Блог пользователя meme

Автор meme, история, 4 года назад, По-английски

This is purely anecdotal, but I have observed that people who have started participating on Codeforces have made it to, say, expert in a shorter duration of time compared to me. I also feel that my rating has risen without considerable improvement in performance, and I remember being completely overwhelmed by the peformance of division 1 users back in my early days. Now, it seems that anyone can become a candidate master (or even master with some improvement).

With all these considerations, how real is rating inflation, to what extent, and how far back in time did it start affecting ratings considerably? Looking for responses backed by evidence. Thanks!

  • Проголосовать: нравится
  • +60
  • Проголосовать: не нравится

»
4 года назад, # |
  Проголосовать: нравится +26 Проголосовать: не нравится

evidence:

a while ago nobody was over 3000

now there are a ton of people over 3000

  • »
    »
    4 года назад, # ^ |
      Проголосовать: нравится 0 Проголосовать: не нравится

    Okay, but that can be explained by increased participation. I'd like to know if it has become easier to gain rating at the same level of proficiency.

»
4 года назад, # |
Rev. 2   Проголосовать: нравится 0 Проголосовать: не нравится

yes may be because the participants have increased dranmatically in past few months so now there are more beginners in this site than ever,,,!! so that could be a reason

»
4 года назад, # |
  Проголосовать: нравится +156 Проголосовать: не нравится

How can rating inflation be real if contests aren't real?

»
4 года назад, # |
  Проголосовать: нравится 0 Проголосовать: не нравится

yes it is real

prove : now even div2C is at beginner level

  • »
    »
    4 года назад, # ^ |
      Проголосовать: нравится +34 Проголосовать: не нравится

    He asked about rating inflation, not about the difficulty of the problems...

    Yea, it is real, but it's good. People who don't participate are being slowly overtaken by regular participants.

    • »
      »
      »
      4 года назад, # ^ |
      Rev. 2   Проголосовать: нравится +3 Проголосовать: не нравится

      People who don't participate are being slowly overtaken by regular participants.

      That is actually what motivated me to post this question in the first place. If the current rating system is so volatile with changing demographics (the percentage of participants who are relatively new to CP, etc.) and a larger userbase, is it possible to tweak it so as to stand the test of time, instead of being favourable to more recent participants?

      I understand that this is a difficult task, and I'm sure that people have tried to come up with such schemes before, but I was hoping to spark an interesting discussion with respect to the rating system.

  • »
    »
    4 года назад, # ^ |
    Rev. 2   Проголосовать: нравится 0 Проголосовать: не нравится

    I don't agree that Div 2C has gotten easier. When I pull out a random C from an old contest, I find them at the same level of difficulty as any of the Div 2 contests of late.

    Also, I checked meme's graph and saw that he became an Expert in just two months.

    • »
      »
      »
      4 года назад, # ^ |
        Проголосовать: нравится +8 Проголосовать: не нравится

      check

      nowadays people complaining about solved only 2 problems or hard round

      believe it or not in my opinion codeforces definitely decrease problems difficulty

      • »
        »
        »
        »
        4 года назад, # ^ |
          Проголосовать: нравится 0 Проголосовать: не нравится

        I think you are under estimating the difficulty because you got stronger over time.

        Can you please provide me a range of contests where you felt the C was harder ? For example, from Contest number 300-350.

        I will check out the problems from that time frame.

      • »
        »
        »
        »
        4 года назад, # ^ |
          Проголосовать: нравится +8 Проголосовать: не нравится

        Older round generally had 5 problems but current rounds usually contains 6 problems.

»
4 года назад, # |
  Проголосовать: нравится +11 Проголосовать: не нравится

Well I do believe that it is easier to gain rating nowadays. But I'm not sure what's the reason behind that. Maybe I'm just getting better than before.

»
4 года назад, # |
  Проголосовать: нравится +98 Проголосовать: не нравится

Here's some data from when the current rating system was established.

Now here's a comparison.

I'm not very experienced in statistics, but looking at it, it seems like there is no serious inflation except at the very highest level. For example, the ratio between experts now and then is pretty much exactly the same as the ratio between grandmasters now and then. I don't feel that the "meaning" of candidate master or international grandmaster has changed much.

The introduction of Div 3 rounds and the changed Div 1 / Div 2 border probably also muddy the waters here.

I notice that OP only joined a few months ago. It's actually a fairly typical thing: new members of a lot of communities perceive very big changes during their first few months. But actually, they just notice the difference between their first impression and reality.

I also feel that my rating has risen without considerable improvement in performance, and I remember being completely overwhelmed by the peformance of division 1 users back in my early days.

Go back to the rounds of your early days. Do they seem as hard as they seemed back then? Sometimes you simply can't see how much you've improved. I remember a contest (Metropolises 2016 day 1) where I could not solve the last two problems of the contest. Half a year later, I looked back and although I didn't "feel" improvement, I could solve the last two problems immediately. So just because you feel you haven't improved doesn't mean you actually haven't.

»
4 года назад, # |
Rev. 2   Проголосовать: нравится 0 Проголосовать: не нравится

Yes. Thanks to coronavirus effect and inflation I gained a fuckton of ratings from crappy performance.