Automatic draw recognition?

What do the players think about implementing automatic draw recognition for Twixt on this server? When a game reaches the point where neither side could win even if all vacant holes other than opposing border holes were filled with that player's pegs, I believe the server should stop the game and declare a draw. TwixtBot can recognize a draw, and LG could also.

Richard what do you think? Thanks.

I think that is a fine idea.

Gets my vote. Richard?

I think it's a good idea.

I think it's a good idea too, but it could be a heavy load if the server had to check for this after every move. Maybe there could be a "claim draw" option for this? When claiming a draw, the opponents holes are immidiately filled with pegs, which of course either results in a draw or the opponent winning. This would cause a smaller load on the server, while serving most of the same purpose, I think.

I think it's a fine idea.

Marius, your idea might work as long as the user sees a warning about what will happen if they go through with this claim draw option. Maybe the screen display could show all opposing holes filled with opposing pegs, but the user could still retract the claim. If the position will reach a draw with best play, but it hasn't gotten there yet, and your opponent does not accept your draw offer, your only recourse is to play the game out until both players are completely blocked, and only then claim the draw.

Yeah, that's probably a good idea. You get to see all the pegs placed before pressing "Send move". And it should only be a legal move to send if it really IS a draw. So it actually becomes more of a "check for draw" rather than "claim draw", I suppose.

At one Twixt move every (roughly) 200 seconds or so, I wouldn't worry about a check that surely takes no more than a fraction of a second.

Once again, Ruby hackery to the rescue!

Or not ... Once again, low numbers.

There have only been 6 absolute draws that would have been detected by the proposed automatic draw recognition algorithm on the server.

The last 4 of them involved TwixtBot. The wrong result was recorded in only 2 of these games, both due to TwixtBot's early non-recognition of draws.

240489, 335000 - not involving TwixtBot, but marked correctly as draws.

986600, 987446 - involving Twixtbot, NOT marked correctly as draws (the human player, World Traveller both times, gave up and resigned).

1025191, 1144239 - involving TwixtBot, but marked correctly as draws.

I still think it's a good idea, for completeness and quality.