YT YouTube2Text

YouTube2Text → Transcripts → По ту сторону медалей. Как устроено олимпиадное сообщество в России? (yet another podcast #56)

По ту сторону медалей. Как устроено олимпиадное сообщество в России? (yet another podcast #56) — Transcript

by Яндекс · 10,513 words · 1,850 segments · language en · Watch on YouTube

Open in transcriber Summarize with AI Download SRT/VTT

Full transcript

0:00Schoolchildren are sitting there,
0:01programming something on the computer.
0:03>> You sit for five hours at your Saturday
0:05club.
0:05>> I'm not an Olympiad participant; I find
0:07it hard to sit still. I definitely can't
0:09sit for five hours.
0:10>> This is IT, and they make a lot of money
0:12in IT. I need my child to become a
0:14programmer.
0:15>> What about programming? Do you think
0:17Andrey Sergeyevich has any prospects?
0:27This is our podcast.
0:30An episode dedicated to competitive
0:33programming.
0:35Fortunately, this sport has some kind of
0:38rest now, a vacation, a short break.
0:41The international competitions ended
0:43recently.
0:44The whole country essentially honored
0:46the winners of our last international
0:48Olympiad.
0:50And everyone involved seems to have some
0:53time, outside of the frenzy, to reflect
0:56, think, and try to make this podcast
1:01interesting for everyone, even for those
1:03who don't really understand what Olympic
1:05programming or sports programming is.
1:08This is how I often apologize for my
1:10stupid questions to guests.
1:12This is Andrey Stankevich, the dean of
1:15ITMO University, better known as the
1:17world's most decorated coach in this
1:20sport.
1:21Tatyana Klenkova, who heads the team at
1:23Yandex working with the Olympic movement
1:26in all its forms.
1:28And you, Tatyana, are you a coach, a
1:30mentor? What are you?
1:31>> I think I'm more of an educator.
1:35But closer to this terminology, then
1:37probably a mentor who is an
1:39>> Educator—normal—
1:40>> says, please, don't worry, nothing
1:42terrible has happened.
1:43>> Can we go out there and do this?
1:45>> No, that's not the case.
1:46>> No? Is that a bit of a different matter
1:48?
1:48>> No, they're more like, "Tanya, what
1:50should we do?"
1:51>> What are we going to do?
1:54So, sports programming, Olympic
1:55programming—it's a sport.
1:58That's true.
2:00How big is it? Well, because it's
2:02obvious it's not in the Olympiad. But
2:04the Olympiad doesn't have a single
2:06intellectual sport. What do we have?
2:08There's chess, checkers, probably some
2:11card games, some poker, complex bridge.
2:15Oh, what, eSports?
2:18Where does your Olympic programming rank
2:20in this ranking in terms of the number
2:23of people involved, well, those
2:25interested at some amateur level, at the
2:27professional athlete level, and in terms
2:30of audience?
2:32I just looked, 600 million people play
2:35chess worldwide.
2:37And fewer watch.
2:40Like, 500 million. A lot? Well, it's not
2:42clear how often they watch, right?
2:44Because it's one thing to play, and
2:46another to watch.
2:47>> Of course, significantly less than chess
2:49, than card games.
2:51Well, card games, it's obvious where
2:53they come from—from some kind of
2:55gambling, apparently. It's not really
2:57sports programming. Well, it's clear,
2:59yes. Competitive programming started
3:01from a completely different direction.
3:04Its origins are probably an attempt to
3:06model computer competitions on American
3:08university sports competitions.
3:10That is, all this terminology, all these
3:13leagues, coaches, trainers, and so on—
3:15it all came from there.
3:17>> If we were to use some numbers, ah, yes
3:20,
3:20>> athletes, for example?
3:22>> Now, no one can say how many athletes
3:24there are, of course, because it's a
3:26somewhat vague concept, like an Olympian
3:28.
3:29>> How many athletes do you have in the
3:30Yandex Cup?
3:32>> In the Yandex Cup finals, I can tell you
3:34exactly how many athletes I have.
3:36If we look at the tracking algorithm
3:38there, which most closely intersects
3:40with the topic of our discussion today,
3:41I have 20 adults,
3:43>> 20 more,
3:45>> children, and 10 employees. And at the
3:47selection, well, at the selection, well
3:49, okay, we're aiming for 20,000
3:50registrations, conditionally, and we
3:52expect that nine-plus of them will be
3:54right on the algorithm track.
3:56>> 10,000, conditionally speaking.
3:57>> Conditionally, yes.
3:58>> And in the ICPC?
3:59>> Well, in the ICPC, if it seems, if you
4:01measure it worldwide, well, probably,
4:03maybe 30 thousand teams, that is, well,
4:04100,000 people.
4:05>> 100,000 people are active athletes.
4:08That's
4:08>> according to the ICPC format, yes. If we
4:10go super lower...
4:11>> And the ICPC is only students, right?
4:13It's
4:14>> only for students, yes. We also have the
4:16All-Russian. This is one of Andrey's
4:18regalia. Andrey is the chairman of the
4:19central subject-methodological
4:21commission. Andrey is the head jury.
4:22>> On the other hand, there is a duplicate
4:24.
4:24>> And if we look at the All-Russian, the
4:26All-Russian has four stages. It's like a
4:28funnel that descends from the wide part
4:31.
4:31>> Well, how many?
4:32>> And in the widest part, it varies from
4:34year to year, because birth rates vary,
4:36and we depend on that, too. There are
4:38600,000 people there.
4:40>> Well, there are 600,000 participants,
4:42but I wouldn't say that all these
4:44participations are even legitimate, let
4:46alone athletes?
4:48Well, legitimate, yes, because at the
4:50school stage, there is no oversight at
4:52all. So how many of these 600,000
4:54participants are actually disguised by
4:57some large Chinese neural network under
4:59the name of a living person, we don't
5:02know.
5:03>> So, we're talking about millions, after
5:04all. Well, look, we even have half a
5:07million, some dubious ones there.
5:11Russia isn't the most populous country
5:12in the world, but there should probably
5:14be a couple of million worldwide, if not
5:16more.
5:17>> If there are 600 million in chess, we'll
5:19probably have a million.
5:20>> By the way, are programming languages a
5:24tool?
5:25Well, I mean, you can't really jump with
5:27springs in the high jump, right?
5:30They have six. It's certified somehow.
5:33What's certified in competitive
5:35programming?
5:36>> I don't know, there's no equivalent, I
5:37don't think it can be done here. Nothing
5:38in competitive programming is certified
5:40at all. In general, competitive
5:42programming is radically different from
5:44all other sports, even intellectual ones
5:46.
5:46>> That's
5:47>> because, like in chess, there's an 8x8
5:50board, knights, bishops, rooks—they know
5:52how to move.
5:54In Go, the board is 19x19, there are red
5:56, white, black, we place the pieces,
5:59everything is fine.
6:00And even in eSports, they run on a fixed
6:04map. In competitive programming, the
6:06tasks are different every time.
6:07>> Yeah.
6:08>> So, in this sense, I don't even know
6:11what other
6:14sports I could give as an example? Well
6:16, maybe some kind of sports
6:18>> confrontation.
6:20No-no-no, well, like any kind of contact
6:23confrontation.
6:25>> There is also a map, to be fair. She's
6:26>> not her own anymore, probably, maybe
6:28she's kind of her own somewhere.
6:29>> Well, how? It's boxing. Every time,
6:32everything is very different.
6:34>> Oh, yes, but the ring is the same, the
6:36gloves are the same, the opponent
6:37behaves differently. That's true. But
6:40the input conditions, they are generally
6:41the same. Here, the input conditions,
6:44what is given, are new every time. And
6:46so, in this sense,
6:47>> there are standard, repeating conditions
6:49. Two or one second, 512 or 256 MB.
6:52>> Well, these are trifles. What is this?
6:55>> No, it seems to me that what makes our
6:56sport athletic are some artificial
6:58restrictions imposed by the jury.
7:01You can have a partial solution, you can
7:02have a complete solution, because you
7:04can solve the problem piece by piece,
7:05part by part.
7:06And you, conditionally, in each problem
7:08there is a certain group of tests, which
7:11and if...
7:11>> Let's do it this way. These are
7:12mathematical, relatively speaking, or
7:14logical problems that need to be solved
7:16with code.
7:17And the code must pass the test.
7:19>> A program that satisfies all the given
7:21restrictions.
