celot (#28),
I do agree with those who say that average players strength in FIDE pool is greater than here, and over the internet in general. Because it takes you a bigger effort to play OTB than it does to play over internet, the noncompetitive players who play just for fun are more likely to play over the internet only. For instance, my 7 year old daughter has an account in lichess, but she would not yet play a rated OTB game elsewhere. Yet, I don't think this is a big factor. The pool with more competitive players tends to have an slightly higher average strength, but not much.
I also agree that lichess ratings are accurate, due to Gliko2 system and a more humanized environment. That's why rating chart shows regular deviation.
The best example of how computer usage can screw up the rating system of an internet club is ICC (chessclub.com). There, for bullet, blitz and standard categories, you are allowed to use a computer account and play rated games against human opponents. That must be why the top blitz ratings there are so inflated, with all-time top 20 ratings are in the 3600-3750 range !!
I believe this is because of the following: For most players, strong computer accounts are unbeatable in blitz. That has driven those accounts to the 3000+ range. Then there are those few smart guys who have learned tricky ways to beat some of those specific computer accounts. They used specific openings and specific strategies to explore a single mistake. Since those computers played very repetitively and do not learn by themselves with their failure, those humans were able to do the same trick over and over, hundreds of times, driving themselves to the 3500 range. But then, greedy enough, those humans accounts rated so high played against other human accounts with GMs titles (some of them which were actually much stronger then them, such as Nakamura and Caruana). Finally, that allowed a selected crowd of GMs to move into that super inflated rating range of 3600-3750.
That does not mean that the entire crowd of players got a 1000 shift in their ratings. Much to the contrary, players below 2000 level do actually get their ratings lowered about 200 points, due to computer accounts pushing them down. While the 2500-3000 range is probably scarce.
Recently ICC has created separate ratings for the 1-minute, 3-minute, 5-minute, 15-minute pools and 25-minute pools, were computer accounts are not allowed. In those pools, the ratings are more modest, all below 3000 range (except only for the 1-minute pool rating of Nakamura).
PS: just a note about my previous post. FM JusticeBot won recently some of the 3+0 games played against Nakamura (maybe one out of 3-4 games I guess). The fact that JusticeBot does not have a GM title is probably an indication that he probably plays much better over the internet than OTB, reinforcing what celot said about humans having a "broad spectrum of ability".
I do agree with those who say that average players strength in FIDE pool is greater than here, and over the internet in general. Because it takes you a bigger effort to play OTB than it does to play over internet, the noncompetitive players who play just for fun are more likely to play over the internet only. For instance, my 7 year old daughter has an account in lichess, but she would not yet play a rated OTB game elsewhere. Yet, I don't think this is a big factor. The pool with more competitive players tends to have an slightly higher average strength, but not much.
I also agree that lichess ratings are accurate, due to Gliko2 system and a more humanized environment. That's why rating chart shows regular deviation.
The best example of how computer usage can screw up the rating system of an internet club is ICC (chessclub.com). There, for bullet, blitz and standard categories, you are allowed to use a computer account and play rated games against human opponents. That must be why the top blitz ratings there are so inflated, with all-time top 20 ratings are in the 3600-3750 range !!
I believe this is because of the following: For most players, strong computer accounts are unbeatable in blitz. That has driven those accounts to the 3000+ range. Then there are those few smart guys who have learned tricky ways to beat some of those specific computer accounts. They used specific openings and specific strategies to explore a single mistake. Since those computers played very repetitively and do not learn by themselves with their failure, those humans were able to do the same trick over and over, hundreds of times, driving themselves to the 3500 range. But then, greedy enough, those humans accounts rated so high played against other human accounts with GMs titles (some of them which were actually much stronger then them, such as Nakamura and Caruana). Finally, that allowed a selected crowd of GMs to move into that super inflated rating range of 3600-3750.
That does not mean that the entire crowd of players got a 1000 shift in their ratings. Much to the contrary, players below 2000 level do actually get their ratings lowered about 200 points, due to computer accounts pushing them down. While the 2500-3000 range is probably scarce.
Recently ICC has created separate ratings for the 1-minute, 3-minute, 5-minute, 15-minute pools and 25-minute pools, were computer accounts are not allowed. In those pools, the ratings are more modest, all below 3000 range (except only for the 1-minute pool rating of Nakamura).
PS: just a note about my previous post. FM JusticeBot won recently some of the 3+0 games played against Nakamura (maybe one out of 3-4 games I guess). The fact that JusticeBot does not have a GM title is probably an indication that he probably plays much better over the internet than OTB, reinforcing what celot said about humans having a "broad spectrum of ability".