Why do Chess Engines Play so Much Better than Bridge Playing Computers?


"We already know bots beat the best players in poker" was your statement. That was the entirety of your claim. I lent clarification, pointing out only in a heads up format has a poker program beaten a very few  professional poker players in a ONE time match. In heads up play, a player can "play the cards". Reading your opponent, recognizing tells, certainly is very important, but strategies revolve around the cards dealt as there are only two hands in play. In a tournament, at the start, 10 players sit at the table. Players "play the player(s). The cards dealt are far less important than in heads up. Who's going to trust a program to read 9 other players? Nobody. To start, the program can not get any "tells", as obviously it can't observe the other players expressions and reactions. So all of Texas Holdem is not just TM where strategies are the same.

MustangMate escribió:

"We already know bots beat the best players in poker" was your statement. That was the entirety of your claim. I lent clarification, pointing out only in a heads up format has a poker program beaten a very few  professional poker players in a ONE time match - which was dependent on the cards dealt as a major variable.

It was a good clarification, you were wrong in stating I was wrong though.

MustangMate escribió:

When it comes to Poker you are clueless PF.  Any such match has NOT been organized and if one is, it is but a single match. Dozens of matches would provide little data. Professionals are not going to put up their own money and play in such. If played with house money ,the betting would be altered. Poker depends on the cards dealt. It's not like chess in which the same position is dealt every time at the start. I played pro for years and it's not just any opinion. You say TH is just TH. Again you are entirely wrong. Ring games, tournaments. heads up play all present different challenges that require totally different strategies. A program that is effective in heads up play would be quite useless at a table of 10 human players in a tournament format. 

I didn't say the strategies were the same nor that a non-heads up match had been organized, I'm merely telling you that a match could be organized just like it has been previously (twice or more in fact), those matches had 4 professional players including Doug who stated that he believed computers would eventually become better than poker players.



The 1st match was strictly heads up matches.

The second match included a six player tournament with the AI program - which won ! The article describes the programming involved, how it employed unconventional strategies, making bets that no doubt threw the humans off. It goes on to describe how the program learned, much like Lc0, that the old programs of heads up play were not applicable to tournament play. Quite amazing the comp won. But it is only one match. Gotta give humans credit for also being able to learn and adapt, to take note of the comps unconventional betting, and perhaps fleece the comp of all it's chips the next time around. 

I think it's very similar to the bridge question, where the added factor of playing with a partner comes into the equation. No doubt "someday" these comps will be far more successful and win many more matches than at present. But the games are dissimilar in one major way. The starting position is always the same for chess, abstract for bridge and poker. It just won't happen that they will win every time at poker as they do in chess.  It'll be quite some time for the bridge programs to compete at the highest levels. It's interesting how the OP says the programs sometimes "cheat" but he hasn't fully explained. Teams must declare the bidding system being used in advance. He stated a program made unnatural bids that the cards did not warrant. I'm wondering if a human monitor of the comp over rode the bidding, threw in a monkey wrench or if the comp did it on it's own. If so ... a bug in programming or it was thinking for itself !



All of the AIs that displayed superhuman skills at two-player games did so by approximating what's called a Nash equilibrium. Named for the late Carnegie Mellon alumnus and Nobel laureate John Forbes Nash Jr., a Nash equilibrium is a pair of strategies (one per player) where neither player can benefit from changing strategy as long as the other player's strategy remains the same. Although the AI's strategy guarantees only a result no worse than a tie, the AI emerges victorious if its opponent makes miscalculations and can't maintain the equilibrium.

In a game with more than two players, playing a Nash equilibrium can be a losing strategy. So Pluribus dispenses with theoretical guarantees of success and develops strategies that nevertheless enable it to consistently outplay opponents. - Copied

The issue with Bridge is that changing strategies is illegal in the bidding phase. After a system of bidding is declared, that system must be followed. It is against the rules to start a round of bidding employing one system (which the opponents know is being played) and then mid stream change to another system, unbeknownst to the opponents.  









What happens is that there are major flaws in  the bidding system and very major flaws when the computers are defending against a contract, 

One thing I have noticed  is that the bots [computers] use the bidding system--sometimes they make a bid that is not in the system such as when my bot partner bid 3 hearts it was supposed to show a 4 card suit but it had a 7 card suit,

Regarding poker--there have not been enough tests and humans will adjust to what the computer does after a few sessions. But I agree that some day the poker playing computers will be better than the best humans.