Repeating moves to add time

[Kempelen]

Have you ever seen an engine that can repeat moves to add time to its clock and then continue playing as normal? would that be difficult to implement?

I think it must be very difficult because it is a path-dependent feature and one always store the pv only, but I am curious if that feature could be implemented..... what is your opinion.

[marcelk]

Have you ever seen an engine that can repeat moves to add time to its clock and then continue playing as normal? would that be difficult to implement?

I think it must be very difficult because it is a path-dependent feature and one always store the pv only, but I am curious if that feature could be implemented..... what is your opinion.

I thought about that once but figured that it only makes sense when there is a time control to meet. With server-type time controls it can only gain a few increments. There must be other ways to make a program stronger than by playing delay moves.

[hgm]

Problem seems to me this also adds time to the clock of the opponent.

In Xiangqi, where you usually can give a perpetual at any time, so delaying moves are always available, this issue can be more important.

[mhalstern]

I've always wondered if any engines had anything programmed in to try and take advantage of other engines time pressure. It should be easy to figure out how much time the opponnent has, based on the time control and the amount of time that it has spent. Would it ever be beneficial to play fast sharp moves, when the opponnent is in time pressure, or would this human strategly backfire in engine matches.

[Andres Valverde]

I've always wondered if any engines had anything programmed in to try and take advantage of other engines time pressure. It should be easy to figure out how much time the opponnent has, based on the time control and the amount of time that it has spent. Would it ever be beneficial to play fast sharp moves, when the opponnent is in time pressure, or would this human strategly backfire in engine matches.

Both yours and Fermin's suggestions seems interesting to me. In your case though, I have read several times that it is not a good tactic for humans to play fast because the opponent is in zeitnot (because of blunders mainly), but maybe it is worth for engines.

About Fermin's idea, it is clear that it would help in some cases. For example, being in move 39 (with TC = 40 moves in 5 min. f.i), 5 seconds left and score +230, if the engine can repeat a move without threefold, it will reach the control without ruining his good position. Almost every chess player has experienced something similar.