[fuburules3]

http://tinyurl.com/7w5s4qw

I don't really understand this one.

[nathan2008]

GIB will always bid 4M when he has a single because he "thinks" 3nt will go down and single card is always good for 4M lol.

[georgi]

Supposedly should stay. With singleton in partner's suits and not having 4th fit as promised is not justified to go over 3N.

[dwar0123]

nathan2008, on 2012-April-27, 02:37, said:

GIB will always bid 4M when he has a single because he "thinks" 3nt will go down and single card is always good for 4M lol.

Well, GIB is often wrong in these cases and perhaps we should change GIB as a result.

[manudude03]

I'm pretty sure I've said this before, but would it be so bad to make a rule where GiB doesn't ever pull a natural 3NT unless it has a known 8 card fit in a major or a 9 card fit in a minor in which case simulate? When I say known, I mean the minimum length "promised" added to the number held.

[Stephen Tu]

I don't like super-hard rules, it would be nice if GIB can pick a moysian or a 5-2 fit *when right*, like experts do. Certainly I've seen it do the right thing occasionally, making 4M when 3nt has no play. The problem right now is it's choosing these in poor spots, not that it does it at all. I think it's mainly because there are default rules in the database that don't fully capture negative inferences about bids not made, so that it knows the fit is *at most* 7 cd fit, rather than *at least* 7 cd fit with 8 cd fit possible. With 8, its sims are going to be successful quite often, and 7 works sometimes, so it tends to pull when it shouldn't.

We can tighten up definitiions,

1. Responder has to bid 2 sp here, this has to be fixed, I don't know any system that prefers 2nt leb over 5 cd sp. This gives opener two notrump bids, 2nt and 3nt, and 3nt can show 19+ hcp, 2nt should probably be about 17-18 hcp.
2. 3nt here should show 19+ hcp. The description says "4+ hearts". It should be 4 hearts only, not 4+ (no 3h bid), then GIB at least knows it's at *best* a 7 cd fit, instead of hoping for 8 cd fit constructions.

If after adjusting these, GIB *still* wants to pull, then maybe we need stricter "3nt ends all auctions without book bid pull (e.g. after transfer sequence)" rules.

[georgi]

3NT explanation 4+ in reversed suit will be fixed to 4 too.

Also the escape disguised as 4+ fit shouldn't happen too.