1

I just defeated a 1601 rated computer in under 3 minutes. Normally, it is a lot more difficult to defeat players having rating even 1500+.

I want to understand how is that possible. Was it a computer that was wrongly rated at 1600+ or did I really outperform it?

[FEN "7k/8/6p1/8/7P/8/3r4/4q2K w - - 0 1"]

It never looked like a computer playing at 1600! What really could have happened?

enter image description here

2
  • 1
    1600 computer is comparatively lot easier to beat than 2000+. But sometimes they play very poorly and sometimes even stronger than the 2000 level bots. It's based on the algorithm to make blunders.
    – m4n0
    May 19 '20 at 18:12
  • 1
    @ManojKumar This should be the answer.
    – Annatar
    May 20 '20 at 10:14
2

A sample of one game should not be enough to make a conclusion. You may have been particularly inspired, or maybe got into a position where that computer player worse than others.

1
  • yes, it actually happened that computer was playing good tactics at start - but, kind of slipped all of a sudden to lose the game May 20 '20 at 7:07
3

I want to understand, how is that possible - was it a computer that was wrongly rated at 1600+? or did I just outperform it really?

Neither.

You don't understand rating systems. There is no absolute rating system in which to say the computer was or wasn't playing at 1600 strength. All rating systems are relative. They measure the relative performance of the players results in that system. There is no judgement of the quality of the moves.

If the computer rating system is accurate then if you set the computer program to play against the same computer program (with you feeding in the alternate moves) on another computer and one is set to play like 1600 and the other like 1700 then over a large number of games the 1700 computer should score a certain percentage against the 1600 computer.

1
  • agreed, but - if that computer had reached 1600+ rating - it could be very well expected to play at certain level. so, even though ratings are relative - the game the computer played was nowhere near to what is expected to be played at 1600 chess.com level May 20 '20 at 7:08
3

Several reasons I can think of:

  • A single game is not enough to draw any conclusions. You might have been lucky, inspired,... in that particular case. Ratings tell about the statistical outcome of a large number of games.
  • Ratings measure the relative strength of players. There is no such thing as a player of 1600 strength. Matching ratings from different websites or OTB chess is inherently impossible because it involves different pools of people (or computers). You can assign ratings to different levels of an engine by having it play against itself many times. However matching this computer rating to a human rating is basically guessing with a huge error margin. Some engines just count the levels (1,2,3...) instead of claiming a fake rating.
  • Making a computer as strong as possible is a well defined procedure (just optimize the parameters, so that it wins most games). Crippling it, so that it plays at a lower level is not well defined. You could for instance have it ignore king safety or ignore material or make a blunder every now and then (but play at maximum strength otherwise), etc.... One way of crippling might suit you more than another player. Say, if you are an attacking player, a computer playing very passively or that ignores king safety would be easier to beat for you than one that is crippled in another way.

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.