[thepossum]

---

I was South at another table who received a close to bottom (quite a few of us) with an ambitious 4H overcall on round 1. But I think this 7 Clubs takes some beating for a top to those lucky enough to by the recipient of the generosity. Whatever people say surely there is something amiss

A new form of takeout by South followed by pass-pass means jump to grand

[thorvald]

Great board.

I have just tested this with the BBO bots in the bidding area of BBO, and here East bid 4!C after the double of 3N - so probably just a glitch in the Matrix

[EarlPurple]

It's a glitch pulling 3NT at all. If they bid it it's because they think it's the best contract. The opponents doubling it is no reason to pull it. It's also cold if they dare to take the winning spade finesse.

[pescetom]

thorvald, on 2022-September-05, 06:04, said:

I have just tested this with the BBO bots in the bidding area of BBO, and here East bid 4!C after the double of 3N - so probably just a glitch in the Matrix

Running to 4!C is not my cup of tea, but I guess if partner can bid 3!H on that hand then it makes some sense.

[smerriman]

I can't find the thread, but I'm positive I've seen this before, when GIB jumps to grand out of the blue after 3Nx. Only explanation I can think of is that some data went missing in the transfer somewhere.

On the other hand, if that were the case, it shouldn't be related to the auction. If I could find the other thread that'd help narrow down the similarities.. anyone remember it?

[smerriman]

OK, it was worse than I thought; it's actually an issue in the bidding database.

Firstly, at least in the 2012 version, GIB doesn't even consider passing 3NTx; it's a reasonably well known issue that it runs from 3NTx way too often. I should check how true that is today, but it's probably still the case, and if so, you should basically always double 3NT.

But what actually happens is that when it simulates, it thinks that a significant proportion of the time, you're going to bid further in hearts. In fact, it predicts that if it bids 5♣, you'll often bid 5♥ (which it can double); if it bids 6♣, you'll often bid 6♥ (which it can double), and if it bids 7♣, you'll often bid 7♥ (which it can double).

So clearly 7♣ is the best way to profit

[smerriman]

Expanding a little further, the GIB database basically treats the double of 3NT as starting a forcing pass situation.

And its forcing pass rules are totally messed up (same is true after redoubling a takeout double), where passing promises 3+ of their suit, and doubling promises 2 or 3 cards in their suit. Anything else, and it's told to rebid its own suit.

Of course, after East bids 7♣, a robot in South will realise via simulation that its own logic is completely stupid, and never bid 7♥.

But East's simulation assumes the South player will blindly follow the database. So East "knows" that if it bids 7♣ and South holds at most 1 club, it'll get a juicy penalty by doubling the forced 7♥ bid. The gains when this happens can easily outweigh the losses when South holds 2-3 clubs in a simulation.

I really have to follow up this 3N doubled thing - if GIB really still never passes like in the past, then you'll be able to steal so many points by doubling 3NT, and this situation of GIB leaping to grand will occur quite regularly. If this information leaks out to the wider community (I post it on BridgeWinners), BBO may have no choice but to work on fixing GIB again