[thorvald]

Follow the play and notice that at trick 10 dummy is high except for the last ♠.

Still West drops ♠A

---

And yes it was a terrible declarer play that lead to that ending :-)

[smerriman]

Another example of this. Some of the simulated deals are saying the spade ace beats the contract by one trick.

[thorvald]

smerriman, on 2022-August-19, 16:17, said:

Another example of this. Some of the simulated deals are saying the spade ace beats the contract by one trick.

It could be interesting to see the deals where the simulation shows discarding ♠A defeats the contract as declarer have 6 tricks in the bag and 3 tricks at dummy

[smerriman]

thorvald, on 2022-August-19, 16:58, said:

It could be interesting to see the deals where the simulation shows discarding ♠A defeats the contract as declarer have 6 tricks in the bag and 3 tricks at dummy

See the thread I linked to - there isn't any simulated deal where it defeats the contract, but it occasionally miscalculates the double dummy result for unknown reasons.

deal 3:
                    S  7
                    H  ---
                    D  J
                    C  T
S  A Q T 9                              S  8
H  ---                                  H  ---
D  ---                                  D  4 3
C  ---                                  C  ---
                    S  K J 6
                    H  ---
                    D  ---
                    C  ---

West to play to DK, D6, S4; no trumps
N/S have taken 6 tricks
SA: 50? SQ: -400 ST: -400
deal 4:
                    S  7
                    H  ---
                    D  J
                    C  T
S  A Q T 9                              S  8
H  ---                                  H  ---
D  ---                                  D  4 3
C  ---                                  C  ---
                    S  K J 6
                    H  ---
                    D  ---
                    C  ---

West to play to DK, D6, S4; no trumps
N/S have taken 6 tricks
SA: -430 SQ: -400 ST: -400

Unlike the other case where about half were wrong, this time only about 1 in 10 deals are wrong.. but when those 1 in 10 result in +10 IMPs vs -1 IMP, the bug means it's basically a coin flip as to whether to throw the ace or not.

[thorvald]

smerriman, on 2022-August-19, 17:01, said:

See the thread I linked to - there isn't any simulated deal where it defeats the contract, but it occasionally miscalculates the double dummy result for unknown reasons.

deal 3:
                    S  7
                    H  ---
                    D  J
                    C  T
S  A Q T 9                              S  8
H  ---                                  H  ---
D  ---                                  D  4 3
C  ---                                  C  ---
                    S  K J 6
                    H  ---
                    D  ---
                    C  ---

West to play to DK, D6, S4; no trumps
N/S have taken 6 tricks
SA: 50? SQ: -400 ST: -400
deal 4:
                    S  7
                    H  ---
                    D  J
                    C  T
S  A Q T 9                              S  8
H  ---                                  H  ---
D  ---                                  D  4 3
C  ---                                  C  ---
                    S  K J 6
                    H  ---
                    D  ---
                    C  ---

West to play to DK, D6, S4; no trumps
N/S have taken 6 tricks
SA: -430 SQ: -400 ST: -400

Unlike the other case where about half were wrong, this time only about 1 in 10 deals are wrong.. but when those 1 in 10 result in +10 IMPs vs -1 IMP, the bug means it's basically a coin flip as to whether to throw the ace or not.

I am reading the old post (Just found out that I had a filter so only pinned and recent post was shown), and as many other I am impressed by your work. It would be very intreresting to get that error fixed. What an improvement that would be for all of us

[thorvald]

How do you generate the output showing the analysis by GIB?

I have search for the command line input to bridge.exe but found none.

From Lorand Dalis AI-project I have: https://github.com/l...cripts/data/gib


but the old document engine.txt that might contain the information is not available on Way back Machine