correspondence chess in the age of NNUE

[Peter Berger]

Even today engines are only a decision
support tool, true with NNue and so on they have improved, but
correspondence chess at higher level also needs human understanding,
..

Maybe ( although the draw rates actually suggest otherwise), but the question is whether any of this "decision making" couldn't be done by a computer program just as well - or maybe even way better.

I remember the posts of Uri Blass about his correspondence chess career like 20 years ago e.g. - nothing he reported felt like things a computer couldn't do in principle.

[jefk]

Peter Berger,
only at the highest levels are we getting almost only draws
in ICCF correspondence chess. At lower levels, it's not a matter
of who's comp is the fastest, but (more) of chess understanding
(especially openings (*) i suggest, but i'm not going to explain all).
The hybrid centaur system comp/human is stronger than comp or
human only, there's a synergy there, but because chess is fundamentally
a draw, (only) at the very highest (almost WC) level we getting mostly draws
it seems, but that's because the playing level if Black then is very high.
Against an unexperienced correspondence chess player, i often win
with White with ease... And then even at top level there sometimes
are still occasional wins at top level, which make the difference, for example
former (2 yrs ago) Russian world champ Dronov apparently did not use
such a fast comp, but simply did a lot of (computer assisted) analysis
with human judgement combined with lots of chess knowledge
(and time, ofcourse) and thus could win a few games (yes, with White).

So nope, it's not as simple as you think, but it's also not everybody's
favorite way of chess, and maybe therefore Uri Blass you mentioned (eg. not willing or not being able to spend enough time on it) got fedup with it.
For the rest you can only judge it you try yourself, i would suggest.
(*) to give a hint what opening book would you use ? (simply the SF .bin book,
well that imho won't make you a correspondence master, that's for sure (not even going to talk about methods of computer assisted analysis, building tree, using Mcts, or whatever advanced methods rather than let your comp crunch one line with alfa beta during a night. And last but not least, although playing time average 3 days looks a lot, if you play a lot of games, time management become important, sometimes very important (yes, you guessed it, again at the highest levels).

NB as for Nnue this is certainly not such an amazing jump in comp program strength for me , during last three decades comps have gradually increased in playing strength sometimes with some jumps, but nevertheless in a more or less gradual way. Iccf players have been allowed already now for a decade or so
to use a comp, nevertheless the results are extremely varied. Anyway you must like chess (usually as result of also playing as human) otherwise you wouldn't
have the patience to mindlessly plugin the comp results.

[Thomas A. Anderson]

I suppose the actual wording should read something like: If a move is played that repeats the position for the third time, either player may call it a repetition, with the repeating side getting 1/4 point, the other side 3/4 point. For engine vs. engine play I suppose it would be automatically scored as such. The main point isn't to avoid repetitions played to avoid a no draws by agreement rule, it is that testing has shown that prohibiiting (or penalising) repetitions reduces draws far more than the stalemate + bare king rules do. There are many drawn endings where if you aren't allowed to repeat, you will lose (many even material King and pawn endings would be in this category). I don't claim that this rule avoids all draws or even most draws; I do claim that it roughly doubles the percentage of decisive games when the level of play is high enough that that percentage is a fairly small number. If you want to eliminate all or even most draws we would need more rules.

Hi Larry.
I assume that you are aware, that this rule will change the character of the game massivly. Not only that KPvKP are effected, but also many other totally equally balanced endings. Even KRvKR and KQvKQ can only try to escape by the 50-move-rule. Chances tobe adopted by ICCF? In my oppinion close to zero. I seen to much reluctantness in accepting rule that affects the game characteristics, even ones, that didn't smell such "unfair" in many situations.
Cheers,

[lkaufman]

I suppose the actual wording should read something like: If a move is played that repeats the position for the third time, either player may call it a repetition, with the repeating side getting 1/4 point, the other side 3/4 point. For engine vs. engine play I suppose it would be automatically scored as such. The main point isn't to avoid repetitions played to avoid a no draws by agreement rule, it is that testing has shown that prohibiiting (or penalising) repetitions reduces draws far more than the stalemate + bare king rules do. There are many drawn endings where if you aren't allowed to repeat, you will lose (many even material King and pawn endings would be in this category). I don't claim that this rule avoids all draws or even most draws; I do claim that it roughly doubles the percentage of decisive games when the level of play is high enough that that percentage is a fairly small number. If you want to eliminate all or even most draws we would need more rules.

