BBO Discussion Forums: Inconsistent robot plays - BBO Discussion Forums

Jump to content

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

Inconsistent robot plays Robots playing differently in same contract and lead

#1 User is offline   mjs4321 

  • Pip
  • Group: Members
  • Posts: 2
  • Joined: 2020-January-01

  Posted 2020-January-01, 12:13

Playing on the masterpoints boards today, I lead a singleton club taken by my robot partner who led a diamond trump back! Contract won by opposition. Checking the other boards I found another robot led back a club which his partner trumped . Contract defeated. How can one robot differ from another with identical contracts and leads?
0

#2 User is offline   hrothgar 

  • PipPipPipPipPipPipPipPipPipPipPip
  • Group: Advanced Members
  • Posts: 15,396
  • Joined: 2003-February-13
  • Gender:Male
  • Location:Natick, MA
  • Interests:Travel
    Cooking
    Brewing
    Hiking

Posted 2020-January-01, 14:00

View Postmjs4321, on 2020-January-01, 12:13, said:

Playing on the masterpoints boards today, I lead a singleton club taken by my robot partner who led a diamond trump back! Contract won by opposition. Checking the other boards I found another robot led back a club which his partner trumped . Contract defeated. How can one robot differ from another with identical contracts and leads?


1. Were you playing with Advanced or Basic bots?

Both advanced and basic bots use simulations to decide how they will play. In turn, this means that they can adopt different lines of play. The advanced bots use the same seed to generate the hands for their simulations. In turn, this means that i the advanced bots are presented with an identical auction and identical opening lead, they should behave identically. The basic bots do not share a seed and as a result, they can make different decisions.

2. If you were playing with Advanced bots, I suspect that there was some difference in the auction that caused the two defenders to take a different view.
Alderaan delenda est
0

#3 User is offline   mjs4321 

  • Pip
  • Group: Members
  • Posts: 2
  • Joined: 2020-January-01

Posted 2020-January-01, 15:48

Thank you. I was hoping that, by playing duplicate in a robot environment, it would eliminate the human randomness/errors and so I would be competing with others under identical conditions. Oh well. Thanks again for your prompt reply.
0

#4 User is offline   zhasbeen 

  • PipPipPipPip
  • Group: Full Members
  • Posts: 240
  • Joined: 2017-September-22
  • Gender:Male
  • Interests:bridge, running, spectator sports, excel, keeping track of all kinds of stuff

Posted 2020-January-01, 20:27

View Postmjs4321, on 2020-January-01, 15:48, said:

Thank you. I was hoping that, by playing duplicate in a robot environment, it would eliminate the human randomness/errors and so I would be competing with others under identical conditions. Oh well. Thanks again for your prompt reply.


While I have been openly critical of GIB, and continue to point out problems, this is one area I would not worry about. In fact, nothing could be fairer from a competitive standpoint than an ACBL BBO Individual Robot Tournament. All humans who enter a tournament have the same partner, while competing against the same opponents, and playing the exact same convention card (GIB 2/1). It all boils down to card play and decision making. By card play I mean both as declarer and defender, although you will probably be declarer more than 70% of the time. All the other noise is factored out.
How can you get any fairer than this?

Every time I've checked to see why I got an unfavorable lead despite being in the same contract as most other pairs, there has been a slight variance in my auction. This phenomena is not unique to robot tournaments, btw.
Tournament results tend to run true to form so far as players with better credentials having the most success. With enough games in the sample there is nothing random about the results. For example, BBO robot players Finn Kolesnik and Jim Munday were #2 and #3 finishers in the most recent Blue Ribbon Pairs earlier last month in San Francisco.

While playing against GIB and friends is bound to frustrate you, I believe that playing can help your game. The hands are challenging. They make you think, and you get to play lots of them in a relatively small amount of time. Most of us need plenty of repetition to improve and this is a good way to get it. The movie replays are a big plus.
0

#5 User is offline   thomas c 

  • PipPipPipPip
  • Group: Full Members
  • Posts: 105
  • Joined: 2012-June-27
  • Gender:Male
  • Location:mississippi
  • Interests:golf, bridge, opera

Posted 2020-January-03, 17:18

i pretty much agree. the bos never forget and falsecard constantly. they can execute squeeze plays, end plays and dup high cards to keep from being endplayed. are there tricks
you can play on them? sure. but there are tricks you can play on human declarers and defenders that work more often.

it is prob better to play conservatively but thats not always fun. also the acbl rournaments are prob a better test of your ability( 12 boards)



as i said before i play 1 to 3 bot tournaments a day.

tom
0

#6 User is offline   nullve 

  • PipPipPipPipPipPip
  • Group: Advanced Members
  • Posts: 2,237
  • Joined: 2014-April-08
  • Gender:Male
  • Location:Norway
  • Interests:partscores

Posted 2020-January-04, 14:28

View Postmjs4321, on 2020-January-01, 12:13, said:

Playing on the masterpoints boards today, I lead a singleton club taken by my robot partner who led a diamond trump back! Contract won by opposition. Checking the other boards I found another robot led back a club which his partner trumped . Contract defeated. How can one robot differ from another with identical contracts and leads?

The auctions may have been different, for example.
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