Why are the bots easier to beat than real players?

Sorry for any grammar errors 😅

[apologies for hoofing an old thread but this had been on my mind since I only ever really play against the bots, usually for a few minutes now & then when I'm on an underground train with no internet access].

The bots obviously don't play exactly like human players, being, relative to their ratings, disproportionately weak in some facets of the game, and disproportionately strong in others.

But my instinct is that the bots' ratings are about right for blitz chess [e.g. 3-5 minute chess.com games], maybe especially towards the faster end of that range. The bots' ratings are too low against a player who's limiting himself to bullet timescales, and too high against a player who's taking a decent amount of time.

I think the bots on this site (at least the lower rated ones) are programmed to make mistakes. I have trouble beating a computer even at the "easy" setting, but I can beat some of the bots here (Catspurrov is probably the highest rated bot I managed to beat after losing a few times). It seems to me that, like humans rated about my level, the bots make mistakes.

I do not know if I am qualified to give advice (seeing that I have very low skill in chess), but I think you need to check if you are hanging a piece before you move. Also check if your opponent is hanging a piece after he moves. 

Pretty much that's how I play as I do not really know any theory and if my opponent does not make an obvious mistake, I usually lose.

I am pretty comfortable with Nelson, Maria, Sven, and Tomas. I beat the Hacker with no takebacks/hints for the first time today, and I play this bot obsessively. I actually thought I was good, until I started getting matched against random players and my rating dropped from 600 to 250-300. I attribute it to several things. First, I like getting into super-complex tangles with the bots and trying to figure out the one little move that unravels everything. People at the 250-300 level are going to default to simpler strategies, so playing against bots is actually bad at teaching you these strategies. Second, there is time pressure. Third, people have ideas as to what they want to do - they want to use the bishop and queen for a quick mate, or whatever, and so I have to figure out what their ideas are, rather than what the next logical move is in a beam search. Fourth, I'm probably just not as good as I'd like to think I am, and 250-300 is my real rating.

It's quite simple. The bot ratings are wrong. It's nothing to do with style of play, difficulties of emulating human-style thinking, or anything complicated like that. It's just that the bot ratings are numbers assigned by someone, not ratings allocated by the statistical process of giving and subtracting points according to how often you win against someone of equal rating. If the bots were thrown into the same rating pool as humans, they would level out at a rating where they win against roughly half of humans at the same rating level.

They might win and lose in very non-human ways, but they'd still do it roughly half the time.

The bot ratings are seriously off, way way off. I'm a 400 player but can win against Antonio, who's supposed to be 1500, so go figure. This is nearly a four-fold discrepancy!

The thing about bots is that you can take as long as you want for your move but you don't have to wait for their move. Use them to train yourself out of a habit of jumping in without thinking.