[myprac]

From practice with basic robots:

---

[calm01]

I agree that GIB doubles inappropriately and because of this its responses to doubles are unduly conservative and doublers rebids can be weird as here.

GIB's doubling style has some similarity with the somewhat aggressive Blue Team style. Each one of the Squadra Azzurra could play in 4-2 fits better than most players play in 4-4 fits so they could get away with an almost any distribution style of double.

GIB is no Garozzo.

Updating GIB doubling style and its responses to doubles is overdue. Usual choice - if its easy to do - just do it. If it is not easy - consider replacing GIB.

[Bbradley62]

calm01, on 2011-December-26, 19:50, said:

consider replacing GIB.

Broken record became tedious long ago.

[myprac]

In GIB's bidding system a takeout double is in effect something of a multi bid, normally showing a minimum point count and what might be called "takeout double distribution" (2- in the suit that is doubled, 3+ in all the others) but also being used with a strong hand that may not have that distribution. In the latter case the plan is to show strength (and alert partner to the possibility of a shape that isn't suitable for a normal takeout double) with a second bid.

I assume that what happened here is that North saw it had strength for the second kind of takeout double but then found it didn't have a suit of good enough quality for a followup bid. One way to avoid that would be to adjust the criteria for the takeout double so that in cases where it means "strong hand, off shape" there has to be a rebiddable suit other than a major suit opened by opponents, so there's a place to go if partner bids your short suit. In this approach North would pass on the first round despite holding 19 total points. NS would end up defending 2♥ with opponents nonvul, which seems like a waste of North's hand but is certainly better than playing in 2♠.

Another possibility would be to make a second bid from doubler mandatory, at least when the result could otherwise be a five-card fit. In this case North would have to bid 2NT, because retreating to a minor could also lead to a five-card fit. I assume this hand falls short of the HCP criterion for 2NT in this situation but it seems like the least unappealing choice given that South's free bid shows at least a few HCP to go along with North's 17, and North holds stoppers in the unbid suits and two in opponents' suit.

[helene_t]

myprac, on 2011-December-29, 10:11, said:

In GIB's bidding system a takeout double is in effect something of a multi bid, normally showing a minimum point count and what might be called "takeout double distribution" (2- in the suit that is doubled, 3+ in all the others) but also being used with a strong hand that may not have that distribution.

I think that is true for doubler's perspective but not for advancer's perspective. At least it used to be that any subsequent call by doubler was explained as showing tolerance for all three unbid suits. To the best of my knowledge this is still the case.

Maybe GIB would double on any 17 count not suitable for a notrump overcall. So there is a need for a more subtle definition of the double. But more importantly, the definition from doubler's perspective needs to be a aligned with the definition from advancer's perspective. I may be wrong but this example suggests that it isn't.

[barmar]

This is just a bug. GIB's criteria for a strength-showing double is too simple (MP is total points):

; 16+ HCP
; 21 MP with a void (18 HCP)
; 19 MP with a singleton
; 18 MP otherwise

In the singleton/void cases, it also needs to check that its long suit is not the same as opener's, since it won't rebid NT.

The bug doesn't bite too often, since hands like this are pretty rare. The DB even has this comment:

; This works well for now. Can tweak if we find GIB doubles (or fails to double) incorrectly