Author: Robert Hyatt
Date: 07:06:12 11/06/99
Go up one level in this thread
On November 06, 1999 at 05:55:44, Amir Ban wrote:
>On November 05, 1999 at 20:19:19, Bruce Moreland wrote:
>
>>On November 05, 1999 at 17:40:59, Amir Ban wrote:
>>
>>>
>>>One benefit of the ban-crafty debate of a few days ago is that I learned how to
>>>use the ICC 'search' command and discovered that it gives much more information
>>>than I thought it did.
>>>
>>>I went looking for the game in which crafty timed out when ban had insufficient
>>>material. I found it, very easily, and it's listed below. It took place on
>>>October 7th, i.e. 4 weeks ago.
>>>
>>>If you add up the crafty times (or easier, use ICC 'examine'), you will see that
>>>crafty had 2:19 minutes on the last move, but did not make a move till time ran
>>>out.
>>>
>>>The first reaction to this by Bob & others was that I was making this up. The
>>>second reaction was that crafty crashed or hung, so I looked for the game crafty
>>>played next, and I discovered that ban started another game with crafty 15
>>>minutes after this game started, meaning immediately after this game ended. So
>>>crafty did not crash or hang.
>>>
>>>I include below the ICC commands I typed in (in quotes), and the ICC output
>>>including the game moves & information.
>>>
>>>I tried messaging Bob on ICC to tell him that, but this is impossible as he's
>>>censoring me. I assume he has interest in this.
>>
>>I don't understand why you bring this up again.
>>
>
>Ok, I will elaborate some, but you have to promise to keep KarinsDad from
>jumping all over me again.
>
>The first reason is because I've been told several times that I'm lying for
>telling about this game.
>
>The second reason is to help the author to track down and analyze this problem,
>to prevent it from happening again (specially against me), to point to a bunch
>of similar behavior that have in common that they are annoying to the opponent,
>and to hope that the program author will take a less indifferent attitude to
>them. This is more in line with my original reason for jumping in these threads,
>and I admit that to continue with it now is of not much use since it backfired
>and I seem to have made a bad job of explaining it. I've explained most of it in
>http://www.icdchess.com/ccc/message.html?76120, i
>
point by point:
(1) the author does not take an "indifferent attitude" to losing on time, or
to not moving and drawing on time. Fortunately, xboard sends a "{Result
1/2-1/2 Crafty ran out of time ...}. I have an automatic script that _always_
looks for those, as they indicate a program crash that needs attention.
(2) I do not write code that is either intentionally (or unintentionally)
abusive in any way. So far as I know, Crafty was the first program that offered
and accepted draws on ICC. And the 'threshold' for doing this varies by player
so that it is 'very easy to get along with'. Of course, it has to know that the
position is drawn, and it isn't perfect there.
(3) I use ICC as my testbed. I test all changes carefully before putting them
up on ICC, but occasionally I have problems. I have had network problems
(local) that allowed Crafty to accept but not play games. I have introduced
learning problems where learning would promptly crash the engine and cause it
to lose on time.
_none_ of that is intentional. I play on ICC to test, but I also play to _win_
whenever possible. And I play to encourage the strong humans to keep coming
back.
>
>>Here is my viewpoint. You are sitting there playing chess, the game is a draw
>>and you are a pawn down. Your opponent, sits there and flags rather than
>>offering a draw or playing on, which is scored as a draw because you have
>>insufficient material. Possibilities:
>>
>>1) The program had some software difficulty and stopped making moves.
>>
>
>No doubt. Also, this problem, in isolation, is meaningless. It's hardly even the
>main thing I'm complaining about (I listed it last in my list. The main problems
>I see are refusal to draw and refusal to adjourn). I'm sure such problems are
>created innocently, but they stay there because of indifference to things that
>don't cause direct harm to the program, unlike other problems, which get hunted
>and fixed right away.
>
I don't have any control over 'adjourn'. xboard automatically rejects all
adjourn requests, with crafty having no say-so about it. Crafty never requests
an adjourn, although ICC will simulate such a request if lag goes over 2 mins
or so.
And _no_ problem gets treated indifferently by me. I go after _all_ problems,
because so many run crafty on ICC a small problem is really a big one.
>
>>2) The operator deliberately threw a wrench into his own program's spokes in
>>order to waste two minutes of your time, then lied about it later.
>>
>
>The program's automatic so this didn't happen.
>
>
>>Sorry, but I gotta go with option 1. If Bob had wanted to screw you over it
>>would have made much more sense to simply grab the newspaper and head off to the
>>bathroom. You would have had to sit there making manual moves in a pointless
>>game, which would have annoyed you for a longer period and would have achieved
>>the same result.
>>
>
>Not in this game, you'll notice. Black loses the knight next move, and it's
>possible to end the suffering in 3-4 moves, if the opponent agrees to make a
>move, that is.
>
>Making manual moves in a pointless game happens very often in games against
>crafty when the program functions flawlessly, becuase it will never take a draw.
>Actually the game is not completely pointless, because you may always lose on
>time, as happened in, e.g. the last game crafty flagged ban. This rarely happens
>to a smooth operator, but it's a pain. Remember that crafty not only drags out
>drawn games to silly lengths, but there's no escape from them: Crafty refuses
>adjudications, penalizes disconnectors, and its author is sure disconnectors are
>cheats and says so.
>
>The experience of playing crafty manually on ICC is quite a bit like arriving
>first day to boot camp.
>
Then _why_ do it?
>
>>If you want to find reason to accuse your opponent of being evil, this is very
>>thin.
>>
>
>I'm not accusing my opponent of being evil, and I never did. Rather indifferent
>about things his program does that are not in its disinterest, and too haughty
>to be bothered with it.
>
>
>>I found something interesting with "search", too. "search crafty ban mode=tm"
>>returns 3 games. Did you look at the third one? It's an old game back from
>>when you were automatic, and it is very similar to this one, with the situation
>>reversed, and only 33 seconds of dead time rather than 2:19.
>>
>
>This is from 3 years ago when Shay was debugging the interface. It was hanging
>left and right. Notice the game was played unrated for testing purposes. He
>didn't manage to find the problem so at the time the automatic interface was
>shelved soon after this. He looked specifically for automatic computers that
>will not bother playing against a program that hangs often, as I'm sure you
>remember since yours was another such opponent.
>
>It's not the same thing at all, if that's what you mean.
>
>Amir
It seems to be to me... we both "hung" due to a bug... nothing more or less..
This page took 0.01 seconds to execute
Last modified: Thu, 15 Apr 21 08:11:13 -0700
Current Computer Chess Club Forums at Talkchess. This site by Sean Mintz.