[1175]

I would normally use a more descriptive title, but I want to post this first as a "what would you bid?" quiz (and follow up more details later). In the duplicate IMPs, you pick up this hand:

The explanation of the double: "Takeout double - 5+ clubs; 2- diamonds; 5+ hearts; 3 spades; 16+ total points"

---

What do you bid?

[pescetom]

GiB system notes say that 2NT is Unusual, which I imagine was explained as 5+♣ 5+♥.
If so, your pass was rather odd and your partner is asking you to wake up and bid 4♥.
But yes, the arithmetic of the explanation of double is dubious

[smerriman]

Yeah, the first pass is already a mistake, but 4♥ is clear here.

GIB's logic for responding to the double here is pretty straightforward:

- determine which suit is your best fit
- determine what level you can bid that at based on points alone
- make that bid, unless there's another bid that will clearly lead to a better result based on thinking about what hands partner can have.

Its best fit is clubs, it doesn't have enough points for 5♣, but before bidding 4♣ it confirms that 4♥ is obviously better, so bids 4♥.

Unless you're playing with a free robot, in which case BBO kill the algorithm before it gets to the word 'unless', even though it's the most crucial part, so I assume the robot bid 4♣.

(At least, that's the principle. With the opponents holding 11 diamonds, even an advanced robot may see nothing wrong with bidding slowly, since with an 11 card fit it 'knows' the opponents will be forced to bid higher based on LOTT and so it will have a second chance anyway.)

[1175]

smerriman, on 2024-June-08, 16:00, said:

Unless you're playing with a free robot, in which case BBO kill the algorithm before it gets to the word 'unless', even though it's the most crucial part, so I assume the robot bid 4♣.

You know the robot well. It did indeed bid 4♣, and I played it there. I agree that 4 ♥ looks clear, particularly since, at the 4-level, you might as well pick the suit where you have a game if it makes.

Fourteen of the fifteen other tables played in game (one in 5♣, and all of the rest in 4♥). At the one other table where South made an Unusual NT call (surprised no one else did), he followed it up with 4♣, which the robot duly punished. I don't like (or even agree) with the treatment of the 4♣ bid ("3- ♦; 3- ♠; twice rebiddable ♣; twice rebiddable ♥; 25+ total points"), but given the system, South can't make that bid. The entire hand (with the bidding at the table discussed last):

---

I have one additional comment (and one additional question). One would think that the robot should trust the opening bid enough (vulnerable versus not, and a robot instead of a human) to figure out that South has one key card and not four (it could have bailed out in 5♥, which makes). Finally, why does the robot often not double when auctions go completely off the rails (as here)? A double of 7♣ (and anything higher) looks automatic.

[smerriman]

1175, on 2024-June-08, 23:43, said:

One would think that the robot should trust the opening bid enough (vulnerable versus not, and a robot instead of a human) to figure out that South has one key card and not four (it could have bailed out in 5♥, which makes).

One could argue the opposite if it believes its partner has the 25+ total points it promised Obviously a stupid definition, but now you may be less surprised why nobody else bid unusual 2NT - you simply can never do so with GIB if you're planning to compete further, or it will play you for game in your own hand.

1175, on 2024-June-08, 23:43, said:

Finally, why does the robot often not double when auctions go completely off the rails (as here)? A double of 7♣ (and anything higher) looks automatic.

In most other situations, because if you haven't learnt from painful experience, virtually all doubles are for takeout with GIB, even at high levels.

But not here; double is penalty, but it generally falls after the 'unless' logic I referred to originally, so can't be done by bid-by-hardcoded-rules-only free robots. (Which makes sense; doubling a contract should be about how many tricks you expect the contract to make, not about how many points you have.)

[1175]

smerriman, on 2024-June-09, 23:41, said:

...but now you may be less surprised why nobody else bid unusual 2NT - you simply can never do so with GIB if you're planning to compete further, or it will play you for game in your own hand.

Good point. I will remember that.

Does the system ever get updated (perhaps by popular demand) when a treatment proves suboptimal?

[smerriman]

It was regularly updated up until February 2019, when all went quiet..