[Vilgan]

Had an interesting thing happen yesterday. Was practicing with partner against 2 GIB opponents. Auction was the following (we are silent, GIBs bidding)

1NT 2♣
2♠ 4♣ (splinter, showing 1- clubs, strong hand)
4♠ all pass

Dummy hit and there was a singleton in dummy in dummy. However it was in hearts! When we clicked on GIB's bid it even stated it showed a singleton club.

So GIB can psyche? Or is this just a bug in their bidding algorithms?

I can find the hand if it would be useful.

Eric

[effervesce]

Perhaps it was an artefact of the single(double?)-dummy simulation engine GIB uses when working out what to bid? Perhaps it simulated (x hands) for the 1NT opener with a 2♠ rebid, and found that a club splinter (pretending to have 0-1 clubs) reached the best spot opposite the 1NT opener. I could be wrong as how GIB decides what to bid though...

[brianshark]

GIB psyched a control showing cue-bid against me the other day. He had Jxx. I guess GIB psyches from time to time. I'm impressed.

[uday]

I think the op might have found a bug ( as opposed to a psych )

[P_Marlowe]

Hi,

GIB does not psych, if a machine could psych, it
would be able to think like a human, and I doubt
that machines will be able to think like humans,
... a least for a couple of years to come, hopefully
after I am dead.

Since it took a while to find the link, will the following
link a 2nd time
http://forums.bridgebase.com/index.php?sho...371&hl=Ginsberg

With kind regards
Marlowe

[barmar]

Not a psyche, but GIB found another auction I've never seen before today. In an MBT, my bot partner opened 1NT, and our unopposed auction proceeded:

1NT-2♣
2♠-3NT
4NT-P

Since when does the NT opener invite a slam? He did end up making 7; I had 12 HCP, they ducked an ace, and then every missing honor card was in the perfect place. But if one card was out of place it could have been much worse, because the opening lead was a diamond, where GIB held Kx opposite my xx (that was the ace they ducked).

But all is forgiven, because in a later MBT, GIB found a business redouble and then bid a cold grand on the last board, to launch me from 5th to 2nd place. No science to this auction, it was

1♠-2♥
2♠-3♣
3♦-6♦
7♦-P

GIB's hand was ♠Jxxxx ♥Kx ♦AQxxx ♣x, mine was ♠void ♥AQxxx ♦KJxx ♣AKQJ.

[helene_t]

Of course GIB could have been programmed to psyche (except that one could argue whether it really is a psyche rather than an exotic part of its system) but it isn't. This is either a bug or an artifact of its algorithm which relies on simulations that occasionally yield weird statistics.

This looks like a bug.

A few days ago it opened 1♦ on a 4-card while it also had a decent 5-card spades.

[hotShot]

GIB used to transfer 1NT opener to his 4-card major (with weak hands) every now and than.....

[helene_t]

hotShot, on May 1 2008, 01:06 PM, said:

GIB used to transfer 1NT opener to his 4-card major (with weak hands) every now and than.....

That makes more sense, when GIB is captain it is fine to use simulations and DD analysis to determine the best call.

When p is captain, as in the splinter example and the 4N slam try by opener, I think GIB should be constrained by some more rigid rules, though. It is really beyond me why the simulation-and-DD algorithm considers such calls in the first place. But maybe I don't quite understand how GIB works. Presumably it has no notion of captaincy.

[hotShot]

Think of the above mentioned board, GIB knows from his simulations that there is a good chance for slam.

GIB also knows that if he bids the ♥-Splinter, the bidding will have to pass 4♠ to discover that you should not bid more than 4♠.

Bidding the wrong splinter allows partner to bid his red suit controls below 4♠. So GIB can bid 4♠ signoff over 4♥. And has the mixed cuebid 4♥ over 4♦. So GIB has some control over the auction here.