Chess.com Computer Playing Algorithm Variation

There are various levels you can choose when playing the computer on chess.com

I am curious to know the differences between the levels.

Like how does the paying algorithm vary with each level?

I guess it maybe varies the depth of the tree search when exploring moves but it would be interesting to know how the search depth varies by level.

I looked on here but could not find the answer.

https://support.chess.com/

I think the level 1-10 computer is the same for paying and non-paying. It runs on your device, so its strength depends on your device.

Chess.com has never revealed anything about the algorithms.

Sure thanks.

But I guess the computer plays better on level 10 than on level 1 so I was really asking if there is any clues as to how the computer plays better.

e.g. maybe on level it explores all moves 1 step ahead. on level 2 is 2 steps ahead etc.

Just trying to get a feel for how it thinks.

Old question, but I would love to have some insight on this too. I'm curious because level 9 is suppossed to have an ELO of 1400, and I beat it about half of the time; however, my "live" ELO is 560, and I win about half of the time there.

Edit: I'm starting to wonder, because the computer starts out playing somewhat tightly, and it suddenly makes ridiculous blunders, like not capturing the queen I neglected...

It's hard to take a general engine algorithm and make it play at a certain strength, unless it was designed that way. Ratings are estimates, though the bot rating should change based on people playing them, so if a 1400 bot is consistently losing to 600 rated players, it would get to a lower rating quickly.

^Thanks Martin.

My OP question was originally asking about the way in which the engine is 'handicapped' or its strength 'adjusted' to give different 'levels' against which members can play.

I am still curious to understand how the engine algorithm is adjusted to allow it to play with the 10 differing levels of skill.

Is that information explained anywhere?