"You all can check the leaderboard—even Legendary Grandmasters are unable to solve Problem F, and Grandmasters and Masters are struggling with E1 and E2. Why waste such hard-made problems on a Division 2 contest instead of saving them for Division 1? What's the point of putting questions that are unsolvable by Division 2 participants? Contest difficulty should increase linearly, but it often spikes abruptly. What do you all think about this? Feel free to criticize, correct, and share your opinion."
→ Обратите внимание
→ Трансляции
→ Лидеры (рейтинг)
| № | Пользователь | Рейтинг |
|---|---|---|
| 1 | tourist | 4009 |
| 2 | jiangly | 3821 |
| 3 | Benq | 3736 |
| 4 | Radewoosh | 3631 |
| 5 | jqdai0815 | 3620 |
| 6 | orzdevinwang | 3529 |
| 7 | ecnerwala | 3446 |
| 8 | Um_nik | 3396 |
| 9 | ksun48 | 3388 |
| 10 | gamegame | 3386 |
| Страны | Города | Организации | Всё → |
→ Лидеры (вклад)
| № | Пользователь | Вклад |
|---|---|---|
| 1 | cry | 164 |
| 1 | maomao90 | 164 |
| 3 | Um_nik | 163 |
| 4 | atcoder_official | 161 |
| 5 | -is-this-fft- | 158 |
| 6 | awoo | 157 |
| 7 | adamant | 156 |
| 8 | TheScrasse | 154 |
| 8 | nor | 154 |
| 10 | Dominater069 | 153 |
→ Найти пользователя
→ Прямой эфир
Codeforces (c) Copyright 2010-2024 Михаил Мирзаянов
Соревнования по программированию 2.0
Время на сервере: 10.11.2024 09:16:00 (h1).
Десктопная версия, переключиться на мобильную.
При поддержке
In top 500 only two F solutions, rainboy and saptarshi8103... In top 500 only one E2 solution, rainboy...
edit: of course, these results are from only official participants
Giving Hard problem to Upsolve
I think the actual issue is not problem difficulty, but that contests are too short for anyone to try E/F unless they have GM tier speed. But I agree it's silly to add them if no one reaches that point, I would prefer 3 hour contests personally.
I would prefer 20 minute contest, because I would get 2200 perf every time...
There aren't a lot of Div 1, so I guess putting challenging problems in Div 2 keeps Div 1 users busy. Also, it happens that difficulty isn't well estimated by authors, but we can't really blame them for that.
To be honest, it might be better to create like 2-3 problems that would cover most ranks of Div. 2 and save these super hard problems to just create a Div. 1 and Div. 2 at the same time. Usually the problems overlap, so it isn't like creating 2 whole different contests. Maybe just a couple problems in the 1400-2000 range might make it less of "speed-forces" and more "skill-forces"
That's true, but I think part of the reason for this is because these problems that fit better for Div. 1 are used for Div. 2-only rounds. If we save them for a few rounds we may be able to have more Div. 1s regularly (although I'm not too sure if we can prepare enough 1F problems for LGMs).
There are much fewer Div. 1 participants who participate in Div. 2 unofficially just because it's unrated, compared to Div. 1 rounds. It's about motivation, and having these problems in Div. 2-only highly demotivates many Div. 1 people. Also, I think it should be natural for Div. 1 participants to think that Div. 2 problems may not be challenging for them (so that they skip these), and it's rather abnormal that they are actually still hard.
This is also true, but if we're seeing the same tendency for months now, maybe it's time to convince ourselves that these problems are likely to be much harder than we initially thought. This year I've almost never seen a Div. 2 round's hardest problem being too easy, and it's almost always the opposite.
That's true, but I think part of the reason for this is because these problems that fit better for Div. 1 are used for Div. 2-only rounds.
Maybe, but as you said, the 1F problems seem to be a bottleneck, since it's quite hard to set problems way above your level. And setters don't seem to lack hard Div 2 problems, since I've seen hard problems quite often recently.
There are much fewer Div. 1 participants who participate in Div. 2 unofficially just because it's unrated, compared to Div. 1 rounds.
Yes, but I also thought about upsolving. I've upsolved most of problems which were in my reach in recent Div 1 rounds, so for a motivated Div 1 user, Div 1 problems aren't enough. Having challenging problems in Div 2, even if I don't take part, is pretty neat. (I also train on other platforms ofc, but still, I appreciate the design of Codeforces so it's one of my favorites)
A lot of the div2 last problems wouldn't and shouldn't be able to appear in div1 though. So its not always like its hard enough for div1 => its fine for div1 too.
In div2, its like "This is too hard to matter whether its a nice problem or not, no real participant is getting separated by this problem" but in div1, you do need to care
Watching Grandmasters struggle with these problems is like watching superheroes lose their powers. Maybe the problem setters were aiming for a real-life comic book plot twist! **
Personal experiences: Sometimes we as setters have no choice. It's not like we have a bank of always-available problems in all levels.
Maybe they didn't realize their problems were that hard. Maybe they had to fill them in to cover some problems they lost in time for deadline. There are various situations for such to come.
In some contests I see a 2500-2700 problem as E, then a 3000+ problem as F, so I think the problem F is sort of wasted, as the contest would've still had a reasonable difficulty curve with 5 problems (for example 949 and 945). Maybe testers were not honest with their opinions on difficulty :(
"Testers were not honest with their opinions on difficulty" is neither a correct nor reasonable claim, honestly. Testers, like everyone, have biases, and while their opinions did contribute to rounds, we should be reminded that statistically they don't represent everyone in Codeforces (especially so when being a good tester requires some specific attributes even).
Also, it's not like testers and setters would all agree on things. For example at Codeforces Round 963 (Div. 2), should I test earlier, I would tell the setters that E is only a *2400-rated at 2250 points (instead of *2700 at 2750 as we're seeing) — and I would still do so even right now knowing that the participants don't share that sentiment. That's just to show testers do have their personal biases independent from each other and from problemsetters' viewpoints as well.
Something to note is rhat div2 ratings especially for higher problems is completely irrelevant. The same problem if put in a div1, could be 400 rating lower or even more. Last contest E2 for ezample deserves atmost 2600-2700 but its predicted to be 2900-3000.
So, its not that testers are not honest with difficulty. Rather testers who give feedback on harder problems are div1 people and they give from their perspective (which is familiarity with div1 ratings and not div2 ratings), and not the fake rating it gets.
The problem rating system is shit.
one good reason is that if now Div.2 only have problems rated up to 2200 then we cannot differentiate CMs from GMs
But your analogy left a huge gap in-between, right? I think a ~2500-2700 rating problem for Div2F is good enough as such a differentiation (and recent "overly hard" problems as stated tend to offshoot that estimation quite far).
Isn't this what Div 1 is for though?
For me there is a problem, that I can create problems with difficulty up to ~2800, but not higher, because I can't understand, how those problems have to look like. I think, problemsetters put ~2800 problems to div 2 contests, because they simply can't create last problems for div 1 contests. One way to solve it is to create div 1.5 contests, which would be rated for people up to ~2600.