If I play ar a pace similar to a rapid time contfol game, I think I win 50% of the games towards a 1300 - 1400 bot. And my actual rapid rating is 1100. So I would say they are something like 2-300 points overrated (compared to chess.com rapid in this rating range).
Ihave beaten two 1700 bots but it took meI think almost ten tries each. Which it statistcally should do if their actual chess.com rating would be around 1400 - 1500.
Lol i was wondering are people ratings are acurate here lol i am just keep winning against hih rated players idk why xd
Ah and for your question... Bots ratings aren't accurate i mean.. Mitten is not 1 elo ofc
And others also same not that accurate
from my exp they are absolutely lower rated than the showed i am barely 500 and can beat with ease Sven who he is 1100 https://www.chess.com/analysis/game/computer/32649197?tab=analysis the blunder of a queen i made on the 26 move was a mouseslip and he even didnt see the blunder. judge for yourself
I'm 980 (1150 at peak), and after some tries i can easily beat bots that are 1800, I just don't think bots think on one level, they mostly blunder sometimes, play well other times, no in beetwen
Their ratings might be accurate for like humans playing with 30 seconds time control or something?
On Noam's profile it says his highest bullet rating was 2725
Bots don't play like human ratings, they'll play hyper-accurate for some period, then blunder to keep the caps at ~the same percentage
For example, when beating Noam for the first time, he made only 1 mistake in the game, and that was blundering a simple backrank tactic, not something any 2200 would do
Their ratings might be accurate for like humans playing with 30 seconds time control or something?
I hope you mean they play like they only have 30 seconds, because in an actual 1|0 game, you can pretty easily beat the 3200 engine
Bots may play different, like looking at depth 4 for 10 possible options, vs a human who can maybe look at depth 6 but only for the 2 most obvious attacks. So to beat a bot, looking at depth >4 for your main lines of attack may work, while a human senses your main lines of attack and goes in higher depth there (say depth 6), so you'd need depth >6 in your main lines of attack to win, which would make it seem like humans are harder to beat. The alternative to beat the human may be to look into alt lines of attack that the human will blunder/miss, but at lower levels that may not happen and you may stick only to your main attacks.
So i guess humans may be more alike and thus be harder to beat. Like playing yourself. Your strengths are your opponent's strengths and cannot be exploited, and your weaknesses cannot be exploited because they are your opponent's weaknesses too.
This doesn't explain why the bot rating does not seem accurate, though 
I was able to beat Catspurrov (1400) on the first try and I'm rated 600 something so bots are definitely over-rated. But I definitely took my time
The fix would be to have the bots elo reflect their win-rate vs the elo of the users’ they play.
No
The fix would be to have the bots elo reflect their win-rate vs the elo of the users’ they play.
Good idea but might not really work because I think many players (like myself) often abort a game against a bot if they don't like the position but still has some winning chances. Or they don't take the game very serious.
I'm literally at a 400 elo (which I know is trash) but have no problem with any of the bots 1400 and below
I wonder why lmaooo
if they gave mittens an accurate ranking it'd scare people
they wanna bait people in thinking the bot is an easy win