7:23And if something happens, and your
7:24solution, for example, isn't very
7:26optimal, you can score some points, but
7:27if it's super, it's
7:29>> not a matter of solving it or not, it
7:30depends on the situation.
7:32>> Different formats, yes. For example, in
7:33solving it or not solving it.
7:34>> In individual competitions, there are
7:36usually points.
7:37The better your solution, the faster or
7:40further you go, the more points. And in
7:42team competitions, solving it or not.
7:44>> And I think you also mentioned memory.
7:47>> And each problem has some limitation,
7:53like the time limit for working on the
7:59written solution.
8:02It's
8:03>> expensive now, too, there's a shortage.
8:05>> Well, we can get into the weeds here.
8:08>> Well, why? What for then?
8:10>> Roughly speaking, usually a computer has
8:12infinite memory, both now and in the
8:14past, and it's possible to gobble up all
8:16of it in two seconds and make the
8:18computer suffer, figuratively speaking,
8:20start slowing down.
8:22>> So, this is some kind of optimization
8:24problem, right?
8:26>> So, to make the computer that's checking
8:29your solution crash, yes, freeze. It's a
8:32small thing; I don't think it's crucial
8:34at all.
8:35Time is a crucial constraint, because
8:38otherwise it's unclear whether the
8:41program will ever stop or never stop—
8:44it's an impossible, algorithmically
8:47unsolvable problem, so to make testing
8:50possible, we have to limit time.
8:54>> Tell me, every sport has its great
8:57representatives who know everything
9:00without even asking.
9:04I'm not sure you watch football, but you
9:06probably know who Messi, Ronaldo, or
9:08Pele are.
9:09Do you have superheroes like that?
9:12>> I don't think everyone knows them, of
9:13course. Programming, after all,
9:15>> has its superheroes.
9:17>> Anyone who knows programming, people in
9:20IT, of course, know Gennady Korotkevich
9:22, or, I think, from more ancient times,
9:24maybe Petya Mitrichev, or maybe Kolya
9:27Durov and Andrey Lopatin.
9:29But people off the street, of course,
9:31don't know anyone.
9:32>> And the main thing for you, as I
9:34understand it, is ratings, right? Back
9:35then, it was TopCoder, and Codeforces
9:37was just starting, I think.
9:40Now, I don't know what's going on there
9:42. What is it? Tell me about these
9:43ratings. What is it anyway?
9:45>> Well, it's not like ratings are the main
9:48thing there. Well, like in chess,
9:51there's something like ELO. They wanted
9:53to come up with something similar for
9:54competitive programming due to the
9:55smaller number of players.
9:57It's less stable there, but nevertheless
9:59, a similar rating, the formula, I think
10:00, was taken directly from there.
10:02The main property of a rating, for it to
10:04be interesting at all,
10:07>> Yeah. It
10:07>> needs to be relevant, so that
10:09competitions are constantly taking place
10:11—the ICPC World Championship once a year
10:13, the IOI once a year—people don't
10:14overlap there at all. You can
10:16participate in the ICPC often, but there
10:17you only have two. Basically, the only
10:19way to measure your achievements in the
10:21ICPC is by some kind of achievement.
10:22Well, I don't know, I can't compare it
10:24to chess, in football, I don't know, the
10:26Russian Championship or there,
10:27>> well,
10:30>> Topcoder, they held competitions every
10:32week.
10:33There were actually three or four a
10:35month.
10:36And so the ratings were relevant. You
10:38could
10:39>> see who was moving up, who was losing
10:41interest, or who was losing interest
10:43.
10:43>> And there were those at the very
10:45beginning, which even I didn't really
10:46experience in Russia.
10:48Topkoder's popularity came later.
10:51At the very beginning, there were cash
10:53prizes right at these Topkoder
10:54championships, right in these night
10:55rounds.
10:56Night, because they were. Well, they
10:58were during the day in America, like in
10:59the evening at 6:00 PM, and in Russia it
11:01was
11:01>> Night.
11:02>> Night.
11:02>> But for money.
11:04>> But for money, you could win $ 200 in
11:06rounds.
11:07That's where one of the rating-related
11:10memes came from—this is a word no one
11:13knows anymore—target.
11:15This is a person with a rating of over
11:183,000, they have a white dot inside
11:20their red circle.
11:22There, too, like on Codeforces now, they
11:24used colors to differentiate people by
11:27rating.
11:28Red was the coolest.
11:29>> Differentiation—pants and lines.
11:32>> And lines, yes, nicknames.
11:33And whoever had a red nickname could
11:36become even cooler, and they would get a
11:38white dot inside that dot that was drawn
11:40next to their nickname on Topcoder.
11:44>> How did I get interested? Because, well
11:46, I'm far from competitive programming,
11:49but I came across some facts about Petya
11:52Mitrichev somewhere online.
11:55This was around 2012, 2011, or 2013. And
11:57these were facts about a guy named
12:00Kinoris, only about competitive
12:02programming.
12:04That Petya Mitrichev doesn't compile
12:07programs, he just checks the compiler
12:09for errors, like, well, like, doesn't do
12:11push-ups, pushes the earth away from
12:14himself, and all that.
12:18Uh, to my utter surprise, five years
12:20later, I found roughly the same list,
12:22only about another super champion of the
12:24next generation.
12:25This one, actually, is about Gennady
12:27Kratkevich.
12:29And I found an English-language version
12:31.
12:32It's not like our guys are keeping up,
12:34it's how they honor a great champion, as
12:36I understand it.
12:38So, Petya Mitrichev and Gena Kratkevich
12:40—they're supposedly the greatest sports
12:43programming Olympic programmers.
12:45>> Yes.
12:46Of the global community.
12:49Besides Gena, apparently, and Petya,
12:51especially now, there are more Chinese.
12:53That is, there are simply more Chinese.
12:55>> Yes.
12:56>> And they are now at the top, the Koreans
12:58, the Chinese, and Southeast Asia in
13:00general.
13:02>> But tell me, yes, again, sports have, uh
13:05, some geographical reference points.
13:10Football is Brazil, everyone knows.
13:12Whoever's a Brazilian, they're a
13:14football player.
13:16Hockey is probably Canada and Russia,
13:18well, I'm not sure, maybe some
13:19Scandinavians are there too.
13:22And, sports programming. It's just that
13:24when I was looking into it, I was amazed
13:26that China and a bunch of socialist bloc
13:28countries were there.
13:31That is, like Poland, Hungary, us.
13:37Has this changed at all or is it still
13:38the same? Romania. And what's the reason
13:41for that? What's the
13:41>> reason that at the dawn of competitive
13:44programming, it was Eastern Europe and
13:46China. China, by the way, didn't exist
13:47at the dawn of competitive programming;
13:49it appeared later, I think. It was
13:50Eastern Europe at the dawn.
13:53This probably has to do with certain
13:55traditions of school education: more
13:57mathematics, more natural sciences, less
14:00of what's now fashionable for soft
14:02skills and the like.
14:05That's why these people, they got into
14:07it, were aiming for IT, among other
14:09things, and proper IT education probably
14:12didn't exist back in the early 2000s.
14:14Simply preparing for competitive
14:16programming competitions, participating
14:19in them—it could have simply been some
14:22form of additional education in
14:24algorithms, IT, and the like. And it was
14:26popular in Eastern Europe, including
14:28because people there perhaps dreamed of
14:30getting jobs there—there's a
14:31>> professional history there.
14:34They don't play IT in the US, but they
14:36have jobs there.
14:37We want to be noticed. Look, we're the
14:40best.
14:40>> Sort of, yes.
14:41>> And how did China join in then? It's
14:44>> just that at some point, competitive
14:46programming and programming olympiads
14:48also became popular there. And the
14:50Chinese, due to their numbers and,
14:53apparently, the growth of this
14:55transformation of school, and maybe even
14:57university, education in China, also
14:59found themselves at the top. We've
15:02>> already talked about top coders and
15:04Codeforces.
15:04>> Yeah.
15:06>> But Codeforces isn't just a Russian
15:08thing; it's also regional, right?
15:10>> Saratov, yes.
15:11>> Saratov.
15:12>> But that's not even the point.
15:14Indeed, Tanya is right; there were
15:16several, so to speak, schools of this
15:18kind of competitive programming in
15:20Yekaterinburg, in Saratov.
15:25>> Novosibirsk.
