Shortest and longest game you've seen?

[BFG]

Just curious what the shortest match (fewest turns) and longest match (most turns) you've ever seen are.

So far in my "mega" tourney, I've had 2 games that ended on turn 13, and one that was declared a draw after turn 250. (It would have gone to 255 if I'd let it.)

[Graham Banks]

CCRL games must have a minimum of 14 moves.
Longest game I've had for CCRL 40/40 would be somewhere around 350 moves.

[BFG]

CCRL games must have a minimum of 14 moves.

To clarify, is that 14 moves (plys), or 14 turns? If it's 14 turns, then I need to delete those 2 games and have them replayed...as I've been trying to use official CCRL rules. And out of curiosity, what's the reasoning behind the (presumed) 14-move minimum? The shortest possible game, a fool's mate, is 4 moves...

I should also mention that I've set Arena to adjudicate a draw at 250 turns, or a draw at 125 turns with both engines calculating 0 centipawn advantage...but perhaps I need to remove those restrictions as well.

[Graham Banks]

CCRL games must have a minimum of 14 moves.

To clarify, is that 14 moves (plys), or 14 turns? If it's 14 turns, then I need to delete those 2 games and have them replayed...as I've been trying to use official CCRL rules. And out of curiosity, what's the reasoning behind the (presumed) 14-move minimum? The shortest possible game, a fool's mate, is 4 moves...

I should also mention that I've set Arena to adjudicate a draw at 250 turns, or a draw at 125 turns with both engines calculating 0 centipawn advantage...but perhaps I need to remove those restrictions as well.

14 moves each side.

[BFG]

Good to know! Makes me wonder if there are any other CCRL testing rules I've missed.

[Graham Banks]

Good to know! Makes me wonder if there are any other CCRL testing rules I've missed.

If both engines think that one side has an advantage of 0.70 or higher for the entire game, the game is excluded as having an unfair opening.

[BFG]

Hmm...that makes sense; unfortunately I'm not aware of any way to review the PGN file and determine when that occurred. Is there any way to enforce that rule besides watching every game?

I've not seen any opening result in an advantage greater than 0.40 so far, but 0.70 could have happened, theoretically.

[Graham Banks]

Hmm...that makes sense; unfortunately I'm not aware of any way to review the PGN file and determine when that occurred. Is there any way to enforce that rule besides watching every game?

I've not seen any opening result in an advantage greater than 0.40 so far, but 0.70 could have happened, theoretically.

Our scripts automatically detect such games.

[BFG]

Ah! Well, I'll do what I can by hand in the meantime. Thanks for letting me know.

[Ray]

If you use the right books this rarely happens or not at all.

[BFG]

That makes sense. I've been using the "Perfect 2010" book and haven't observed it yet.

[BFG]

Here's another question...are games replayed if an engine loses due to time forfeit, and that engine did NOT crash? I've had 8 out of 4,830 games end in a time forfeit - 6 lost by Mustang and 2 by Bouquet. To me, an inability to manage time correctly should reduce an engine's ELO, so I have allowed them to stand.

In general, I'm trying to determine if there are any other rules - such as the "28ply minimum" or "<70 opening centipawn" rules - that would cause the CCRL to discard a game. Interistingly, I also had a game recently where both engines scored EXACTLY a 70 centipawn advantage after the opening - discard or no?

[Graham Banks]

We don't allow time losses.
If the game is clearly won, lost or drawn, we'll edit the result as necessary.
If the result is unclear, we'll replay.
If an engine obviously has a large problem with time losses, we don't bother testing it.

[Graham Banks]

Just had a game last 369 moves before ending in a draw by the 50 move rule.