[pescetom]

This board followed and put a premature end to any chances in the tournament.

N bid his hand at least once too often, but W had what GiB expected.
So why pull 4♥ (recognised as "strong, rebiddable") with 2 card fit to show just 10 HCP already bid and ♣JT2 ("described as 4+") ?

[pilowsky]

You get what you pay for.

[smerriman]

I ran this through the old version of GIB. Despite it also having a definition of 4+ clubs for 4♥, every single simulated hand it gave West held 5 clubs.

I still haven't achieved my goal of fully reverse engineering exactly how the simulated hands work. It may simply be a case of that being the only way it can make up the required number of points in West's hand. It's true West has their hand, but North and South doesn't.. if GIB is assuming South has an extra king for their bid, West is going to need a lot more distribution to make up their hand.

GIB does tend to believe its opponents way too much; if it is assuming West must have a lot of extra distribution, it may be slightly more understandable that 5♣ could work out better than 4♥x.

[LBengtsson]

pescetom, on 2022-December-05, 17:08, said:

N bid his hand at least once too often, but W had what GiB expected.

I think you were screwed over by North. I have a feeling North knew what was going to happen that is why he bid 3♠ knowing that he would not penalized. Penalty doubling 4♥ was assuming it was going down - that is a gambler's bid - but why East went AWOL with 5♣ knowing you have 6-2 ♥ fit is crazy! Even a amateur would leave 4♥X bid in imo.

Human with 6♥5♣ shape would rebid 4♣ over 3♠ surely? GIB already assumes you have that hand and goes to five level and worse contract. Not good, as you say.

[pescetom]

pilowsky, on 2022-December-05, 18:11, said:

You get what you pay for.

Not in this case.
I paid 1.50 for a 14 board tournament plus 0.29 for the robot, but I might as well have packed up after just 4 boards.

[pescetom]

smerriman, on 2022-December-05, 22:44, said:

I ran this through the old version of GIB. Despite it also having a definition of 4+ clubs for 4♥, every single simulated hand it gave West held 5 clubs.

....

GIB does tend to believe its opponents way too much; if it is assuming West must have a lot of extra distribution, it may be slightly more understandable that 5♣ could work out better than 4♥x.

I guess a lot depends on what it thinks West's hearts suit must be. "Strong" is strong, but is "rebiddable" context dependent and must it imply 6 cards here ? I have seen 5 card interventions described as "twice rebiddable", so I'm wary of these definitions.
If it expects 6 card, then I find it hard to imagine simulations showing 5C to be a better contract than 4H, even assuming 5 card clubs.

[smerriman]

Yes, all the hands held 6 hearts as well as 5 clubs.

But you're right; there is something else going on. When I tested this earlier, I ran simulations when North passed 4♥, because something was going funny when North was doubling and it wasn't showing simulated hands at all. East was always passing 4♥, but I saw 5♣ wasn't too far behind, and thought maybe that meant a penalty double could cause it to try the other.

But I've just realised what the 'something funny' was. As everyone is well aware, you should never dare a penalty double playing with GIB, since it usually pulls it. While N/S aren't GIB at all, East assumes they are. In 100% of simulated hands here, East sees that if it passes, the database will tell South to bid 4♠, so it thinks there is a 0% chance of being left to play in 4♥x. I guess that faulty assumption, combined with the expected 6-5 shape for West and North having some hearts, makes it reasonably harmless to bid 5♣ now.

[pescetom]

smerriman, on 2022-December-06, 13:35, said:

Yes, all the hands held 6 hearts as well as 5 clubs.

But you're right; there is something else going on. When I tested this earlier, I ran simulations when North passed 4♥, because something was going funny when North was doubling and it wasn't showing simulated hands at all. East was always passing 4♥, but I saw 5♣ wasn't too far behind, and thought maybe that meant a penalty double could cause it to try the other.

But I've just realised what the 'something funny' was. As everyone is well aware, you should never dare a penalty double playing with GIB, since it usually pulls it. While N/S aren't GIB at all, East assumes they are. In 100% of simulated hands here, East sees that if it passes, the database will tell South to bid 4♠, so it thinks there is a 0% chance of being left to play in 4♥x. I guess that faulty assumption, combined with the expected 6-5 shape for West and North having some hearts, makes it reasonably harmless to bid 5♣ now.

Though this be madness, yet there is method t'it