UCI chess engine with adaptive playing strengh?

[PeterO]

Looking for UCI chess engines with adaptive playing strengh.
1. Do you know one?
2. I heard from Stockfish DD - but I can not find a download.

Thanks for your answer!

Peter

[Ajedrecista]

Hello Peter:

Looking for UCI chess engines with adaptive playing strengh.
1. Do you know one?
2. I heard from Stockfish DD - but I can not find a download.

Thanks for your answer!

Peter

Stockfish DD (SF 4.5) can be downloaded from the official SF web:

https://stockfishchess.org/download/
↓
Old (archived) releases of Stockfish
↓
Dropbox - Stockfish 4.5 (DD)

Regards from Spain.

Ajedrecista.

[PeterO]

Thank you Ajedrecista!

Does anybody know another engine with adaptive playing strengh?

Peter

[pocpit]

Searching for "stockfish adaptive" I found this site: https://blog.bradbeattie.com/adaptive-stockfish/: "It’s a chess algorithm that tries to match its current advantage with a given target. With a target of 0, it tries to play as well as its opponent. "

[mvanthoor]

Thank you Ajedrecista!

Does anybody know another engine with adaptive playing strengh?

Peter

Shredder + its own GUI can automatically adapt playing strength per game and give you a "Shredder ELO" rating (not comparable with any other ELO rating; it only means that if your rating goes up, you get better).

IMHO, adapting playing strength during the game, which is what the DGT Centaur does, I believe, is bad. That just means that the computer will either:

- Keep the position a bit better for the player (Friendly mode)
- Keep the position completely equal (Challenging mode)
- Keep the position a bit better for the engine (Expert mode)

So it means that, as long as you don't make a massive blunder, the engine will keep the position within a certain evaluation range. It means it is very hard to win in Challenge or Expert mode, because the position will always be equal or worse for you, except if you make such a brilliant move that the engine won't compensate (on purpose).

[Fritz 0]

Fritz GUI also has the "Friend mode" option, where the engine tries to adapt its play to the opponent's strength. It may work for other engines too, not just Fritz, but I've never tested it. In practice, the engine just plays stronger the next game when it loses and weaker when it wins, regardless of the actual quality of the opponent's play. In other words, it adapts to the result, not the playing level. And it often makes unnatural and "unhuman" mistakes, so it's not a very realistic sparring partner, except maybe in blitz and fast rapid, where humans tend to play nonsense too.

[CornfedForever]

Fritz GUI also has the "Friend mode" option, where the engine tries to adapt its play to the opponent's strength. It may work for other engines too, not just Fritz, but I've never tested it. In practice, the engine just plays stronger the next game when it loses and weaker when it wins, regardless of the actual quality of the opponent's play. In other words, it adapts to the result, not the playing level. And it often makes unnatural and "unhuman" mistakes, so it's not a very realistic sparring partner, except maybe in blitz and fast rapid, where humans tend to play nonsense too.

Isn't the whole idea of 'adaptive' a bit of a chimera being chased? I mean, by nature one plays better in one game than another and what kind of mistakes made in one may not be repeated...they come in many colors.

I think ideas like Larry Kaufmans method of pegging Dragon to specific elos and playing against that is more helpful to a player. It's not going to be 100% perfect either, but probably better. If one wants to "play a Swiss tourney' of sorts with different rated opponents, just play one game with Dragon set to different elo depending on your result in the previous game.

[Rowen]

Looking for UCI chess engines with adaptive playing strengh.
1. Do you know one?
2. I heard from Stockfish DD - but I can not find a download.

Thanks for your answer!

Peter

Hi
The stockfish derivatives-Honey,weakfish etc used to have an adaptive strength parameter, also the Lucas Chess GUI had the option to select parameters such as 'slightly better/worse/about the same' etc when playing against an engine.

[Fritz 0]

Fritz GUI also has the "Friend mode" option, where the engine tries to adapt its play to the opponent's strength. It may work for other engines too, not just Fritz, but I've never tested it. In practice, the engine just plays stronger the next game when it loses and weaker when it wins, regardless of the actual quality of the opponent's play. In other words, it adapts to the result, not the playing level. And it often makes unnatural and "unhuman" mistakes, so it's not a very realistic sparring partner, except maybe in blitz and fast rapid, where humans tend to play nonsense too.

Isn't the whole idea of 'adaptive' a bit of a chimera being chased? I mean, by nature one plays better in one game than another and what kind of mistakes made in one may not be repeated...they come in many colors.

I think ideas like Larry Kaufmans method of pegging Dragon to specific elos and playing against that is more helpful to a player. It's not going to be 100% perfect either, but probably better. If one wants to "play a Swiss tourney' of sorts with different rated opponents, just play one game with Dragon set to different elo depending on your result in the previous game.

I completely agree.