So just one more view I had playing some games here, they seem exactly the same bots as the chess.com app bots, they have been up there a long time already and the website seems to be using them from now on, since I played a lot on mobile I can give some differences I spotted.
The ratings are closer to FIDE ratings than before, 1500- elo are harder and very punishing on blunders, but the 1700+ elo are slightly easier than before.
They punish blunders like a savage, but if you get ahead they are more likely to blunder themselves more than before, at least for lower rating bots.
They are much more aggressive than the older bots, if they spot any weaknesses, they go all in on it.
The openings they play exactly as their older counterpart.
overall the ratings being closer is better, but they play much less like real people, they spot blunders and punish them in a way their rating wouldn't allow them to, but they also hang some pieces that in their rating it shouldn't happen, so I think they are worse to play than the older ones since its less human-like.
Wow, thanks for looking into this. I don't have the expertise to verify your suppositions, but your work sounds spot-on. Really wish I could be a fly on the wall at chess.com headquarters when they discuss how to tweak the bots. Would have preferred the retired the old bots and introduced new ones with the new attributes, actually.
So just one more view I had playing some games here, they seem exactly the same bots as the chess.com app bots, they have been up there a long time already and the website seems to be using them from now on, since I played a lot on mobile I can give some differences I spotted.
The ratings are closer to FIDE ratings than before, 1500- elo are harder and very punishing on blunders, but the 1700+ elo are slightly easier than before.
They punish blunders like a savage, but if you get ahead they are more likely to blunder themselves more than before, at least for lower rating bots.
They are much more aggressive than the older bots, if they spot any weaknesses, they go all in on it.
The openings they play exactly as their older counterpart.
overall the ratings being closer is better, but they play much less like real people, they spot blunders and punish them in a way their rating wouldn't allow them to, but they also hang some pieces that in their rating it shouldn't happen, so I think they are worse to play than the older ones since its less human-like.