Would a 75 Move Rule be better than a 50 Move Rule?

MARattigan: And that 115 move solution contains plenty of pawn moves interspersed along the way ...

@EndgameStudier

I've looked it up here:

http://galen.metapath.org/egtb50/

The longest mate in two knight v pawn under FIDE Competition Rules is 128 moves.

This is another reason for not having a fixed move limit. Humans (tournament arbiters excepted - but do they count as human) think in DTM terms. I.e. what's the quickest mate or what's the most robust defence? Fixed limits such as the 50 move rule are only suitable for computers (and not even computers if FIDE change the number of moves in the limit every few minutes). Chess should be a game for humans.

But we come back to this point. You're highly forgiving of the errors of the 'attacking' side, but not of the 'defending' side. It has to be consistent. Both players make errors, and you have to put a limit at some point for purely practical reasons in competitive situations.

As far as I'm concerned, we can never get a perfect rule, but it's very unfair on the defending side that the criteria you want is basically equivalent to 'maintaining the advantage' (our further discussion clearly indicates this, despite what you try to claim in #51). In other words, you can hardly ever get a ruling in favour of the defending side. As far as I'm concerned, there needs to be a limit, irrespective of the material configuration, whether that's 50, 75 or maybe 100 is much of a muchness to me, but I strongly suspect that they've kept it at 50 because the overwhelming majority of all endgames are winnable well inside 50 moves. Many of those longer wins had already been found when the rule was introduced.

I'm not sure you're aware that the EGTBs are automatic. They're completely mapped out. Every single move. In other words, if you play correctly, you'll beat any program whether it's using tablebases or not.

Earlier I said that the 50-move rule makes the NBK vs K endgame more challenging, but that isn't actually the main goal of the 50-move rule.  The goal is that games don't go on forever.  It also punishes inaccurate play, but that is only a side effect which isn't bad. In general the evaluation of a position can change if you make enough inaccurate moves. The 50-move rule doesn't change the nature of chess. 

People can agree to draw in drawn positions, but this rule is necessary for really stubborn opponents and players who fail to convert a winning position in a reasonable number of moves. You could make some arguments that it should be 75 moves instead of 50, but a limit is necessary. I agree that giving more moves for the mentioned 2 knight endgame is reasonable. However how often does this endgame occur? Apparently they changed it from 75 to 50, so they must have reasons for that. 

No. You're wrong. In most positions in KNNKP the DTM tablebase contains multiple accurate moves for many positions. In positions where the 50 move rule can occur it is the case that if one program with DTM EGTB chooses a particular sequence among those moves it will be mated within the 50 move rule while another choosing different moves will not. The DTM EGTB gives no guidance on which moves to choose because the 50 move rule is no part of its objective.

If you play correctly or incorrectly you certainly will not beat any program using the appropriate tablebase (DTM for Basic Rules or Syzygy for Competition Rules) unless you start off in a theoretically won position for the rules in question. I'm not sure that you're aware that accurate play and the evaluation of positions is different depending on which rules are in force. 

MARattigan: No, you don't understand. The tablebases have these endings completely mapped out. If you make a mistake, the number of moves to mate is recalculated. In other words, if you play correctly or make only minor inaccuracies, it doesn't matter which engine or tablebase you're using, you will win (provided that you're playing the advantage side).

If you play correctly and the endgame is theoretically won, the engine equipped with tablebases won't be able to save the position. (For clarification, I'm not talking about the 50 move rule being applied here, I'm talking about playing out the endings in question.)

As you say the 50 move rule was designed to cope with pig headed opponents who fail to convert a winning (or drawn) position into a win in a reasonable number of moves. I think the number 50 in the rule assumed that KBNK was the most difficult endgame that anyone could play and added 50% to the maximum moves for that ending for good luck. The only way they had to measure progress at the time was captures and pawn moves.

Times have moved on (and will move on). Endgame theory has progressed to the point where 50 moves is insufficient to accommodate analysed endgame positions and has been for a century or so. We also have better ways of assessing progress in endgames with seven men or less.

My progress rule doesn't allow games to go on forever. If an obstinate player is randomly progressing, marking time or regressing, then his probability of surviving k draw claims decays exponentially with k. If he's in a useless KNNK or KQKQ position he will be terminated (not literally) at the first claim.  

I do understand. It's just that you've changed what you're saying from your original post.

What you now say is true under the Basic Rules (so long as you mean by "minor inaccuraces" inaccuracies which don't lose). But do you understand that your phrase "play correctly" means different things according as the FIDE Basic Rules or the FIDE Competition Rules apply?

If you play accurately in traditional terms (DTM) and you play a program with a Syzygy tablebase then there are winning positions that you will fail to win if Competition Rules apply. If you play accurately in traditional terms and you play a program with a DTM tablebase then it's a lottery if Competition Rules apply.

MARattigan: Read what you wrote first, and you'll see that you were saying the exact opposite of what you're saying now. Currently I'm not discussing the move limits of competition, I'm only talking about playing against an engine, just as you were (you weren't talking about competition rules). You were saying that even if you can win that 115 move K2N v KP position against one engine equipped with tablebases, there's no guarantee that you can defeat another engine and its tablebases. I only pointed out that of course you can, because if you play correctly, there's no defence. No engine can change that fact since the tablebases are the result of sheer number crunching - they contain every possible position from seven men downwards.

I have read what I wrote and it was explicit that I was talking about the 50 move rule. It's you who have decided to stop talking about the 50 move rule. What I said was:

If you can win a 115 move position against a program using a DTM EGTB within the 50 move rule, it doesn't necessarily follow that you could win against a different program using the same EGTB.

That is true. I explained it to you in the simplest terms I could in #151.

@Numquam

Earlier I said that the 50-move rule makes the NBK vs K endgame more challenging.

Absolutely true. Too challenging I think for human players. I doubt if anybody in the world could reliably win the relevant positions against a Syzygy tablebase, let alone emulate a DTM50 tablebase. That's part of the reason I'm against fixed limits in the draw rule. They're for computers only and I think they should give us our game back.

Lol.

Most fitting would be a 1024-move rule, as the longest pawnless ending checkmate is that long.

Even two bishops versus knight frequently take up to 75 moves, so, in case a longer threshold is not implemented, what do we do: declare an obviously won KBB vs KN ending as drawn?

@Lyudmil_Tsvetkov

Most fitting would be a 1024-move rule, as the longest pawnless ending checkmate is that long.

Intriguing! I would have expected it to be much longer.

Did you work that out yourself?

Yeah, and the first pawn move isn'y until about move 86 i believe

MARattigan: My apologies (I'm now blushing with shame). However, my point still stands. If it is possible to win the specific K2N v KP ending inside 50 moves, and you play correctly, you will be able to win it against all engines with tablebases. However, if the specific endgame position is not a win inside 50 moves without capture or pawn moves, you won't be able to win it inside 50 moves against any engine equipped with tablebases. If you play an engine equipped with tablebases it's not going to play the weaker moves, it will play the best moves against your moves. Talking specifically about the 115 moves position, it is winnable without allowing 50 moves to pass without capture or pawn moves.

That sounds like an unusual method. You then have only 29 moves to mate after the pawn reaches its fourth rank.

That is fascinating that it's a binary number

Can I just ask, where have you find this information about the 115 move position? I've seen the exact same position, only mirrored, in The Encyclopaedia of Chess Ending (#1746) where the first pawn move occurs on move 14, the second on move 27, and the third on move 68. By my reckoning, that leaves 47 moves to mate by move 115. I'd be very curious to compare the solution that you've found.