15:26>> In Novosibirsk, to a certain extent, yes
15:28.
15:28>> In St. Petersburg and Moscow.
15:29>> In St. Petersburg and Moscow, in fact,
15:30several of each, as we already said.
15:32Even now, there are more of them. In
15:33general, school Olympiads and student
15:35Olympiads are two completely different
15:36stories.
15:37>> Yeah. They're
15:38>> similar because the problems are the
15:40same and the rules are similar.
15:43But it seems to me that the motivation
15:45in different periods is also different.
15:48>> I remember 13 years ago, when I
15:50interviewed Andrey's colleague, Mr.
15:52Parfenov, who was then, as I understand
15:54it, in your position, yes, the dean,
15:56what did he say?
15:59The Chinese wave is coming, we need to,
16:03and we're losing ground.
16:05>> Yes, there are a lot of Chinese, and
16:07Parfenov sensed that even then.
16:09>> But he could probably also sense that
16:11there are a lot of Indian citizens.
16:1450%of the people who go, they're
16:17supposedly from India. And they go to
16:21BigTech.
16:24Google, Apple, anywhere.
16:27And
16:27>> other startups and not only that, yes.
16:30>> This, of course, is not competitive
16:31programming.
16:33Let's call it that, but this is
16:35programming to pay for a living.
16:41How did you get involved? Tell me.
16:43Here's your path.
16:46Well, because I went to three Olympiads
16:48, I think three math ones, two physics
16:49ones.
16:51Something there, somewhere somehow.
16:54But there's also life.
16:57There's also, I don't know,
16:59non-intellectual sports.
17:02And in general, it's quite difficult,
17:04you have to slog through it.
17:06And I didn't get into it.
17:08But you, obviously, maybe in different
17:10ways, got into it.
17:12How? For
17:13>> me, it was, first of all, a long time
17:15ago. Even in school, I was always
17:16interested in computers.
17:18Computers were my dream at first, then,
17:20when I already had a computer, it was
17:22the opportunity to do something with it
17:23.
17:24I was always extremely interested. I
17:26don't know why, there's no rational
17:28explanation. And then, in fact, it's
17:30very important to get into a community,
17:32when I...
17:33>> Into a community or this trainer-teacher
17:35?
17:36>> For me, it was into a community, maybe
17:38for someone else, some kind of
17:39trainer-teacher.
17:41For me, it was exactly this: first a
17:43community, then a coach-teacher, because
17:45at first I was somehow studying on my
17:47own.
17:48Then I found out somewhere that there
17:50was a programming club somewhere else,
17:52in our area.
17:54I went there.
17:55Then I...I had enough to get into the
17:58Olympiad myself.
17:59So, there in class, the teacher told me
18:01to come to the Olympiad. I went, went
18:03somewhere, and ended up in St.
18:05Petersburg at the training camp at
18:07Sirius. Well, and there I met Durov, and
18:09Lopatin, and everyone else.
18:11And I ended up in the community.
18:13>> Durov Sr., yes, Nikolai.
18:15>> Nikolai.
18:16>> And Lopatin, who later became a coach at
18:17St. Petersburg State University, right?
18:19>> St. Petersburg State University. And
18:21then there
18:22>> I had my first teacher.
18:25It was, in fact, ITMO professor Sergei
18:28Evgenievich Stolyar, who was my first
18:31teacher.
18:33And then, already at university, the
18:34coach there, Mark Sandler, Matvey
18:36Kazakov, Roman Elizarov, are already my
18:38teachers.
18:39>> You could say that you, at least for
18:41yourself, also had a great battle.
18:44I understand that you lost a couple of
18:46times at the ICPC to Durov and Lopatin's
18:49team, right?
18:50Was that a story?
18:51>> To put it mildly, back then, they were
18:53probably an order of magnitude stronger
18:55than us in terms of athletic performance
18:57. They had been training there since
18:59almost the seventh grade.
19:01>> So, it matters when you start?
19:02>> Since the eleventh grade. Personally, I
19:04achieved serious success since the
19:06eleventh grade. I trained there since
19:07the ninth. My teammates—well, maybe the
19:09tenth. But we tried our best. We trained
19:12a lot back then, we really focused on
19:15trying to catch up with them. We didn't
19:17succeed both times, of course, and they
19:19beat us. Which grade should we start
19:21with?
19:22In our case, it probably wasn't even as
19:24important as the fact that they were
19:26simply tougher.
19:27>> Listen, Durov is still an amazing
19:30character.
19:32Because he's completely private, it's as
19:34if he's not celebrated the way Ratkevich
19:36and Dmitrichev are.
19:38And they're probably on the same level.
19:40>> Well, he was before the social media era
19:42.
19:43>> Because he wasn't able to. Because the
19:46younger brother created social media,
19:48and
19:48>> they created it with his younger brother
19:51, Davik, right back in the early 2000s,
19:53when his Olympic career was already over
19:55.
19:56>> And that's why we don't really know
19:58about it, right?
20:00>> Yeah.
20:01>> It was a very unexpected take. Why
20:04doesn't anyone know about the elder
20:06Durov?
20:08Because the younger brother came up with
20:09the idea of creating social media after
20:11he retired from competition.
20:13It's really funny. The club turned out
20:16to be the most important thing for
20:18Andrey. I understand that you're
20:20basically involved in the club, right?
20:22It's your main job. And how did you end
20:24up in the club?
20:25>> Already, well, being a relatively adult
20:28, like not a child,
20:30>> yep,
20:31>> I worked for a company that had a good
20:33relationship with MIPT, uh, with the
20:35university.
20:37And so, as part of their new office,
20:39they donated, uh, a wing of MIPT so that
20:42students could come to classes and so
20:44that the faculty who don't live on
20:47campus but live in Moscow and work, for
20:49example, at Yandex or somewhere else, uh
20:52, they could, uh, meet somewhere in
20:54Moscow and, uh, so these busy people—
20:57the faculty—don't have to travel to
20:59Dolgoprudny.
21:02And following the donated wing,
21:03following the donated classrooms, the
21:05Olympiads came into my life.
21:07And two key people appeared in my life,
21:10I think, who, uh, influenced my entire
21:13career in the future.
21:14>> yep.
21:15>> It wasn't Andrey. Well, that's the third
21:17one.
21:17>> Great.
21:20>> First, of course, we have a wonderful,
21:23beloved woman named Elena Vladimirovna
21:25Andreeva.
21:27>> Right. She's
21:28>> currently, uh, the chair of the computer
21:29science department at the Specialized
21:30Educational and Scientific Center of
21:32Moscow State University. And, uh, the
21:33second important person is Denis
21:35Pavlovich Kiriyenko.
21:38He's a teacher at School No. 179. And he
21:40and Andrey are already members of the
21:41All-Russian jury.
21:43>> As someone who's been observing from the
21:45outside, uh, how long is the first step
21:47?
21:49What grade? When should you join a club
21:51?
21:52>> We say you can start in the seventh or
21:54eighth grade, uh.
21:56>> Seriously? Not earlier?
21:58>> I think Andrey and I had a situation and
22:00a phenomenon in our careers.
22:03We talked about famous adults like Petya
22:05and Gena, and among the children, we
22:07also had this phenomenon like Yegor
22:09Lifar.
22:10And so, Yegor Lifar, like Gena, started
22:13in fourth grade. Except Gena went to the
22:16international competition in fourth
22:17grade, and Yegor came to the All-Russian
22:18competition in fourth grade.
22:20And I like to joke that it was our first
22:21All-Russian competition together. I came
22:23back then just to watch how things were
22:25going.
22:26>> He came to the All-Russian Olympiad, the
22:27>> finals of the All-Russian School
22:29Olympiad. It was in Yekaterinburg, and I
22:31remember that he was
22:32>> younger than everyone else. By how much
22:33?
22:34>> By a lot.
22:35>> Well, by two or three years, for sure.
22:36>> By three years, yes. It's clear that now
22:39in our community there are some outliers
22:43like Yegor, some phenomena like him, who
22:45started early, who are studying, who are
22:47succeeding, and so on and so forth.
22:51But, unfortunately, this has also had a
22:53negative effect, because there are
22:54parents who live by their own
22:56unfulfilled ambitions, which they apply
22:58to their children.
23:00In seventh grade, my child went to his
23:01first international competition. And so
23:03, the parents who follow him, who live
23:05by the paradigm that this is IT, that IT
23:08makes a lot of money, I need my child to
23:10become a programmer. They learn about
23:13competitive programming and the benefits
23:15it offers.
