BBO Discussion Forums: Fixed by RNG - BBO Discussion Forums

Jump to content

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

Fixed by RNG

#1 User is offline   1eyedjack 

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

Posted 2014-October-19, 10:54



MP, Best Hand South, 25c Robot tourney, Pro GIB 30

Somewhat lazily I played by doubleton Hearts in order low-high, at tricks 1 and 2. A third Heart followed, limiting me to 9 tricks.

At another table where the auction was identical, the first two tricks were identical except that declarer echo'd in Hearts, presumably to tell dummy that he had a doubleton. At that point East, who had exactly the same information as at my table (all missing Hearts now accounted for, and no helpful discards), decided to play a Club. Declarer rose with the Ace for 10 tricks.

This was not an Instant tourney where the excuse could be put down to a different version of GIB at each table. Clearly the collection of simulated deals generated by East in each case differed by reason of the order in which the Heart 3 and Heart 4 were played, despite that they are equal cards.

It may be a tall order to ask that GIB runs its sims identically in these situations, but I feel compelled to comment that in my opinion the current setup is undesirable.
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

#2 User is offline   uva72uva72 

  • PipPipPipPip
  • Group: Full Members
  • Posts: 245
  • Joined: 2014-June-09

Posted 2014-October-19, 18:41

Non-material "false-carding" of this type can, unfortunately, have a major impact on results. I have reported similar instances, agree that this is completely unsatisfactory and strongly urge that the issue be addressed. Simulations should not produce different results when identical information is available.
0

#3 User is offline   steve2005 

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

Posted 2014-October-19, 19:32

You cant save hand simulations across tables. I believe it violates the rules of bridge. (Program would in effect know what happened at other tables)

I suspect the problem is the small number of simulations run.
The more simulations run the better and more consistant results you get.
Sarcasm is a state of mind
0

#4 User is offline   1eyedjack 

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

Posted 2014-October-20, 12:40

View Poststeve2005, on 2014-October-19, 19:32, said:

You cant save hand simulations across tables. I believe it violates the rules of bridge. (Program would in effect know what happened at other tables)

I suspect the problem is the small number of simulations run.
The more simulations run the better and more consistant results you get.



I don't think that you have really understood the point.

GIBs of the same version in the same tourney *WILL* behave absolutely identically at different tables given identical starting conditions at the point of measurement. They start with the same random number seed, run the same number of simulations, which generate an identical population of hands, from which its algorithm will then derive an identical conclusion as to optimal line.

Currently the "identical starting conditions" mentioned above include the exact identical bidding and identical play up to the decision point. There is no "saving hand simulations across tables", whatever you mean by that. It simply runs the simulations independently at each table but inevitably comes up with the same answer, as a direct consequence of how random number generation works in computers. I don't see that as a violation of the rules of bridge. No GIB at any one table is aware, when running the sim, that there even has been (or will be) any other table in the tourney that precisely replicates the action at its own table.

All that I am asking is that the process of simulation should be varied so as to extend the principle of generating an identical population of sample hands where there have been certain minor deviations in previous play that can have no impact on the remaining hidden cards.

I am not saying that it would be easy to program. I am quite certain that it would be one of the more difficult requests to honour. I believe that it is within the realms of possibility. I was not expecting anyone to post that it would be actually undesirable.






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

#5 User is offline   steve2005 

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

Posted 2014-October-20, 12:48

If they are not producing the same results from simulation I suspect your premise that they are using the same seed could be wrong. I have no knowledge of how GIB picks it's seed but when I programed I multiplied a base seed by a function of the lower decimals of the time clock to randomize the seed even more.

Obviously they're doing something to randomize the seed or the simulations would give the same result.





Sarcasm is a state of mind
0

#6 User is offline   Bbradley62 

  • PipPipPipPipPipPipPipPip
  • Group: Advanced Members
  • Posts: 6,542
  • Joined: 2010-February-01
  • Gender:Male
  • Location:Brooklyn, NY, USA

Posted 2014-October-20, 13:18

View Poststeve2005, on 2014-October-20, 12:48, said:

Obviously they're doing something to randomize the seed or the simulations would give the same result.
No. All GIBs (in a tournament) start with the same seed, so that if exactly the same thing happens at multiple tables, multiple GIBs will take the same actions. (We have been told this repeatedly, and no one has ever posted evidence to the contrary.) However, "exactly the same thing" includes that the human must play his cards in exactly the same order. In the hand posted here, EW start by cashing two top hearts. Human South can do one of two things: (a) he can play 4 on the first one, then 3 on the second, or (b) he can play 3 then 4. At all tables where Human South chose (a), East behaved consistently; at all tables where Human South chose (b), East behaved consistently. This is because all GIBs started with the same seed. OP's (valid, imo) point is that Human South's selection of (a) vs (b) should not be enough to cause East to make different decisions, since after two tricks all tables are at exactly the same point, regardless of how they got there. But, apparently, GIB's decision tree is constructed such that once there has been a variation, the branches don't re-merge. Unfortunate, but I suspect that fixing this is beyond the plans of BBO.
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