[Bbradley62]

After taking the first 10 tricks, GIB makes a can't-win play of exiting a club and pitching a potential winner instead of a sure loser.

---

With 3 cards to go, GIB-North knows that East has the winning club; East's other cards must be:
a) two hearts,
b) two diamonds, or
c) one of each.

If a), the winning play is to exit the club and discard a diamond, endplaying East.
If b), the winning play is to lead a heart and take the finesse.
If c), the winning play is to cash hearts from the top.

GIB's chosen line can never win. I would have expected GIB to simulate hands that fit a), b) and c) and to make the winning play associated with whichever holding occurred most often in the simulations.

[hatchett]

It was playing for (a) but when LHO pitches a heart it knows
RHO has at least one diamond to cash, so the heart ten will never win a trick.

[dwar0123]

They play in a way that a human never would, but hatchett is right. Both sides are simulating and it isn't like east doesn't know that the king of diamonds is the setting trick before he exits a heart. After declarer pitches north's 10♥, east knows it doesn't matter which order the tricks are played, 1 each. Had north kept the heart, he would of course have taken the king of diamond first, same result.

[Bbradley62]

hatchett, on 2012-May-14, 01:18, said:

It was playing for (a) but when LHO pitches a heart it knows RHO has at least one diamond to cash, so the heart ten will never win a trick.

Ah. I didn't catch that. Thanks.

[Bbradley62]

I wonder if GIB would have played differently at matchpoints. With no particular reason to believe that any of the three alternatives was more likely than the others, playing for c) guarantees 11 tricks while trying for 12, whereas the others risk making only 10.