23:17So, they're all trying to send their
23:20child down the path of Yegor Lifar,
23:22despite completely ignoring the fact
23:24that Yegor had been trained much earlier
23:26, that he simply had a predisposition,
23:29and that's what it is.
23:31And that's why, in Yegor's case, some
23:33things worked out, some didn't, but...
23:35No
23:36,
23:36>> well, if you do it earlier, you'll
23:38probably achieve more, or you'll quickly
23:40test a hypothesis that it's not right
23:42for you.
23:43>> Testing a hypothesis and stopping if the
23:46results are unsuccessful is one thing.
23:48>> Right.
23:49>> But most parents weren't deterred by
23:51that, at least in that situation.
23:53There's
23:54>> another question: what if you stop
23:56testing a hypothesis if it's not your
23:58thing? It could be a false start, a
24:00wrong answer. Maybe it's not your thing
24:02yet, but in tenth grade it will be your
24:04thing, and you already think it's not
24:06your thing.
24:07The ideal situation is if someone is
24:09interested in it at that time.
24:11That's when they first became interested
24:13.
24:13If they're interested in programming,
24:17not playing computer games, like I was.
24:21I think maybe games are more
24:23brain-consuming now than they were in
24:25the 1990s, although that's not a given.
24:28You could start in fifth, sixth, or
24:31seventh grade.
24:33But it probably makes sense to take your
24:36first classes specifically focused on
24:38solving Olympiad problems in seventh or
24:40eighth grade.
24:42It's definitely not too late to join the
24:44community in ninth grade.
24:46>> And when is too late?
24:47>> And in nine and ninth grade, it can
24:48already be a conscious choice, including
24:50the student themselves, not just
24:51something they're abstractly interested
24:53in, but something that will eventually
24:55lead them into the community or into a
24:56certain class within the community.
24:59And it's never too late. I'll give you
25:01an example, although it's a bit outdated
25:03, but I had a student who became a
25:04two-time world champion in programming
25:06in his third and sixth years. His name
25:08is Zhenya Kapun.
25:09He studied physics at school and didn't
25:11know how to program. In his first year,
25:13he programmed with one finger, like this
25:15. He became world champion in his third
25:18year. After that, in his fourth year, he
25:20tried but failed to make it to the
25:22finals.
25:23In his fifth year, he was already world
25:24champion on a different team, but failed
25:26to make it to the finals. In his sixth
25:28year, he didn't give up.
25:30He made it to the finals again and again
25:32became world champion on a different
25:33team.
25:34>> Another real gem.
25:36>> So it's never too late, but from the
25:38perspective of school Olympiads, eighth
25:40and ninth grades are probably optimal.
25:43Seventh isn't too early, tenth isn't
25:46exactly late, but unless you're growing
25:48really fast, it'll just be hard.
25:52Maybe, and if you have really high goals
25:55, then starting in tenth grade is
25:57probably a good idea.
26:00It's realistic, considering that only
26:02then, if you're really...
26:04>> But we have another example. We have
26:07Tikhon Evteev, who learned programming
26:08when he was writing a selection circle.
26:10These are
26:10>> isolated examples that only confirm
26:13statistics.
26:15If we look at the larger statistics, the
26:17majority of people start between seventh
26:19and ninth grades.
26:21>> Speaking of results, he later, in
26:22addition to winning the All-Russian
26:24competition, also won two ICPC gold
26:26medals.
26:26>> Starting in seventh grade.
26:28In reality, a parent has little, or even
26:33less, influence on their child at this
26:37age.
26:39Because the child, of course, is
26:40beginning to understand everything on
26:41their own, and the parent is already
26:42outdated.
26:43But the particularly persistent can
26:46poison this well, right? Because, in
26:49theory, well, not all children are
26:52inclined to meticulous, constant,
26:55scrupulous work, yeah, right.
26:59If they force you to do this on top of
27:01homework, then there's probably a chance
27:03that this interest will simply be killed
27:05.
27:06Is that common? You
27:09>> should do what interests you, and, of
27:11course, it's also desirable to have some
27:14prospects.
27:15If we're talking about, uh, specific
27:18material goals now, like going to
27:20university or building a career, then
27:23maybe, in addition to what's really
27:26interesting, you can also do something
27:29that, yeah, with some kind of...
27:31>> What, yeah, something that has a real
27:33future, yeah.
27:33>> And there are many of these areas, so
27:36focusing on just one is the reason it's
27:38all the hype right now. Firstly, it's
27:41all the hype now, but in five years it
27:42might not be.
27:44Take competitive programming, for
27:45example, which we're talking about now.
27:47Yeah, right. Five years ago, no one
27:49needed to sell anything, so to speak,
27:51everyone understood perfectly well that
27:54they needed to get into programming.
27:56You'd be an IT specialist, a programmer
27:58, and then you'd go to Yandex, and maybe
28:00even somewhere else.
28:01And then you'd be
28:03>> Yandex? Well, if they don't hire you,
28:05then you'll go
28:07,
28:07>> and, um, everything will be fine.
28:11Five years have passed.
28:13Just a week ago, I met at a respected
28:15physics and mathematics lyceum in St.
28:17Petersburg, including with applicants
28:19and their parents, and people like, what
28:21do you think about programming, Andrei
28:23Sergeyevich, are there any prospects?
28:27Will anyone even need it? I can't
28:30deceive people anymore.
28:32I don't know, just like no one knows
28:34what will happen to these guys.
28:36He
28:36>> didn't answer, right?
28:38>> So I replied that I don't know, but
28:40developed brains, the ability to think
28:43and act in unconventional situations,
28:45will always be valued, with or without.
28:50And, of course, we can't predict the
28:51button-pressing routine right now, but
28:53it seems certain others will be needed.
29:01>> Let's finish this sports analogy.
29:06Sports is usually a field that people
29:09leave, except for those very few who go
29:12into coaching, somewhere around the age
29:15of thirty, yes, in practically all
29:18sports.
29:21When do your athletes finish their
29:23careers?
29:24>> Well, it seems to me that the majority,
29:26well, again, if we're only grading
29:28schoolchildren,
29:29>> yep. When they finish school.
29:30>> Well, yes, it's logical when they finish
29:32school, because there's no other
29:34>> time?
29:35>> Because not all of them go on to play in
29:37college.
29:40If someone wants to continue, well,
29:42someone wants to continue competing in
29:44the Olympics, then, naturally, they go
29:46and write rounds and train for and
29:48participate in
29:50>> Rounds for...Oh, my God. Okay, I get it
29:53.
29:55>> There. And in that case,
29:56>> there is.
29:57>> Infinity.
29:59>> And in that case, it seems like their
30:02career ends either after five
30:04semi-finals or two finals, as it were.
30:09We simply have restrictions on
30:10participation, yes. You
30:12>> can't participate once, you
30:13>> can't participate in the final more than
30:15twice, and you can't participate in the
30:16semi-finals more than five times, or in
30:18the Northern Eurasia final.
30:19>> And that used to be the case.
30:22Fifteen or thirteen years ago, the best
30:25people traveled the world and collected
30:27, albeit small, prizes from all the big
30:29companies.
30:32Basically, you could travel around and
30:34collect some money.
30:36And what's more, all the best people,
30:38they really loved those photos with the
30:40huge checks they hand out there.
30:43And that's supposedly a nice supplement
30:45to your pension; you could earn, like,
30:47several thousand or even tens of
30:49thousands.
30:51>> The story remains the same, because,
30:53well, that's why our community is so
30:55dynamic, and so on and so forth.
30:59And so it's believed that basic
31:00knowledge and so on are best imparted to
31:03you by mature, established teachers
31:04who've read the topic a billion times,
31:06presented it a billion times in
31:08different situations, and understand how
31:10best to explain it so you understand it
31:12.
31:14But specific tricks, tips, new, I don't
31:16know, approaches to problem solving, and
31:18so on are imparted to you by current
31:20students who are still learning
31:22themselves.
31:24Well, if you look at the attitude of
31:25schoolchildren towards students. Because
31:27there, if people constantly write rounds
31:29on Codeforces or participate in other
31:31competitions and train in the same way,
31:33they have a greater understanding of the
31:35world. It seems like a career is ending
31:38, an active career, probably, mainly
31:40during student years, but at the same
31:43time, we know a billion twenty-twenty
31:46years, yes.
