to force everyone through the same hole, and destroying intellectual curiosity in the process.
Personally, my experience with this educational system turned me off from "math." It was only much later in life when I learnt that the mindless algorithms that I had to personally execute for years on end to get into university could barely be considered math at all. Only then could I begin to appreciate how elegant real math could be. And I doubt I'm alone in my experience.