Блог пользователя VLamarca

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

Hi!

In an attempt to verify by myself if rating inflation is real, I will be participating virtually in this contest from 2015 in roughly 15 minutes (21h05 UTC-3). I invite you to do the same!

I will do this for the next few weeks, and maybe gather some statistics in the end.

UPD: Among 3 participants that I consider very trustworthy, the rating change was very small. (Around +10 for each one). The rating variation was estimated by looking at people with similar original rating that got similar ranking in the contest as well. So this very small sample indicates that rating inflation is negligible.

  • Проголосовать: нравится
  • +65
  • Проголосовать: не нравится

»
4 года назад, # |
  Проголосовать: нравится +38 Проголосовать: не нравится

What is your hypothesis?

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

    I would expect a very small variation in rating, so I definitely disagree With people that believe That nowadays it is much easier to gain rating.

    • »
      »
      »
      4 года назад, # ^ |
        Проголосовать: нравится +67 Проголосовать: не нравится

      I actually think the opposite. I have found that old problems tend to be easier / more typical than recent problems with same difficulty rating.

      • »
        »
        »
        »
        4 года назад, # ^ |
          Проголосовать: нравится +5 Проголосовать: не нравится

        Not to mention the crazy amount of people participating and ranks drop in matter of 2-3 mins if you don't get A-B done in time.

        Certainly doesn't concern blues and above but to those are below I think it has been causing more rating drops than raise.

        Only those are above blue are probably feeling some kind of inflation

        My personal views from my recent experiences in CF contests. They may not be accurate.

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

          I don't think blues feel inflation. Inflation mostly exists at upper bounds of division: deep cyan->blue in div3, deep purple->yellow in div2. Most notable inflation in my opinion has been in deep purple->yellow as number of continuous div2 contests without any div1 contest in between is too much. When 4 consecutive div2 contests occur, first 2 send most fallen yellows back to yellow and last 2 create some inflated yellow who wouldn't be yellow if they were competing in div1 against yellows.

»
4 года назад, # |
  Проголосовать: нравится +16 Проголосовать: не нравится

I tried this contest and got ~2350-2400 performance(based on official participants with roughly the same place and their rating changes). I also don't feel like there is an inflation. Moreover, I don't understand why C and D are 2000 and 2200 respectively, they are slightly easier, in fact.

I'm quite surprised that there were no observation/constructive/formula/math problems, unfortunately, these days Codeforces has a lot of them

»
4 года назад, # |
  Проголосовать: нравится +37 Проголосовать: не нравится

https://cfviz.netlify.app/virtual-rating-change.html

You can use this to estimate correct delta.