31:47Well, up to twenty-three, let's say.
31:50But at the same time, the
31:51>> oldest person on Codeforces is 70+.
31:54He plays something.
31:56>> We also know other situations, for
31:59example, where adult competitive
32:01programmers—well, since I'm throwing
32:03around names, there's Lesha Gusakov, our
32:08>> CTO. The
32:09>> CTO of search and artificial
32:11intelligence.
32:12He's still playing, by the way. And he
32:14writes the algorithm qualification every
32:16year.
32:17And last year, he participated in the
32:19Moscow ICPC quarterfinals with the
32:21Yandexoid team and went to the
32:23semifinals.
32:25So. My point is, even as an adult, the
32:28CTO of a major business group like
32:31Yandex, a family man, a dad, a husband,
32:34he still finds time to write contests
32:37and
32:38>> weightlifting.
32:42As someone once wrote, it's impossible
32:44not to notice that a networker who can
32:47bench press 195 is worthy of special
32:49mention, sir.
32:51>> But he still continues writing contests
32:53; it drives him incredibly. So, after a
32:55hard day at work, he can find five hours
32:57to write a workout with his teammates
33:00before midnight, and then still have to
33:02go home or go to the gym, or do
33:03something else.
33:04>> No, listen, there are such things,
33:07there's no escape.
33:09At the same time, I think you're both
33:12telling me that this discipline is
33:14actually somewhat isolated from real
33:17work, yes, a bit speculative.
33:22Do you acquire any skills that help you
33:25advance in your profession, or not?
33:30>> On the one hand, many people, especially
33:33before joining, were primarily focused
33:36on technological knowledge—
33:38technological, as they say in IT, stacks
33:41, the ability to use tools, connect
33:44libraries, properly build an application
33:47architecture, and debug.
33:51Well, in general, the work is far from
33:53solving mathematical puzzles, like...
33:55Well, that's
33:56>> the kind of technical stuff that
33:59requires ingenuity, and for
34:01>> some, it's a great help that, having
34:04simply solved contests in school or
34:06later at university, they developed this
34:09kind of cataloguing thinking, the
34:11ability to construct certain components
34:14in an unconventional way, to apply
34:17something, which they then figured out
34:19how to apply in an unconventional way in
34:22IT and there they did something
34:25unexpected, which then led to a sharp
34:27leap forward.
34:28>> Why am I asking this question?
34:31Because
34:34this is, of course, a media issue, but
34:37it seems like at one point the Durovs—
34:40well, especially the younger one—talked
34:43a lot about how it was the Olympiad
34:45participants who created VK, then
34:48Telegram, yes; that small teams could
34:50pull off such things, and other things.
34:55And they always paid a lot of attention
34:58, tried, maybe publicly, maybe actually
35:00to support them.
35:04So, they had this aura of being an
35:06Olympiad company.
35:10>> Listen, they started many companies when
35:12they were small, startups, or at the
35:14initial stages of development.
35:18Every engineer, so to speak, every
35:20developer was super valuable.
35:23They couldn't hire hundreds and
35:24thousands of employees
35:28to experiment with various products,
35:30like Google did back in the 2010s. We'll
35:33make a product, we'll see, no, it's some
35:35kind of crap, we'll shut it down.
35:37>> Oh, yeah, there was a whole series of
35:38>> stories.
35:40And after that, the question becomes
35:41whether the team, the company, can set
35:43up teams like that or not. Yandex, for
35:45example, couldn't do it for a long time
35:48, from my perspective,
35:50>> um,
35:51>> and then they learned, and now they can
35:53.
35:54Some companies have recently taken off,
35:57mostly related to AI.
36:00They have a bunch of Olympiad winners
36:02there too.
36:07>> Well, come on, I think we've reached the
36:09point where this scary beast of AI is
36:12coming onto the scene.
36:14How do you see
36:19AI as someone who's worked with
36:21programming all their life?
36:23There's a view that this is raising the
36:25level of abstraction.
36:27We've
36:28>> always been told that you're solving
36:30your own mathematical puzzles
36:33.
36:34>> No need.
36:35>> With known solutions.
36:36>> um,
36:37>> no one needs all that.
36:40You need to know Java, Python,
36:42distributed systems libraries, backend,
36:45frontend—that's what's needed.
36:49What you do there, no one needs. And
36:51it's always been that way. In this sense
36:54, even the opposite, maybe there's a
36:56slight pullback, so now we can laugh at
36:58them too, just like we know these
37:00libraries of yours better than you, and
37:02why the hell do we need to know them.
37:05We know ab exist, and thank you.
37:08>> And maybe we'll soon forget.
37:09>> Yes. But we've trained our brains and
37:12are now supposed to rule, meaning AI is
37:14better than you, because you only knew
37:17libraries, and your libraries are no
37:19longer needed. But this is all a
37:21simplification, as in any such story.
37:24>> Well, it seems optimistic.
37:26>> Optimistic on the one hand, but on the
37:28other, there's the question of mass
37:30adoption. And no one knows yet.
37:33Everyone is looking at this with a bit
37:36of concern; some are restructuring their
37:38processes, some are moving forward, and
37:41there they're introducing more, as it's
37:43fashionably called, yes, AI-assisted
37:45coding, while others are doing less.
37:49Since no one knows, everyone is watching
37:50with trepidation, wondering what will
37:51happen.
37:53But this is an economic situation,
37:54economic laws are complex, how will this
37:56all be resolved?
37:58But there are already many stories where
38:00, roughly speaking, tokens ate up annual
38:03budgets in a month, and what to do next
38:05is unclear. How this System will reach
38:07equilibrium, where it will be, we don't
38:10know.
38:11In this sense, I say, I always give
38:12advice to everyone, especially
38:14applicants, invest in yourself, learn
38:16fundamental mathematics and the basics
38:18of programming to understand what's
38:20going on, because how can a neural
38:21network generate anything for you? If
38:23you don't understand, you will
38:25definitely be of no use. And if you do
38:27understand, it's unknown, but the
38:28chances are better.
38:30>> This is true.
38:32>> And now, regarding the transition to a
38:35programming language, English or Russian
38:37.
38:38>> Yes.
38:39>> Now we are witnessing a slightly curious
38:42thing.
38:43In ancient times, I think, even before
38:45the foolishness...
38:47>> I burst into tears. Okay, good.
38:49I should have come with the dinosaurs.
38:52>> Dinosaurs.
38:54When dinosaurs roamed, they programmed
38:55on punched cards, even before assembler
38:57, they punched out some
38:59>> switches.
39:01>> In switches, then in assembler, then in
39:03C ++, which compiles to assembler, which
39:05compiles to machine code, and now in
39:07Python, which...Well, it doesn't compile
39:10to C ++, but it also compiles some
39:12intermediate language, which compiles to
39:14assembler, which compiles to machine
39:16code.
39:18Now we're talking about programming in
39:20Russian, which, well, or in English,
39:22compiles to Python, which compiles to
39:24its internal representation, which
39:26compiles to assembler, which compiles to
39:28machine code.
39:30The question is, why this whole chain?
39:32Why not immediately create machine code
39:34from Russian? If it's some kind of
39:36startup or something, then I'll agree.
39:39There's a new programming language, but
39:41right now, this process isn't a
39:42programming language.
39:44And there are some differences here.
39:47It's not that we give commands in
39:49Russian, like assign this value to a
39:51variable, create this cycle, or create
39:53this function.
39:56We're setting the task directly in
39:57Russian at the
39:59>> Meta-Level.
40:02Yes.
40:03>> And the resulting code is in our
40:05existing programming language.
40:09And this is, in fact, well, as they say
40:12now, a new reality. Yes, in programming
40:15—a new situation, let's use a slightly
40:18less pompous word, a new situation where
40:25instead of wasting your time
40:28implementing your idea in code, you can
40:32spend your time
40:37looking at how your idea is implemented
40:39in code on demand, and, if necessary,
40:42adjust it somehow.
40:44It doesn't look like this person, who,
40:46strictly speaking, used to be
40:48programming longer, just had to type
40:50these lines. IDEs helped him, suggested
40:53which libraries were available, what was
40:55available. But now he immediately
40:56receives ready-made code, by the way, of
40:58different types, quality varies. The
41:00company is now focused on improving it.
41:03And then they interact with him in one
41:05way or another, sorting things out.
41:08But the thoughts that turned into this
41:11code still came from humans.
