Re: [Bug-gnubg] gnubg Offered Odd Backgammon Resignation.

[Ian Shaw]

Hi Ralph, 

I got an error message when I pasted your 9wAAAGzfSQgAAA:MAvgAEAAQAAE position 
into gnubg.

Illegal match ID 'MAvgAEAAQAAE'
Dice 0 0, player on roll 0 (turn 1), resigned 0,
doubled 0, cube owner -1, crawford game 0,
jacoby 0,
match length 7, score 4-8, cube 1, game state 3

Looking at the score (4-8) It looks like the match is over.

I've just tested NwAAAGoSYj8AAA:EQEAAAAAGAAA with the a certain gammon and the 
score 3-0 to 7, holding a 2-cube. Gnubg resigned a gammon.

I put a few gnubg men back in my inner board NwAAAGoSYo8CAA:EQEAAAAAGAAA and 
gnubg didn't resign a backgammon. It played on until the backgammon was saved, 
then resigned a gammon.

I think the original design idea for resignation was to  avoid any accusations 
that gnubg was not giving full value when resigning. 

It's meaningless to keep any overall tally of points in matches. A backgammon 
match is not like a football match, where goal difference might count at the 
ned of a season. Gnubg won't try to win or avoid a meaningless backgammon if it 
means giving up chances to win the match. Nor will it try to reduce a winning 
margin from, for example, 7-2 to 7-3; it will always to everything in its power 
to win the match, however large the final margin.

If you keep a tally that records points-difference in matches, you will be 
getting misleading information about your performance. If you are trying to 
maximise points difference and gnubg isn't, then you aren’t really playing the 
same contest.

It only makes sense to look at overall score in money (unlimited) games.

I hope this helps.

-- Ian 

-----Original Message-----
From: Ralph Corderoy [mailto:address@hidden 
Sent: 29 April 2017 19:32
To: Ian Shaw <address@hidden>
Cc: address@hidden
Subject: Re: [Bug-gnubg] gnubg Offered Odd Backgammon Resignation.

Hi Ian,

> I'm guessing that it's something to do with the match score that 
> affected how gnubg offers a resignation. For example, maybe 2 points 
> was enough for you to win the match anyway, so it didn't matter 
> whether gnubg conceded 2 or 3.

I think you're right, but that it is wrong for gnubg to do this.  Yes, it 
doesn't affect the match, but it does affect the overall points, if some 
external tally is being kept.  More importantly, it confuses novices, like me.  
"How could I have backgammoned it from that position?
What am I missing?  I only managed a gammon."

Since I'm sure gnubg knows I'm going to gammon it at best, it should only offer 
a 2 resignation.  Is this the right place to make that argument, or do I need 
to open an issue?

I poked around the saved SGF file for the game and found that it records cube 
actions, but not the offer of a resignation, whether it was 1, 2, or 3, and 
that it was rejected.  I looked because we here are uncertain on which of two 
board positions gnubg offered the resignation and expected the saved file to 
say.

Similarly, when a resignation is accepted, the panel showing the moves says 
"Resign" at the end, but not a qualifying "gammon", for example.

Since I can't pinpoint the backgammon resignation offer, I took more care the 
next time it occurred.  Position 9wAAAGzfSQgAAA:MAvgAEAAQAAE was 6-4 to me in a 
7 match, and gnubg offered a 2 resignation.

Once last thing, since I'm making suggestions...  When gnubg moves a counter 
one point, to an empty one, and it was the only thing on the old point, then it 
could avoid making such a long sharp arcing loop, but a squatter one instead.  
It would take long enough to be seen, but not too long when its destination it 
typically clear from the off.

--
Cheers, Ralph.
https://plus.google.com/+RalphCorderoy