BBO Discussion Forums: Appalling declare play by the Robot - BBO Discussion Forums

Jump to content

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

Appalling declare play by the Robot

#1 User is offline   1175 

  • PipPipPip
  • Group: Full Members
  • Posts: 76
  • Joined: 2024-May-10

Posted 2024-May-28, 03:10

On one hand in the duplicate IMPs, I wondered how I could have lost IMPs (less than one, but the board seemed incredibly flat) on a particular hand. Sadly, I did not take contemporaneous notes on this hand (including variations in the auction) - I have only a screenshot of the position after eleven tricks (I played this before I registered for an account). I recall every table in 5 by East, and the contract easily making at every table except for one. What happened at that table actually shocked me. This occurred on board 10 (all vulnerable with East as dealer), and with East on lead after eleven tricks, with East-West having taken nine tricks up to that point, and North-South two tricks.

         S
         H 7
         D 2
         C

S               S
H               H Q
D KJ            D 5
C               C

        S
        H
        D QT
        C


As you can surmise (only one way exists to go down), the Robot (at this table) led the D5 to the DK, and conceded a diamond. This obviously makes no sense to me (I can't wait to see the explanations of why the robot did this :) ).

I have a couple of only somewhat related questions.

First, does some way exist in the hand editor to create a position like this (later in the hand) with the exact auction and card play up to this point forgotten? Just the remaining cards, but showing the board number, vulnerability, dealer, contract, number of tricks taken by each side, and player on lead.

Second, what does it cost to rent an "upgraded" Robot, and where can you use one (only for tournaments, or for something like the duplicate IMPs)? It seems no way exists to see the prices without actually depositing money. Obviously, I would expect even the most basic Robot not to make mistakes like this. :)
0

#2 User is offline   mycroft 

  • Secretary Bird
  • PipPipPipPipPipPipPipPip
  • Group: Advanced Members
  • Posts: 7,346
  • Joined: 2003-July-12
  • Gender:Male
  • Location:Calgary, D18; Chapala, D16

Posted 2024-May-28, 10:54

Click on (the name line of) each hand when you want to move to the next one.

Or, of course, since it is text once inserted into the post
[hv=pc=n&s=shdqtc&w=shdkjc&n=sh7d2c&e=shqd5c]399|300[/hv]
build a full one and delete the cards from the text.
When I go to sea, don't fear for me, Fear For The Storm -- Birdie and the Swansong (tSCoSI)
0

#3 User is offline   smerriman 

  • PipPipPipPipPipPipPip
  • Group: Advanced Members
  • Posts: 3,992
  • Joined: 2014-March-15
  • Gender:Male

Posted 2024-May-28, 13:57

I expect South had shown out of hearts and something in the bidding (or the lead, like last time) convinced East that North must have had an extra 2 HCP, no matter what. It's a classic GIB issue to not distinguish 100% lines from lines that are only 100% based on its assumptions.

Yes, with a 3 card ending it is entirely trivial to generate all 20 arrangements of the unknown cards, and at least break ties on its original calculations based on all 20. But GIB doesn't do a lot of trivial things; even with advanced robots I've seen cases where it simulated nearly 300 deals of 3 card endings and it went wrong as all 300 were identical, because its assumptions were too strong when influencing the dealer algorithm, even when it is meant to allow some level of mismatches.

That is one thing that Lorand was looking at fixing in this mystical new version.

Any paid tournament uses advanced robots; but if you go into Casual and click 'Start a table 3 with robots' it will tell you the rental price; I forget what it is currently, though it has skyrocketed over the last few years.
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