41:14And in this sense, it seems that
41:16although the profession will change,
41:18just as it did with other programming
41:20languages,
41:22a
41:27higher level of abstraction, a truly
41:29higher level of problem-solving, really
41:35comes to the fore.
41:37And technology, knowledge of certain
41:40libraries, the ability to use them, or
41:44>> even punctuation are no longer so
41:45important.
41:46>> What are the alternatives? Either a
41:48singularity we don't know what will
41:50happen, enslaving us to AI, or
41:51everything will collapse and everything
41:53will return to normal—which, frankly, is
41:56also hard to believe.
41:58Or the system will transform into some
42:00other stable state, in which you need to
42:02be prepared to occupy the highest
42:04possible niche.
42:06To do this, you need to invest in
42:07yourself.
42:14>> Let's go back to the optimistic scenario
42:17of a sports programmer or an Olympian.
42:22But they train themselves; they go to
42:24their own gym, right? And what does that
42:27gym look like? How long does he study?
42:29What does he do there? What kind of
42:31clubs are these? So, the regimen, the
42:32sports regimen, what does it look like?
42:36>> Generally speaking, if it's already
42:38clear that this is a child who's
42:40interested, engaged, and developing in
42:43competitive programming, then his...
42:45well, depending on his level, well, if
42:47we take the general population, then his
42:50week—it looks something like this: well
42:53, he has school, he has it, she has it—
42:55it doesn't matter.
42:58But besides school, there's also some
43:00kind of additional education.
43:03For example, we have...Yeah,
43:05>> right.
43:06>> And there's a project called Yandex
43:07Circle.
43:09We work in three areas there:
43:10competitive programming, artificial
43:12intelligence, and competitive
43:13mathematics.
43:15This additional education adds an
43:18estimated 20 hours to a typical
43:20schoolchild's weekly schedule, where you
43:23spend five hours in a club on Saturdays
43:25, then another five hours writing a
43:28thematic essay on the topic covered in
43:30that club, and another five hours
43:33writing a remote essay on the next
43:35difficult one...What's a
43:38>> tour?
43:39>> Well, a tour is essentially a contest.
43:41A contest is a set of problems.
43:44>> So. A
43:46>> thematic tour means it's a set of
43:48problems united by a single topic.
43:50>> So.
43:51>> For example, you were taught a segment
43:53tree, and you have a segment tree
43:55contest of some complexity.
43:58You were taught, God forbid, the fast
43:59Fourier transform, and you have a
44:01five-hour fast Fourier transform contest
44:03.
44:03>> Hours per week.
44:05>> Naturally, 20 hours is some kind of
44:08super-upper limit, which is probably
44:10followed by some more or less serious
44:12competitive programmers.
44:15But in general, a club is about solving
44:18problems.
44:19You don't have to spend all 5 hours on a
44:21topical contest, but rather on as much
44:23as you can.
44:25You can spend less than 5 hours on a
44:27remote round, but only as much as you
44:29can, or not even do it at all during
44:31that particular week.
44:33So, there's always some kind of
44:35optionality.
44:36>> Even if it's not all there, it's about 3
44:38hours a day, or more precisely, two
44:39five-hour days, and
44:42>> Andrey wants to add more. It's a
44:44>> serious workload.
44:45>> Well, of course, that's a maximalist
44:47assessment.
44:49I don't know, I think my opinion is that
44:51no one, even the top ones, should study
44:52that much.
44:55And with schoolchildren, with students,
44:58I really did have classes three times a
45:00week during the peak classes.
45:03But, admittedly, we didn't have any
45:05homework. I worked with schoolchildren
45:07from Lyceum 239 in St. Petersburg. They
45:10have enough homework there without me.
45:12There are psychological problems with
45:14homework, like at home. That's
45:16>> not even cheating, but simply forgetting
45:18.
45:18>> At home, there's little opportunity to
45:20solve problems, yes. And this is
45:21connected not only with the lack of time
45:23, but also, for many schools, with the
45:25workload at school.
45:27Ah, there are schools with less workload
45:30, where, in principle, children can
45:31study at home on their own if they find
45:34the energy.
45:36Now students also usually have two...
45:38Well, students are, after all, more
45:39mature people, they can stay later, so
45:41we still have a
45:43>> more adult life, plus dancing, meetings
45:44.
45:45>> I actually did contests with them.
45:48That is, we had two five-hour contests
45:50every week, except around the exam
45:52period, ah, where we still got some rest
45:54, well, so that we could still pass the
45:55exam.
45:57>> This is some kind of standard,
45:58>> has become a standard.
45:59>> Olympiad time?
46:01>> Five-hour Olympiads, are these
46:03>> Olympiad rounds of some kind significant
46:05lasting 5 hours? It
46:05>> lasts 5 hours, yes.
46:06>> Can you, uh, or can't you?
46:10Can you just go to the toilet?
46:11>> Yes, you can. You can go to the bathroom
46:14, you can even eat something; people
46:16bring their own food and drink to the
46:18tour. Five hours is a long time. It's
46:20>> not like you just take out containers
46:22and eat, just in case.
46:24>> Oh, well, during training, of course,
46:25people just eat whatever they want. At
46:26the competitions themselves, yes, there
46:28...
46:29>> Well, this is the perfect time, I think
46:31, to bring it back, because, of course,
46:33I found that video, uh, where the dean,
46:35this Stankevich, disqualified the
46:37cabbage.
46:39What kind of outrage is this? A man sits
46:41for five hours and decides why it's
46:43necessary...What kind of...This guy
46:46>> was acting preemptively back then. He
46:48disqualified a head of cabbage from the
46:49trial round.
46:50>> From the beginning. We will, of course,
46:52show that video where you are there, uh
46:54, it seems like the most decorated coach
46:57in the world is laughing his head off at
46:59that moment.
47:00>> Of course.
47:02>> But he disqualified a head of cabbage.
47:04>> Well, the word "disqualified" isn't
47:05quite correct.
47:07Of course, this was a bit of a humorous
47:09episode, and everyone involved,
47:11including the student, I think Petr
47:13Losev, who took part in it, understood
47:16that.
47:17We wouldn't have done that to some
47:19unknown student.
47:21Oh, and an unknown student would hardly
47:22have brought a head of cabbage.
47:24There's a certain boundary in the rules
47:27that says you're allowed to bring food
47:29on the tour,
47:31>> right?
47:31>> But it has to be, and then there's this
47:34rather informal description that it has
47:36to be in the original packaging.
47:38>> It has to be square and soft. It
47:40>> doesn't have to be square, but in the
47:42original packaging.
47:43It shouldn't have any strong odors or
47:45anything that would disturb others, and
47:48eating it shouldn't be noisy, so as not
47:50to disturb others.
47:52>> The student brought it?
47:53>> The student brought a head of cabbage,
47:56saying, "No, of course not."
47:58I don't know, maybe he was planning on
48:00eating it, but it never got to that
48:02point. He brought a head of cabbage and
48:04>> placed it next to the computer.
48:06>> Next to the computer, yes. We decided,
48:08so to speak, to study whether it was a
48:12>> head of cabbage, whether
48:13>> it was food according to our rules. We
48:15noted
48:16>> one point about a student being able to
48:18bring a talisman in addition to food.
48:20But that's
48:20>> not allowed, is it? While it's allowed
48:22at the official final round of some
48:24Olympiads, it's prohibited at the final
48:26round of the All-Russian Olympiad.
48:29>> Children's talismans are usually small
48:32or not-so-small soft toys that symbolize
48:34something to them.
48:36>> Sharks, hedgehogs.
48:38>> We had large hedgehogs, and some small
48:42whales.
48:44It varied, and
48:45>> a cabbage could be, like this, a plant
48:48of emotional support.
48:50>> We conducted, yes, a humorous
48:52examination together with my colleague
48:54Denis Pavlovich Karasev,
48:57>> and they suggested something.
48:59>> That it's
49:01>> not food because it doesn't meet the
49:03definition of "not in its original
49:05packaging" and is a talisman.
49:08And since it's a talisman, talismans are
49:10prohibited.
49:11>> They confiscated the cabbage and
49:12destroyed it?
49:13>> No, why destroy it? We simply
49:15confiscated it, and probably the
49:17management told
49:19>> me, and I did, and so, yes, they handed
49:21me the cabbage, and so, we have a
49:23tradition at the All-Russian School
49:25Olympiad of broadcasting the Olympiad.
