Do humans still provide value in correspondence chess? — LessWrong

x

Do humans still provide value in correspondence chess? — LessWrong

5 comments, sorted by

Click to highlight new comments since: Today at 12:34 AM

[-]Yair Halberstadt3y100

Presumably you could test this directly. Have 100 players play a correspondence game against each of the top 3 chess engines, where you gave them a max amount of compute time they could use (say no more than 2 hours per move) and they just play directly against the computer.

My guess is they would have a slight advantage, at least over stockfish, if only by exploiting known bugs/adversarial situations.

FWIW, the two main online chess sites forbid the use of engines in correspondence games. But both do allow the use of opening databases.

(https://www.chess.com/terms/correspondence-chess#problems, https://lichess.org/faq#correspondence)

[-]Jonathan Paulson3y20

https://www.iccf.com/ allows computer assistance

[-]RHollerith3y74

Do humans still have something to add? The continued existence of correspondence chess, IMO, suggests that they do.

Sorry, but that is absurd: the fact that humans continue to play correspondence chess is only the most teeniest tiniest evidence that humans have something to add.

I am interpreting, "humans have something to add," as "the best human player using computers still has some kind of advantage over a computer alone where no human is allowed to intervene in the operation of the computer over the course of the game except perhaps to ensure a steady supply of electricity".

Note that I am not opining either way on your overall question. I am commenting only on what conclusions we can draw from the fact that people still play correspondence chess.

[-]Jonathan Paulson3y30

Why do you think so?

Presumably the people playing correspondence chess think that they are adding something, or they would just let the computer play alone. And it’s not a hard thing to check; they can just play against a computer and see. So it would surprise me if they were all wrong about this.

[-]RHollerith3y*79

Human social behavior is complex. Maybe some or all of the winners of ICCF tournaments won by merely parroting the moves chosen by an engine, but they chose not to admit it out of a worry that admitting it would cause a change in the rules that would disadvantage them in future tournaments.

A document titled "2023 ICCF Rules" does not exactly explicitly encourage the parroting behavior I just described (though it does not explicitly disallow it either):

In ICCF event games, players must decide their own moves. Players are permitted to consult prior to those decisions with any publicly available source of information including chess engines (computer programs), books, DVDs, game archive databases, endgame tablebases, etc. . . . No other consultation with another person concerning analysis of an active position is allowed . . .

The only things I omitted from the paragraph I just quoted have to do with humans playing as a team.

I have quoted from the public written rules of the tournaments because public written information is all I have access to. Communities often develop unwritten rules that strongly influence human behavior -- rules we would have no way of knowing about without asking a community member in a context in which the member has some basis for trusting us.

It might be the case that ICCF's leaders see it (correctly IMO) as impossible to enforce a rule against chess engines, so they allow them as a practical measure, but they don't like them, and most of the winners know that, which again would tend to cause anyone who won by merely parroting moves chosen by an engine to choose not to announce that fact.

Or it might be that the majority of those with a megaphone that reaches the correspondence-chess community maintain that chess engines have ruined the once noble and delightful correspondence-chess scene, with again the same effect.

If the organizers of a tournament explicitly declared that one of the purposes of the tournament is to determine whether human-computer teams can outperform computers alone, then that would start to be evidence worth considering (against, e.g., the evidence provided by the overwhelming dominance of computers over human-computer teams in chess with other rules (other time controls to be specific) -- particularly if there was decent prize money.