Re: Naught gnubg

On Wed, 4 Mar 2009 05:19:55 -0800 (PST), pauldepstein@xxxxxxx wrote:

On Mar 4, 9:56 am, David C. Ullrich <dullr...@xxxxxxxxxxx> wrote:
On Tue, 3 Mar 2009 05:43:52 -0800 (PST), pauldepst...@xxxxxxx wrote:
On Mar 1, 8:15 pm, bob <bob_k...@xxxxxxxxxxx> wrote:
On Feb 27, 7:26 am, pauldepst...@xxxxxxx wrote:

On Feb 26, 10:13 pm, Walt <walt_ask...@xxxxxxxxxxxxxx> wrote:

David C. Ullrich wrote:
gnubg offered to resign a single game in the following
position. I suppose this doesn't count as officially a bug,
since a player is free to make whatever offer he feels
like at any time. But surely it's not what the authors
intended. (And of course positions like this are not that
common, but still...)

 >  GNU Backgammon  Position ID: AwAAuD19AAAAAA
 >                  Match ID   : ASkAAAAAEAAA
 >  +12-11-10--9--8--7-------6--5--4--3--2--1-+     O: gnubg (Cube: 2)
 >  |                  |   | O  O     O  O  O |     0 points
 >  |                  |   | O        O  O  O |     On roll
 >  |                  |   | O        O     O |
 >  |                  |   | O        O       |
 >  |                  |   | O                |
 > ^|                  |BAR|                  |
 >  |                  |   |                  | XX
 >  |                  |   |                  | XX
 >  |                  |   |                  | XXX
 >  |                  |   |                X | XXX
 >  |                  |   |                X | XXX 2 points
 >  +13-14-15-16-17-18------19-20-21-22-23-24-+     X: user
 >

What's the match length?  It's not an incorrect resignation if it's a
match to 4.
Besides David's comment about the indication-by-omission-of-match-
length, it could only be a match to 4 if one of the players had made
at least one enormous cube blunder.  If a match to 4, one would ask
why gnubg didn't do an automatic redouble.

Paul Epstein

   If there is no market loss possibility then the redouble to 4 is
optional. It is not necessarily the case that a cube error was made.

Bob Koca

But gnubg is an automaton.  Surely it redoubles without wasting time
checking that there aren't market losers.

??? It may not "understand" the _concept_ of market losers, but
it doesn't follow that market losers are irrelevant to the way it
plays. If a position has the property that you're better off not
redoubling now, whether because of non-existence of market
losers or any other reason, then that position has the property
that your equity, (or I suppose MWC in match play) is larger
if you don't redouble. gnubg tries to make the play that
maximizes its equity - concepts like "market losers", etc,
are just tools we humans use to try to do the same thing.

David C. Ullrich

"Understanding Godel isn't about following his formal proof.
That would make a mockery of everything Godel was up to."
(John Jones, "My talk about Godel to the post-grads."
in sci.logic.)

I was talking about the mandatory-double situation where doubling is
clearly optimal (because the cube's value already gives your opponent
the match [ignore gammons now for simplicity]) but where holding may
also be optimal in the absence of market-losers.

Ok. Yes, in that situation you may well be right, there's no reason
it's going to sit around looking for other optimal plays once
it's found an optimal one.

I'm sure that, in
this situation, bots redouble rather than checking against all 441
immediate sequences to check that there aren't any market losers and
that the hold is also ok.

I think confusion arose because the discussion started when I gave the
view that your position would not have arisen in a match-to-4, so I
was looking at that match score in particular. You seem to have
thought I was talking about a much more general situation. I agree
with you that _if_ I were addressing the general situation, my comment
would have been wrong.

Bob Koca said that even in a match-to-4 played correctly, your
position could have arose. My view (which doesn't exactly contradict
Bob) is that the position would be extremely unlikely in a match-to-4
scenario. Part of the reason I deem it extremely unlikely is that I
can't imagine gnubg holding the 2 cube. (Actually, I can imagine it.
That use of "imagine" is a figurative way of expressing my certainty.)

Paul Epstein

David C. Ullrich

"Understanding Godel isn't about following his formal proof.
That would make a mockery of everything Godel was up to."
(John Jones, "My talk about Godel to the post-grads."
in sci.logic.)
.

Relevant Pages

  • Re: Naught gnubg
    ... since a player is free to make whatever offer he feels ... why gnubg didn't do an automatic redouble. ... checking that there aren't market losers. ... "Understanding Godel isn't about following his formal proof. ...
    (rec.games.backgammon)
  • Re: Naught gnubg
    ... since a player is free to make whatever offer he feels ... why gnubg didn't do an automatic redouble. ... checking that there aren't market losers. ... thought I was talking about a much more general situation. ...
    (rec.games.backgammon)
  • Re: My talk about Godel to the post-grads.
    ... at all) that a term can play a role even though it plays no ... Remember the definition of a union? ... It is a redundant sign. ... "Understanding Godel isn't about following his formal proof. ...
    (sci.logic)
  • Re: Is it correct to redouble when you are about to lose a gammon and the match with it?
    ... one general rule would be you need to have Market Losers. ... you have no shots, you don't need to redouble yet. ... GNU Backgammon Position ID: d7cNAAC2bYEBYA ...
    (rec.games.backgammon)
  • Re: Naught gnubg
    ... plays once it's found an optimal one. ... what you say is meaningless bullshit. ... It's nonsense to say that's nonsense. ... "Understanding Godel isn't about following his formal proof. ...
    (rec.games.backgammon)