49:29This is when parents, future students
49:31who will go to the All-Russian Olympiad
49:33or who recently attended the All-Russian
49:35Olympiad, and now students, watch the
49:37broadcast, watching how their
49:38schoolmates are doing.
49:41If they're parents, they monitor their
49:42children, watching how they write, how
49:44many points they score, and so on.
49:47So, during the broadcast of the tour, we
49:49had a studio, like we're sitting here
49:51now, and in this studio, we set up a
49:53special cabinet where we basically
49:54collected some of the children's
49:56talismans, and the centerpiece of it all
49:58was a head of cabbage.
50:01And given that the All-Russian
50:02Tournament lasts a week, meaning there
50:04are two rounds and a day off between
50:06them, and there were also lots of lamps
50:08heating the space and so on, the head of
50:10cabbage started leaking.
50:12It turns out
50:13>> we were right to disqualify it. It was
50:16perishable.
50:17>> As it turns out, yes, within certain
50:19limits.
50:20So, well, during the
50:22>> first round, it heroically survived the
50:24broadcast, and by the second round,
50:25because, well, the hype chain had
50:27already been launched, we bought a
50:29second head of cabbage so that...Yeah,
50:30well, we gave it to the presenters for
50:32the broadcast, because it doesn't really
50:34remove the smell for them.
50:37>> There is cabbage.
50:38>> Well, not exactly.
50:40The thing is, the story with the cabbage
50:42head caused a bit of a stir there.
50:45It was in April, roughly, when the new
50:47school year and the new Olympiad year
50:49began, and the first Olympiad was when
50:51the hype was renewed.
50:53It was the St. Petersburg team Olympiad
50:55for the All-Russian selection. And
50:57several teams had already brought
50:58cabbage heads.
50:59>> For the selection for what?
51:01>> Well, there's the
51:03>> ICPC—a student team Olympiad, and
51:05there's also a school team Olympiad,
51:06similar to the ICPC, and it's called the
51:08KOSHP.
51:10>> You have cabbage, you have Nutella, you
51:13have ARVI.
51:15>> Well, we have a lot of things, I think.
51:16It's all more or less local, because
51:18generations of children change, and, as
51:20a rule, it seems to me, it's the
51:21children who bring the most of these
51:23things.
51:25>> Listen, let's wrap this up somehow,
51:27because I'm not an Olympiad winner, I'm
51:29having a hard time sitting still.
51:32>> 5 hours.
51:33>> Uh, I definitely won't be able to sit
51:34still for 5 hours.
51:40Uh, it seems to me that for those who
51:42still believe that this is a way to
51:45develop these problem-solving skills, uh
51:48, whether they're stem-based or not, and
51:51who are perhaps considering this path
51:53for their child, it's important for them
51:56, uh, it seems to me, to somehow
51:59determine the limit when this workload
52:02is too much, when the child might, well
52:04, simply not cope, or lose interest.
52:08Is there some kind of thermometer? You
52:11look and it's too much for the child.
52:15And maybe someone else finds it for
52:16themselves.
52:17>> Well, first of all, I think that if a
52:19child simply doesn't want to go to a
52:21club, or doesn't want to solve problems
52:23, you can give them a little time, maybe
52:26, like, "You haven't figured it out yet
52:28, maybe you'll like it later." But more
52:30often than not—I don't know, but from
52:32our experience working with
52:34schoolchildren—it's almost immediately
52:36obvious that the student has gotten into
52:38the wrong club. We recently had an
52:40in-person audition for a club
52:43in St. Petersburg.
52:45Well, not just for the Yandex club,
52:46there are others.
52:49So we had a club, and I went there to
52:50see how it was going.
52:54And my colleagues were holding it, and
52:56there were schoolchildren sitting there
52:57programming something on the computer,
52:59including some pretty young ones, like
53:01fifth, sixth, seventh graders.
53:03They look especially funny when they're
53:04really focused, programming something
53:05from the sidelines.
53:07Ah, that looks interesting. And there
53:09was a schoolboy in the hallway, I was
53:11just walking into the classroom, he was
53:13on the phone, "Mom," he said, "No, I'm
53:15not going to deal with this nonsense,
53:17it's just, I'm not interested.
53:19" Well, then don't bother, what's the
53:21point really? It's a
53:22>> very simple signal. A
53:23>> simple signal, yes.
53:25But, ah, if you look at it from a
53:27slightly longer perspective, if you've
53:29been studying for a while, your results
53:31aren't improving, that is, you're not
53:34succeeding in any competitions, and you
53:36don't want to.
53:38>> So, is this more stressful for you than
53:40enjoyable, right?
53:41>> So, you have, uh, an important sign that
53:44a person is using neural networks when
53:46they're clearly prohibited. If, for
53:48example, instead of solving problems,
53:51they're fed to DeepSeek, it's absolutely
53:53useless.
53:54Or, I don't know, Alice.
53:56If a person tries to cheat to achieve
53:58some result, that's also a sign.
54:01So,
54:02>> to get their parents to leave them alone
54:03, which is also what's happening now.
54:07>> Whenever I meet with schoolchildren and
54:08talk to them about something, I say, "
54:09You do all your homework with DeepSeek,
54:11" and they say, "Not in all subjects."
54:14So, if it's a subject where DeepSeek is
54:16used for homework, then, well, it's
54:18pointless. DeepSeek won't do anything.
54:20Maybe it will learn to solve problems
54:21better, but that's not a given.
54:23>> Alice is better.
54:25>> A schoolchild definitely won't learn. So
54:26, if you don't want to go to class, you
54:28use neural networks to solve your
54:30homework. It's
54:33not worth it. A bad sign, yes. A
54:36>> bad sign.
54:37>> And if a schoolchild is ready to spend
54:40the rest of the night solving something
54:43, or not, "I won't go there today, I
54:45don't know where, I'll be writing a
54:47dissertation."
54:49Well, that might also be a bad sign for
54:51something they won't attend, but it
54:53means they've become interested in
54:55sports competitions—the term "
54:57competitive programming" is more
54:58commonly used for schoolchildren.
55:01>> Listen, here's another question.
55:04By the way, I think that programming,
55:07well, as one of the options, maybe in
55:10this world beyond, uh, that singularity
55:13wall we can't see, it could very well
55:16take on some kind of gym.
55:19You need to go to the gym, or else
55:20you'll fall apart, right?
55:23And if you're putting everything into
55:26Alice and Deep Six, then you probably
55:29also need to go to an intellectual gym,
55:33otherwise you'll be a real test.
55:37>> I agree with the following caveat.
55:40We don't have to look far now; we don't
55:43need any singularity or Alice.
55:45It's long been known that people who
55:47engage in cognitive activity—there are
55:49very extensive studies showing that
55:51people who engage in some kind of verbal
55:54activity suffer less from age-related
55:56diseases.
55:58From Alzheimer's or some other memory
55:59problems.
56:01And even down to such banal things as
56:04solving crosswords or something else or
56:07trying to learn languages in adulthood.
56:10The same will happen here. It's a
56:12different matter that competitive
56:13programming has a fairly high entry
56:15barrier; you need to know how to program
56:16.
56:17But at some point, when programming
56:20skills suddenly become widespread, it's
56:22unclear why programming would decline,
56:25but theoretically, it could.
56:28It's already been a form of intellectual
56:31leisure for a while. Take Codeforces for
56:33a certain number of people.
56:37For those who don't program at work, by
56:39the way, because now a digital slave
56:42programs instead.
56:44It could probably return to the context
56:47as an option. Yes, the hands remember.
56:50>> And as for physical education. There's
56:53an obvious myth.
56:56And it seems to me that it's not a myth
56:58, but a statistically observable
57:00deviation, that the top students in any
57:02discipline are those who don't cheat.
57:07Are many of your athletes real athletes
57:10, physically fit ones?
57:13Is this a stereotype I came up with and
57:16the general public, or is this reality?
57:19>> There are several components to this
57:22image.
57:24I think glasses are a historical legacy
57:27of reading in low light, and therefore
57:31...
57:32>> Well, glasses are fine now anyway.
57:34>> Well, glasses are all the same, but in
57:36the sense that, uh, small children don't
57:38have them to the same extent, I think,
57:41now. Plus, they invented lenses.
