I would guess that of draws in high-level chess are by agreement from a position such that there is a significant chance the game will still be decisive, but I haven’t checked. Maybe you get to draws effectively-by-agreement from unclear positions if you also include threefold repetitions that aren’t that forced, but I’d still guess weakly against.
Most draws are by agreement, but my guess is that a good chunk of these are from positions which are very drawish (in particular, often from endgame positions which are known draws).
Anyway, I agree that a bunch of draws are by agreement from a position with many pieces on the board, and it’s interesting why these exist.
I think it’s often about conserving resources as Kennaway said (mostly for other games in the same event). I think it’s occasionally about just not feeling like playing. I think it’s also reasonably commonly about effective utilities of outcomes not matching 0⁄0.5/1 for other reasons. Like, it’s easy to set up tournament situations in which a certain draw is better for a player than eg winning with probability and losing with probability . Utilities of chess outcomes don’t have to be 0⁄0.5/1 even if that is the number added to your tournament score. (Your utility also doesn’t have to match your elo.) For example, maybe you could keep pushing from an equal position and get a win probability and a loss probability, but this is the last game in the tournament and you already know a draw is sufficient for winning the tournament by 0.5 pts whereas losing the game would lead to second place for you, and winning is important enough[1] that continuing to push has negative EV for you (even though you have an advantage in the local context of the game). It’s easy to fill in more details so that continuing to play is also bad for the player with a disadvantage, so a certain draw is a better outcome for both players than continuing to play. In general: a chess game often doesn’t in fact have the payoffs it naively appears to have, and in particular it often isn’t a constant-sum interaction.
I would guess that
Most draws are by agreement, but my guess is that a good chunk of these are from positions which are very drawish (in particular, often from endgame positions which are known draws).
Anyway, I agree that a bunch of draws are by agreement from a position with many pieces on the board, and it’s interesting why these exist.
I think it’s often about conserving resources as Kennaway said (mostly for other games in the same event). I think it’s occasionally about just not feeling like playing. I think it’s also reasonably commonly about effective utilities of outcomes not matching 0⁄0.5/1 for other reasons. Like, it’s easy to set up tournament situations in which a certain draw is better for a player than eg winning with probability
eg because it’s the candidates tournament and you want to qualify for the world championship match