Hi Larry.
I assume that you are aware, that this rule will change the character of the game massivly. Not only that KPvKP are effected, but also many other totally equally balanced endings. Even KRvKR and KQvKQ can only try to escape by the 50-move-rule. Chances tobe adopted by ICCF? In my oppinion close to zero. I seen to much reluctantness in accepting rule that affects the game characteristics, even ones, that didn't smell such "unfair" in many situations.
Cheers,

Usually, in fully "equal" endings KRvKR and KQvKQ, the game will ultimately be drawn by 50 move rule instead of by repetition, no big change here since the players can agree to a draw without waiting for 50 move rule. But there will be many equal material endgames including those two where one side has an advantage but not enough to win, yet enough to force the opponent to repeat or lose, so those endgames will get 3/4 of the point. I consider this a good thing, but of course not everone will agree. The only alternative to this or similar rule changes that would make correspondence chess playable again at the top level is to mandate bad openings or a bad startposition (with paired games), as in TCEC. I leave it to the chess world to decide which is the better option, or whether Correspondence chess should fade away.

[Uri Blass]

In reply to Peter Berger, no, the suggestions for rule changes
are not pointless, certainly not for Fide/ICCf correspondence
chess where comps are allowed. Even today engines are only a decision
support tool, true with NNue and so on they have improved, but
correspondence chess at higher level also needs human understanding,
and the ability to use different tools; opening theory is especially
important in such slow games, as wel as the ability to improve
on existing theory. In addition, in middle game strategy its a
matter of deep analysis, not simply running one engine for a night.
This may be especially true for the transition between middle
game and endgame.

Think of centaur (computer assisted), comp/human hybrid chess
but then for slower times.

For the rest such a discussion belongs better in the other
thread i noticed now about correspondence chess, ie.
what's the 'use' of correspondence chess. Well some Otb
players may find it interesting for example to gradually
improve their opening knowledge, other players may
like the scientific approach, rather than the tactical fight
of faster Otb games. whatever.

So some rule modification to address the drawing problem
in top correspondence chess certainly is relevant and i'm
going to address in a next message.

I do not agree that opening theory is important today with slow correspondence games.
You can basically use the moves of the engine without opening book at slow time control.

Engines may not find the best move at faster time control but they play better at slower time control and I guess one hour per move is usually enough not to lose.

My experience as a correspondence player is that even at the high level humans in most games are not stronger than engines.
analysis by engines usually did not help me to find better moves that the engine did not suggest and a case when it helped was an exception.

[Uri Blass]

Peter Berger,
only at the highest levels are we getting almost only draws
in ICCF correspondence chess. At lower levels, it's not a matter
of who's comp is the fastest, but (more) of chess understanding
(especially openings (*) i suggest, but i'm not going to explain all).
The hybrid centaur system comp/human is stronger than comp or
human only, there's a synergy there, but because chess is fundamentally
a draw, (only) at the very highest (almost WC) level we getting mostly draws
it seems, but that's because the playing level if Black then is very high.
Against an unexperienced correspondence chess player, i often win
with White with ease... And then even at top level there sometimes
are still occasional wins at top level, which make the difference, for example
former (2 yrs ago) Russian world champ Dronov apparently did not use
such a fast comp, but simply did a lot of (computer assisted) analysis
with human judgement combined with lots of chess knowledge
(and time, ofcourse) and thus could win a few games (yes, with White).

So nope, it's not as simple as you think, but it's also not everybody's
favorite way of chess, and maybe therefore Uri Blass you mentioned (eg. not willing or not being able to spend enough time on it) got fedup with it.
For the rest you can only judge it you try yourself, i would suggest.
(*) to give a hint what opening book would you use ? (simply the SF .bin book,
well that imho won't make you a correspondence master, that's for sure (not even going to talk about methods of computer assisted analysis, building tree, using Mcts, or whatever advanced methods rather than let your comp crunch one line with alfa beta during a night. And last but not least, although playing time average 3 days looks a lot, if you play a lot of games, time management become important, sometimes very important (yes, you guessed it, again at the highest levels).

