Help
Page 1 of 1
down in cold contract
#1
- Group: Advanced Members
- Posts: 6,542
- Joined: 2010-February-01
- Gender:Male
- Location:Brooklyn, NY, USA
Posted 2011-November-27, 23:16
Unnecessary play from dummy at Trick 1 causes NorthGIB to go down in a cold contract. The lack of second heart finesse also costs a trick.
#2
- Group: Admin
- Posts: 21,541
- Joined: 2004-August-21
- Gender:Male
Posted 2011-November-28, 00:59
I suspect the point of the trick 1 play was to win the trick in dummy if you were leading from Axxx, so it could take a heart finesse.
Not repeating the finesse is an old, known problem that we haven't figured out yet. Fixing problems with GIB play is difficult, since there are few rules, it's almost all simulations. The critical piece is in how it selects the hands in the simulation, and that's complicated stuff.
My guess is that it comes down to the old "double dummy" assumption. Suppose East has ♥KJ doubleton. From its perspective, it doesn't matter if it plays K or J to the first finesse, because it "knows" declarer will not take the finesse the second time. And the converse of this is that North knows that East could do this, because the way it selects hands are by comparing what was actually played with what it would consider a plausible play with that hand. GIB doesn't know about restricted choice explicitly, but I think the theory is that it should come to the same conclusion as restricted choice dictates when it does its simulations.
BTW, while I doubt GIB is capable of this coup, there are stories of famous bridge experts playing the K from KJ in situations like this. The reason is that if they won with the J, declarer would set up another suit instead of trying the Q finesse, and that alternate line was going to work. But when you win the first finesse with an unnecessarily high card, they assume that a repeat of the finesse is more likely to work. GIB can never try this because it assumes declarer is playing double dummy, so it can't be fooled.
Not repeating the finesse is an old, known problem that we haven't figured out yet. Fixing problems with GIB play is difficult, since there are few rules, it's almost all simulations. The critical piece is in how it selects the hands in the simulation, and that's complicated stuff.
My guess is that it comes down to the old "double dummy" assumption. Suppose East has ♥KJ doubleton. From its perspective, it doesn't matter if it plays K or J to the first finesse, because it "knows" declarer will not take the finesse the second time. And the converse of this is that North knows that East could do this, because the way it selects hands are by comparing what was actually played with what it would consider a plausible play with that hand. GIB doesn't know about restricted choice explicitly, but I think the theory is that it should come to the same conclusion as restricted choice dictates when it does its simulations.
BTW, while I doubt GIB is capable of this coup, there are stories of famous bridge experts playing the K from KJ in situations like this. The reason is that if they won with the J, declarer would set up another suit instead of trying the Q finesse, and that alternate line was going to work. But when you win the first finesse with an unnecessarily high card, they assume that a repeat of the finesse is more likely to work. GIB can never try this because it assumes declarer is playing double dummy, so it can't be fooled.
Page 1 of 1