57:43Ah, and as for physical activity, well,
57:46there are 24 hours in a day, not 48. If
57:49you're interested in math, programming,
57:52or whatever, physics, or anything else,
57:55of course, less time is devoted to
57:58sports. Ah, and what's there to talk
58:00about, what's there to play ball with,
58:02they don't even know how to program.
58:04This doesn't even verbalize the idea.
58:08It may exist, it's just that I'm more
58:09interested in talking to a computer than
58:11to...
58:12>> or to a person who understands what...
58:14>> who understands it.
58:16In this sense, we also have, in addition
58:18to the club we talked about, summer
58:20schools, for example, and winter schools
58:22, there are training camps, and there's
58:24the Sirius center, which is simply a
58:27production line, there's a nonstop shift
58:29every month, there are about 500 people
58:31in science in total.
58:33And plus the internet. With the advent
58:35of the internet, this problem has become
58:37much less of a problem, because you can
58:39use specialized chats or forums, like
58:41the Yandex Circle chat, to chat with
58:43people just like you.
58:44>> Listen, I have one last question. But
58:46let's get back to Yeshechka.
58:48Do you think you can adapt the type of
58:52competition itself, your sport, to this
58:59new reality?
59:01>> Well, first of all, it's pointless to
59:03compete with AI. Even now, last year—
59:05that is, a year ago—for AI at this stage
59:07of development, a year is a complete
59:09abyss.
59:10A year ago, it already beat everyone
59:12there in the finals.
59:14And now it's already solving problems
59:16like those of the Finns. Well, there are
59:17nuances there, we won't go into them now
59:19. I think it will soon be the same even
59:21without them. So, competing with AI
59:23directly in this sense is pointless.
59:27But is it possible to solve problems
59:31with AI? There are attempts here now.
59:36They've been relatively unsuccessful so
59:38far, because it turns out that humans
59:39are the weak link in this case,
59:41specifically in the competitive sphere.
59:44Competitions are no different.
59:46There's a clearly defined problem, clear
59:47evaluation criteria, and a clear
59:52solution. And when a person
59:57gets into it, it turns out that for this
1:00:02kind of field, humans are better off,
1:00:06even at the solution-generation stage.
1:00:11There were competitions where it was
1:00:13allowed to use [systems], and the
1:00:14authors specifically tried to design
1:00:16problems so that people couldn't solve
1:00:18them on their own, but had to use
1:00:20systems.
1:00:21But it turned out that the authors
1:00:23simply, firstly, came up with the
1:00:24problem two months earlier than the
1:00:26competition; secondly, well, they didn't
1:00:28spend as much money on tokens as some
1:00:30participants ended up spending, and with
1:00:31a slightly more powerful model, they
1:00:33managed it without any special help,
1:00:35without a human.
1:00:37>> So, that is to say, there's no option
1:00:39yet. There aren't
1:00:40>> many options for such workers right now
1:00:42.
1:00:44Either we ban it everywhere, and people
1:00:46simply compete with other people, or
1:00:49competitions degenerate a bit. So far,
1:00:52so good. Do
1:00:53>> you think
1:00:56Olympiad mathematics has more promise?
1:01:00Because in what you're talking about, it
1:01:03seems like the programming part is
1:01:06already a bit anachronistic.
1:01:09>> But no, what's the problem? It's still
1:01:10the same in Olympiad mathematics.
1:01:13Mathematically, like some of the
1:01:15competition authors, there was hope that
1:01:17the programming part could be
1:01:19transferred, and the math part would
1:01:21remain.
1:01:23But as I just said, no, it didn't work
1:01:25out. He also solves the math part better
1:01:27than people. But he solves the Filto
1:01:30problem. And it's clear that
1:01:31incomparable resources were used there.
1:01:34And at Olympiads, the problems aren't
1:01:36world-class, but rather, millennia-long
1:01:38.
1:01:39>> I'm also afraid that not everyone will
1:01:40understand these problems.
1:01:42Tell me, in my childhood, for children,
1:01:44the reference point was, in particular,
1:01:47all sorts of little books that could
1:01:49draw them into the world of science.
1:01:52There's wonderful physics, wonderful
1:01:54mathematics, and well, something like
1:01:56that, yeah, there.
1:01:58Is there anything like that, like an
1:02:00entrance, like a litmus test, some kind
1:02:02of entrance?
1:02:05Or maybe, I don't know, maybe there's
1:02:07some kind of mathematical animated
1:02:09series? Well, so what? There's Rick and
1:02:11Morty.
1:02:12It's actually quite highly mathematical
1:02:15in some places.
1:02:17Something.
1:02:19So how can you test your fourth-or
1:02:21fifth-grader, whether he's into it or
1:02:22not, using a game method?
1:02:24>> Take him to some kind of math festival
1:02:26and see if he likes it or not.
1:02:29Of course. Where else?
1:02:31>> Well, to see what a math festival is
1:02:33like. A
1:02:34>> festival is the perfect entrance and
1:02:36test for your, well, child, whether he
1:02:38likes it or not, of course.
1:02:39>> There are a lot. That's precisely why
1:02:41there's less of that in programming, not
1:02:44much in programming, but in mathematics
1:02:46...Well, probably, maybe in fifth or
1:02:48fourth grade, it's really
1:02:51easier to test in mathematics. There's a
1:02:54huge collection of clubs and clubs from
1:02:56the Soviet era that have survived and
1:02:59developed further, and there are
1:03:01festivals, regattas, and festivals.
1:03:05Where children come, it's not five
1:03:07o'clock, but half an hour or an hour to
1:03:08solve problems.
1:03:10The problems are formulated so that a
1:03:12fifth-grader, and maybe even a
1:03:13first-grader, can understand them.
1:03:15>> Buy it for a child.
1:03:16>> What, what? There's a
1:03:18>> math magazine for adults called Quantum
1:03:20, and there's a magazine with math
1:03:22problems for children.
1:03:24It's called Quantum.
1:03:26Well, and see how the child reacts to
1:03:27what you give them, or where you take
1:03:29them, and whether they like it in
1:03:30general, whether they want to return or
1:03:32not.
1:03:34>> Well, I mean, we probably don't need
1:03:35this wonderful math book anymore, we
1:03:37shouldn't use Perelman, right?
1:03:39What's
1:03:40>> wrong with him? I don't know, I think it
1:03:42>> seems relevant, yes. This is exactly
1:03:45what
1:03:46>> I did when I was working with
1:03:47schoolchildren.
1:03:48I almost wanted to force them to read
1:03:50some of these books, because they felt
1:03:52like they had a problem—we knew all the
1:03:54algorithms, but we couldn't come up with
1:03:56a solution.
1:03:57I said, "Well, solve math problems if
1:03:59you can't come up with one."
1:04:01I gave them books to read, but of course
1:04:03, they didn't read them.
1:04:05Programming is more interesting than
1:04:07solving math problems, because there's
1:04:09something about programming olympiads
1:04:11that doesn't exist in math olympiads.
1:04:14It's from psychology theory—immediate
1:04:16reinforcement. In programming problems,
1:04:18you submit a problem and immediately get
1:04:19the result.
1:04:21The testing system tells you whether
1:04:22you're right, you're wrong, and you can
1:04:22even fix it.
1:04:27That's why programming olympiads are
1:04:28simply more interesting for the
1:04:28participants; they're more dynamic. The
1:04:29testing is done right away.
1:04:30The result is immediate.
1:04:32>> Well, then, let's submit our work to the
1:04:34audience for review, and see what they
1:04:36give us. Thank you.
1:04:38>> Thank you.
1:04:39>> Thank you.

About this transcript

This page contains the full transcript of По ту сторону медалей. Как устроено олимпиадное сообщество в России? (yet another podcast #56) by Яндекс, generated from the public captions YouTube serves with the video. The transcript has 10,513 words across 1,850 segments, with the original timestamps preserved so you can click any line to jump to that moment in the embedded player.

What you can do with it

Use the transcript to take notes, quote the speaker, build a study guide, generate a summary with ChatGPT or Claude via the YouTube Summary tool, or export it as a timed subtitle file with YouTube to SRT. You can also re-open it in the transcriber to choose a caption track available on the video.

Free YouTube transcript tool

YouTube2Text is a free YouTube transcript generator — no signup, no daily limit. Paste any YouTube link and get the full transcript instantly, with timestamps, click-to-jump, caption language selection, AI prompts for ChatGPT, Claude, and Gemini, and exports to TXT, SRT, VTT, or Markdown.