[1175]

If I understand, the free Robot does no simulations during the bidding. I guess that would explain the actions taken on this hand (I do wonder what the full Robot would do - sometimes just counting tricks works as well as any simulation). This hand occurred a while ago. I don't recall the auction at my table, but instead took a screenshot of (I believe) the most common auction:

---

East-West have three 8-card fits, but have (helpfully for North) stumbled into their diamond fit. North has a clear 5♣ bid (instead of 4♣) at its second turn - virtually cold opposite xxxx,xxxx,x,Kxxx, even with 2-0 trumps, given that South has a singleton diamond from the bidding. At one table, South even made a negative double, then supported clubs without North bidding them again (something like 1♣-1♦-X-2♣-P-2♦-3♣-3♦-4♣-P-P-P).

[pescetom]

I have one human partner who likes to negative double with support in that way. I could understand it better from West with decent 5-4 majors.

[smerriman]

 1175, on 2024-June-27, 05:52, said:

I do wonder what the full Robot would do - sometimes just counting tricks works as well as any simulation.

'Counting tricks' and 'simulation' are synonymous, so of course any robot which considers 5♣ via such means, as opposed to the free robot's method of counting points, will bid it here.

Unfortunately, at least in the old version, GIB has been 'optimised' to only consider a certain subset of possible alternative bids via some rules that I haven't quite figured out yet; in this specific example, 5♣ wasn't included

Edit - upon debugging, I believe GIB is told that if a database rule tells it to make a bid specifically based on the LOTT, then it is not allowed to consider higher level bids of the same strain. If I relax the LOTT condition on the 4♣ bid, it still defaults to 4♣ but does consider 5♣ and bids it.

[1175]

 pescetom, on 2024-June-27, 08:25, said:

I have one human partner who likes to negative double with support in that way. I could understand it better from West with decent 5-4 majors.

While I certainly don't mind 2♣ (South looks like he has a "one bid" hand, and South might guess that East-West will end up playing the hand), I don't mind a negative double, either. North could have only three clubs, or even the same shape as South. I actually might have tried a double with the East hand (even if a bit light on values). Anyone who has followed this forum over the past couple of months knows that I like making multi-suit bids.

 smerriman, on 2024-June-27, 14:23, said:

Edit - upon debugging, I believe GIB is told that if a database rule tells it to make a bid specifically based on the LOTT, then it is not allowed to consider higher level bids of the same strain. If I relax the LOTT condition on the 4♣ bid, it still defaults to 4♣ but does consider 5♣ and bids it.

Obviously, I have to ask here: Doesn't the LOTT tell North to bid 5♣? South surely has four clubs for raising without a rebid by North, so North should know that North-South have 11 clubs.

[smerriman]

 1175, on 2024-June-27, 18:13, said:

Obviously, I have to ask here: Doesn't the LOTT tell North to bid 5♣? South surely has four clubs for raising without a rebid by North, so North should know that North-South have 11 clubs.

Possibly, but it doesn't have that programmed in. It's just following a rule that it can compete to the 4 level if {LOTT function call returns >= 10}, but a jump to 5♣ requires strength, and that function call, while checking for at least a 10 card fit, appears to also prevent 5♣ from showing up in simulations as a side-effect.

[1175]

 smerriman, on 2024-June-27, 18:24, said:

Possibly, but it doesn't have that programmed in. It's just following a rule that it can compete to the 4 level if {LOTT function call returns >= 10}, but a jump to 5♣ requires strength, and that function call, while checking for at least a 10 card fit, appears to also prevent 5♣ from showing up in simulations as a side-effect.

I always thought that the LOTT existed to tell you when to bid without strength.

[pescetom]

 1175, on 2024-June-27, 19:34, said:

I always thought that the LOTT existed to tell you when to bid without strength.

At least when to bid despite lack of strength... but that doesn't seem to have registered with GiB.
The application of LoTT is erratic in general, at times it will boast "the LAW" in an explanation but more often it will misapply or blatantly ignore it.
Particularly irritating (and embarassing with opps/TD) is it's inability to comprehend that delayed support after an initial pass cannot promise a true fit.