BBO Discussion Forums: Does GIB not have any bidding heuristics? - BBO Discussion Forums

Jump to content

Page 1 of 1
  • You cannot start a new topic
  • You cannot reply to this topic

Does GIB not have any bidding heuristics? Because if it does then this must be a bug

#1 User is offline   wbartley 

  • PipPipPipPip
  • Group: Full Members
  • Posts: 289
  • Joined: 2010-July-23

Posted 2016-July-21, 16:08

GIB (a Just Play Bridge robot) is North and holds:

KQT84 K T942 984

South deals and the auction goes

1-3-3-Pass
3-Pass-4-Pass
4NT-Pass-5-X
6-All Pass!

Surely after South has bid two suits naturally with the opponent preempting in clubs and GIB holding three clubs, no reasonable heuristic would allow for a pass here.
0

#2 User is offline   1eyedjack 

  • PipPipPipPipPipPipPipPip
  • Group: Advanced Members
  • Posts: 6,575
  • Joined: 2004-March-12
  • Gender:Male
  • Location:UK

Posted 2016-July-21, 17:07

I suppose it depends how you define "heuristic", which normally refers to some form of learning from experience. Gib certainly does not do that. It does, sometimes, depending on circumstance, run simulations, whereby it generates a population of possible hidden hands, runs double dummy analyses on those selected hands, which then influences its decision. But even that takes no account of a memory of hands that have gone before.

Furthermore, the simulations in bidding only ever kick in with "advanced" Gib which cost (extra). If you are playing "just play bridge" then you are using the "basic" version that makes no pretence at heuristics or sims in the bidding and relies entirely on defined rules. If that rule has not been defined for the situation (or if there is a bug in the definition) it all falls apart.

Here I have no idea what was meant by 6C but it seems likely to be a rule-based bug. Thereafter Gib often defaults to Pass when it has gone off the rails (in rule-based mode)
Psych (pron. saik): A gross and deliberate misstatement of honour strength and/or suit length. Expressly permitted under Law 73E but forbidden contrary to that law by Acol club tourneys.

Psyche (pron. sahy-kee): The human soul, spirit or mind (derived, personification thereof, beloved of Eros, Greek myth).
Masterminding (pron. mPosted ImagesPosted ImagetPosted Imager-mPosted ImagendPosted Imageing) tr. v. - Any bid made by bridge player with which partner disagrees.

"Gentlemen, when the barrage lifts." 9th battalion, King's own Yorkshire light infantry,
2000 years earlier: "morituri te salutant"

"I will be with you, whatever". Blair to Bush, precursor to invasion of Iraq
0

#3 User is offline   wbartley 

  • PipPipPipPip
  • Group: Full Members
  • Posts: 289
  • Joined: 2010-July-23

Posted 2016-July-21, 17:35

Heuristic does not imply learning. Heuristic would mean any kind of logic for deciding what to bid that isn't a wrote decision tree. An example of a heuristic would be GIB answering the question can I pass this bid by ascertaining whether the partnership has agreed on a trump suit (in this case, no), if the partnership has a known 9 card fit in a suit other than the last bid suit (in this case, yes), or if the partnership, based on the bidding is likely to have less than a 6 card fit in the last bid suit (in this case, yes). So by two heuristic measurements out of the three I mentioned, GIB should not consider pass a viable action.

Of course, it could also check if the last bid suit is neither the first nor the second suit bid naturally by its partner and if one of the opponents has shown extreme length in that suit which heuristically would indicate that partner's last bid is some sort of cue bid.

I've been led to believe that GIB does use heuristics in the form of Monte Carlo hand generation combined with double dummy analysis. I don't know if any of that is true. If it were true then what better time to employ such methods than when GIB has "gone off the tracks" as far as its wrote decision tree is concerned. It could easily generate a set of hands matching partner's initial two bids, maybe even include the opponent's bid (after all, GIB is "off the tracks"). It seems likely that any such analysis would indicate that the double dummy result of a 6C contract is not statistically very good as compared to say, a 6D contract.
0

#4 User is offline   steve2005 

  • PipPipPipPipPipPipPip
  • Group: Advanced Members
  • Posts: 3,148
  • Joined: 2010-April-22
  • Gender:Male
  • Location:Hamilton, Canada
  • Interests:Bridge duh!

Posted 2016-July-21, 18:58

yes the default to pass when confused is ridiculous . should default to cheapest known fit if that is pass fine.
Sarcasm is a state of mind
0

Page 1 of 1
  • You cannot start a new topic
  • You cannot reply to this topic

1 User(s) are reading this topic
0 members, 1 guests, 0 anonymous users