[BunnyGo]

It's unlikely this is a simulation error, since only one of ten GIBs playing this hand got the play correct, and it was helped by the lead (different auction = helpful lead). I see no reason the correct line in diamonds shouldn't be taken (at least one of the GIBs should have produced a 4-0 trump split and seen that cashing the A first is the only way to play the suit). Is there a bug? Was it confused somehow and threw out the 4-0 splits?

There were a few GIBs that got lucky. They led the 9 of diamonds next and East covered--that just makes 2 incorrect plays on the same trick. The only GIB that made legitimately won the first trick (a club) in dummy and led small to the Ace of diamonds.

---

Edit: I don't know how many simulations GIB runs, but I figure that if it runs 100 simulations, it should have a 87% chance of making this hand.

This post has been edited by BunnyGo: 2011-June-26, 12:55

[barmar]

GIB plays the odds, it doesn't know about safety plays. This is a 5% layout, so it doesn't affect the odds much.

[BunnyGo]

barmar, on 2011-June-26, 20:44, said:

GIB plays the odds, it doesn't know about safety plays. This is a 5% layout, so it doesn't affect the odds much.

Sorry, I thought GIB ran simulations and then took the line that worked in the most simulations. If *any* of those simulations had a 4-0 split onside, that would make GIB take the safety play (that's the meaning of safety play).

And as I said above, I don't know how many simulations GIB runs (or even if my understanding of how GIB works is correct), but if I'm right and GIB runs 100 simulations (say) then it will almost surely encounter a 4-0 onside split.