[AAr]

The biggst flaw I've seen with GIB is with reopening takeout doubles is that they'd be the perfect bid except they often pass them on some inapproprate hands, often allowing the GIB oppoent to make a doubled bid with an overtrick when the contract we should've been in makes or we at least could push them into a non-making contract.

Example of such auction:

I have a mimimum opening hand with 4-4-4-1 Dist. (Like S: x, H: KQxx, D: QJxx, C: AJxx.)
LHO Me RHO P:
1S X 2S P
P X* P P
P

*- takeout.

With the way the hands are, 2Sx makes with an overtrick. 4H our way makes. Should I start bidding 4-card suits instead?

Comments?

[AAr]

Well, one example just came up:

I had A763, 2, AKQ8753, A

I opened 1D. LHO overcalls 2C. Partnet and RHO both pass.

I thought about bidding 5D, but I wanted to look for a Spade fit, rather playing in a 4-4 Spade fit with the Diamonds likely running than in 5D. Partner passes the double! 2Cx is down one, but 6D makes. BTW, gib partner had T4, AK943, J4, J865, where the correct call on a reopening double is 2H (in which I bid 5D over it), not pass.

Are reopening takeout doubles REALLY to be avoided when you don't want partner to pass?

ETA: It appears my plan in the above hand was totally wrong, as most of the Field was in 3NT, which made with two overtricks. Still, my point about asking whether to use reopening doubles when I don't want my partner to have them unless the bid suit is also my GIB partner's suit stands.

[helene_t]

It has been discussed before that GIB converts doubles too often.

The way GIB works is that it simulates a number of hands consistent with the bidding, computes the double dummy results for each of the simulated hands for (for example) 2♣X as well as some alternative contracts, and from this it computes the average total points scored by converting double. If it is better than the alternatives it will convert the double. (I am not sure exactly how the alternatives to consider are chosen, maybe someone can enlighten me).

Anyway, a crucial element of that algorithm is the assumption it makes about the meaning of the double, i.e. which hands it should consider consistent with the auction and include in the simulations. Many doubles are not really defined as take-out but rather as adding no extra information to what is already known from the auction. In that case it will include some hands with more enemy trumps than what would be plausible for a take-out double. And this would bias the results in favor of converting the double.

With Axxx-x-AKQxxxx-A I think it is advisable not to double when playing with GIB. I would bid 3NT.

[Bbradley62]

Too funny. I just made what I thought was an obvious penalty double, in pass-out seat in a competitive auction, and GIB took it as takeout. Playing matchpoints...

---

[cloa513]

Bbradley62, on 2011-December-02, 11:09, said:

Too funny. I just made what I thought was an obvious penalty double, in pass-out seat in a competitive auction, and GIB took it as takeout. Playing matchpoints...

---

Also funny that if you rearrange the minor cards so that East has 5C and North 3D then I don't think you can take the 3S contract down.

[Bbradley62]

Yes, there's always a worst-case scenario. I guess I'm one of those who believes that if you don't occassionally eat a double, you're not doubling enough.