NB as for Nnue this is certainly not such an amazing jump in comp program strength for me , during last three decades comps have gradually increased in playing strength sometimes with some jumps, but nevertheless in a more or less gradual way. Iccf players have been allowed already now for a decade or so
to use a comp, nevertheless the results are extremely varied. Anyway you must like chess (usually as result of also playing as human) otherwise you wouldn't
have the patience to mindlessly plugin the comp results.

At lower levels players do not use engines or decide not to choose better moves that the engine suggest because they do not like them
or decide to trust some bad books instead of giving the engine to analyze also in the opening.

I believe that at the higher levels many players simply play the engine move without thinking.

[DrCliche]

You are right that there's little that can be done about the influence of computers in correspondence chess, but the topic here is draw reduction, which is a completely different issue. You can either change the start position (i.e. chess960 or mandated openings), or change the rules. For me the only interesting rule changes are those that reduce draws with the least effects in most positions, keeping it as close as possible to normal chess. If you want a very different chess, shogi is the best option in my opinion.

I suppose I was reading between the lines a bit too much, then. I would say the reason the draws are a problem is because it's now becoming clear that computers are close enough to perfect that they rarely make moves that are exploitable by their peers, and human input and exploration usually make little to no difference.

Changing rules to reduce or eliminate draws would certainly change the appearance of tournament cross tables, but wouldn't change at all the fact that humans largely have no place in the process of playing games. I thought that was the real problem, as evinced by the fact that top correspondence players are quitting in droves and the correspondence game appears to be dying.

[Uri Blass]

I think that it may be more interesting to have a correspondence tournament when the target is to beat the machine.


Let say you have 1 year and the target is to produce a win against stockfish when stockfish play at higher level than the level that your opponents give it.
Stockfish may use a single core and 4096 mbytes hash(so stockfish is deterministic) and you have to use a fixed version without an opening book.

Every participant is going to send a win against stockfish at x nodes per second when the participant who can show a win against the bigger x after one year is the champion.

It is possible to have one championship for white and one championship for black.
I think that we can learn from this type of championship what is possible in correspondence chess because I am tired of seeing
only draws in mark young's fair games of stockfish that are 100% draw at long time control
and I would like to see games when stockfish lose at long time control.

[lkaufman]

You are right that there's little that can be done about the influence of computers in correspondence chess, but the topic here is draw reduction, which is a completely different issue. You can either change the start position (i.e. chess960 or mandated openings), or change the rules. For me the only interesting rule changes are those that reduce draws with the least effects in most positions, keeping it as close as possible to normal chess. If you want a very different chess, shogi is the best option in my opinion.

I suppose I was reading between the lines a bit too much, then. I would say the reason the draws are a problem is because it's now becoming clear that computers are close enough to perfect that they rarely make moves that are exploitable by their peers, and human input and exploration usually make little to no difference.

Changing rules to reduce or eliminate draws would certainly change the appearance of tournament cross tables, but wouldn't change at all the fact that humans largely have no place in the process of playing games. I thought that was the real problem, as evinced by the fact that top correspondence players are quitting in droves and the correspondence game appears to be dying.

Let's suppose that the rules of chess were changed in some way to eliminate draws while keeping the chances of White and Black fairly equal. Just for example, we've already shown with many tests that forbidding Black from castling short while giving Black draw odds is a fairly equal drawless variant of chess. There are many ways to achieve this goal, let's not worry about which way is chosen for now. The question then is, will correspondence players with a lot of skill score way over 50% in this variant against players who rely only on an engine (assuming engines are modified to play well under the new rules)? I believe that the answer is "YES", although we can't be sure without actually testing this hypothesis.

[carldaman]

Can a test be run to see if top human corr players (playing as centaurs) can defeat unaided top engines running on some 'regular' hardware?

If the answer is no, then what is the point of seriously continuing with corr chess in its current form, where players can be assisted by engines and are able to choose their own openings?

If the answer is yes, then maybe there is still a future in (computer-aided) corr chess and these unaided engines could even be entered as participants in top competitions to spice things up by adding to the overall win percentage.