Your network blocks the Lichess assets!

lichess.org
Donate

Science of Chess: Let your Brilliance be your guide

Blessings, much continued success in your research!

Blessings, much continued success in your research!

@DIAChessClubStudies said ^

Blessings, much continued success in your research!

Thank you!

@DIAChessClubStudies said [^](/forum/redirect/post/8d1mrn82) > Blessings, much continued success in your research! Thank you!

@NDpatzer

I wonder why the highest percentage of wins led to decreasing gains after a time. Maybe cos burnout.

Analyzing losses is psychological demoralizing and the people who say 'only analyze losses' never say how.

They say find the mistake but they never say how. And they never say how to improve from knowing a mistake.

Analyzing win is psychologically good cos it make u happy.

@NDpatzer I wonder why the highest percentage of wins led to decreasing gains after a time. Maybe cos burnout. Analyzing losses is psychological demoralizing and the people who say 'only analyze losses' never say how. They say find the mistake but they never say how. And they never say how to improve from knowing a mistake. Analyzing win is psychologically good cos it make u happy.

"Lichess players between 1600-1800 ELO."
"one month of Blitz play (3-5 minute time controls during May 2023)"

  • This sounds questionable. A blitz game is not worth to analyze - Nezhmetdinov.
    What does analyze mean? Quickly glance at? Request engine analysis and do learn from your mistakes?
    1600-1800 3-5 minute blitz is most about blunder checking, not analysis.
    How were the respondents selected?
    The win rate higher than 50% indicates a biased selection, a balanced sample should have 50% win rate.
    Progress is most difficult at higher ratings.
    More relevant would be 2000+ and rapid or classical, with some additional question to define what 'analyze' means.
"Lichess players between 1600-1800 ELO." "one month of Blitz play (3-5 minute time controls during May 2023)" * This sounds questionable. A blitz game is not worth to analyze - Nezhmetdinov. What does analyze mean? Quickly glance at? Request engine analysis and do learn from your mistakes? 1600-1800 3-5 minute blitz is most about blunder checking, not analysis. How were the respondents selected? The win rate higher than 50% indicates a biased selection, a balanced sample should have 50% win rate. Progress is most difficult at higher ratings. More relevant would be 2000+ and rapid or classical, with some additional question to define what 'analyze' means.

@NDpatzer

I read the paper you linked.

The paper defined 'analysis' as games which were analyzed by the computer on Lichess. (People clicking analyze your game button).

Computer analysis does not mean that the game was actively analyzed by the player.

Studying losses cannot be equated with getting a computer analysis after the game. It leaves out self-study without the engine.

Clicking the computer analysis button is a long way from 'studying losses'!

@NDpatzer I read the paper you linked. The paper defined 'analysis' as games which were analyzed by the computer on Lichess. **(People clicking analyze your game button).** Computer analysis does not mean that the game was actively analyzed by the player. Studying losses cannot be equated with getting a computer analysis after the game. It leaves out self-study without the engine. Clicking the computer analysis button is a long way from 'studying losses'!

"games which were analyzed by the computer"

  • And how to know if it is the winner or the loser who requested the computer analysis?
"games which were analyzed by the computer" * And how to know if it is the winner or the loser who requested the computer analysis?

@tpr said ^

"Lichess players between 1600-1800 ELO."
"one month of Blitz play (3-5 minute time controls during May 2023)"

  • This sounds questionable. A blitz game is not worth to analyze - Nezhmetdinov.

Your mileage may vary, but I think a lot of players find it worthwhile to go back and look at Blitz games. I do agree that it would be interesting to take a look at a regression like this with longer time controls, but I don't think it's inherently a problem to limit yourself to Blitz play here.

What does analyze mean? Quickly glance at? Request engine analysis and do learn from your mistakes?
1600-1800 3-5 minute blitz is most about blunder checking, not analysis.
How were the respondents selected?

You can find this in the paper (link in the blog post) but this was a random sample.

The win rate higher than 50% indicates a biased selection, a balanced sample should have 50% win rate.

Again, they randomly sampled players from the target time period. Also the win rate I see here is slightly under 50% (they report ~47.5%) which doesn't strike me as a worrisome fluctuation around 50.

Progress is most difficult at higher ratings.
More relevant would be 2000+ and rapid or classical, with some additional question to define what 'analyze' means.

Ah, but more relevant to whom? I'm a player in this range and I play a lot of Blitz, so this is quite relevant to me!

@tpr said [^](/forum/redirect/post/zM0uy2HW) > "Lichess players between 1600-1800 ELO." > "one month of Blitz play (3-5 minute time controls during May 2023)" > * This sounds questionable. A blitz game is not worth to analyze - Nezhmetdinov. Your mileage may vary, but I think a lot of players find it worthwhile to go back and look at Blitz games. I do agree that it would be interesting to take a look at a regression like this with longer time controls, but I don't think it's inherently a problem to limit yourself to Blitz play here. > What does analyze mean? Quickly glance at? Request engine analysis and do learn from your mistakes? > 1600-1800 3-5 minute blitz is most about blunder checking, not analysis. > How were the respondents selected? You can find this in the paper (link in the blog post) but this was a random sample. > The win rate higher than 50% indicates a biased selection, a balanced sample should have 50% win rate. Again, they randomly sampled players from the target time period. Also the win rate I see here is slightly under 50% (they report ~47.5%) which doesn't strike me as a worrisome fluctuation around 50. > Progress is most difficult at higher ratings. > More relevant would be 2000+ and rapid or classical, with some additional question to define what 'analyze' means. Ah, but more relevant to whom? I'm a player in this range and I play a lot of Blitz, so this is quite relevant to me!

@RuyLopez1000 said ^

@NDpatzer

I read the paper you linked.

The paper defined 'analysis' as games which were analyzed by the computer on Lichess. (People clicking analyze your game button).

Computer analysis does not mean that the game was actively analyzed by the player.

Studying losses cannot be equated with getting a computer analysis after the game. It leaves out self-study without the engine.

Clicking the computer analysis button is a long way from 'studying losses'!

Sure, that's fair and it is a limitation of what they can do with the Lichess API. The advantage of getting this kind of data is leveraging the size of the database, but the downside is that we don't know how thoroughly they studied the game. Even with a much shallower review of the game, however, there is still a relationship to explain in the data that favors getting an analysis of wins vs. losses - It would be neat to see a controlled experiment that enforced richer review, but this would also be very hard to execute.

@RuyLopez1000 said [^](/forum/redirect/post/xgqc3f2a) > @NDpatzer > > I read the paper you linked. > > The paper defined 'analysis' as games which were analyzed by the computer on Lichess. **(People clicking analyze your game button).** > > Computer analysis does not mean that the game was actively analyzed by the player. > > Studying losses cannot be equated with getting a computer analysis after the game. It leaves out self-study without the engine. > > Clicking the computer analysis button is a long way from 'studying losses'! Sure, that's fair and it is a limitation of what they can do with the Lichess API. The advantage of getting this kind of data is leveraging the size of the database, but the downside is that we don't know how thoroughly they studied the game. Even with a much shallower review of the game, however, there is still a relationship to explain in the data that favors getting an analysis of wins vs. losses - It would be neat to see a controlled experiment that enforced richer review, but this would also be very hard to execute.