I want to know the simulation results...

[pretender]

Today's ACBL robot duplicate #3875

Board 3

 

The auction went 1C by player, 1D by GIB N, 1H by GIB E, 1NT by player, 3NT by GIB N ending the auction.

 

GIB held

 

-

63

AQJ6543

KT96

 

and thinks 3NT was the best contract?

 

Also, why did GIB E bid 1H holding 6 spades and 5 hearts?

[junyi_zhu]

Today's ACBL robot duplicate #3875

Board 3

 

The auction went 1C by player, 1D by GIB N, 1H by GIB E, 1NT by player, 3NT by GIB N ending the auction.

 

GIB held

 

-

63

AQJ6543

KT96

 

and thinks 3NT was the best contract?

 

Also, why did GIB E bid 1H holding 6 spades and 5 hearts?

Because there is no strict rule that Gib has to bid its longest suit first....

The basic framework of gib system is very flawed. Gib doesn't follow any strict rules of how to bid a two suiter. Often, you may find gib in competitive auctions rebidding and only promises 4 or more. Often, you may find gib rebid a 5 card suit and bypass a 5 card lower ranked side suit. As I said again and again, the bidding should really be based on rules and seldom the bidding should be based on simulations, because the results of simulations are random in a limited sample size, which usually leads to unpredictable weird behavior and unpleasant experience. Also, because the bidding system is so bad, in gib's simulation, it assumes that the human's bidding is as bad as gib is. Basically, the original programmer is very poor in basic bridge bidding, a lot of basic bridge bidding rules are lacking in the gib system, including the one I mentioned above.

 

Also, to make simulations effective, a set of basic rules still should be implemented, like:

0, respect the rules. If an auction is forcing, it is forcing. If an auction shows 4 or more, you can't bid a 3 card suit frequently.

1, bidding one's longest suit first, except when raising partner. (response to takeout double can be considered as a raise).

2, understand that bridge is a single dummy game. For example, bidding NT when it's a reasonable spot, which should never be based on a double dummy analysis because bridge is a single dummy game and many double dummy unmakable NT are very decent in a single dummy base.

3, limit the strength range as fast and as narrow as possible. For example, 1x 1y double pass 2z, this 2z shows something like 12 to 22. And when x = z, this rebid only promises 4 cards. This is actually a problem coming from the rule 2, since gib doesn't like bidding NTs, it has to invent a lot of bad or faked suit bids. Often NT bids defines the playing strength very fast and well.

4, Evaluation in a trick taking base in many situations. In competitive auctions, often it makes more sense in a trick base not HCP base. Often, you may see something like 1S p 4S 5C: showing 20 points or more, which is a nonsense. 5C just shows a hand that plays well in clubs with a lot of tricks. Also as a result, gib often raises the bid to 6 with some very bad cards, including KQ in spades. Also, if you can have a expectation of how many defensive tricks you can take, it's not difficult to make penalty doubles, which gib rarely makes.

[calm01]

junyi_zhu,

 

What you say about the lack of few a basic bridge guidelines in GIB and its tendency not to no double for penalty very often, its unwillingness to bid NT to limit its hand and its poor slam bidding jugdment etcetera all resonates with my experience of GIB as a bidding partner or bidding opponent.

 

This all makes very grim reading.

 

I could add to your comments on GIB mis-bidding two suiters to its frequent mis-responding to a partner (GIB or human) showing a two-suiter. What seems to be missing is being able to communicate or recogne distributional information.

 

GIB also does not always support with support (especially for hearts) losing much of the advantage of a 5 card major system.

 

This sounds like a recipe for abandoning all hope.

 

However, from what Ari has written in many recent posts, it seems that correcting a few of the more commonly occurring bidding descriptions and ensuring GIB understanding corresponds to its descriptions will make a very significant improvement in the GIB bidding experienced by human partners.

 

Also its database or similar situations may help to improve its bidding over time.

After all it is not a bad declarer. And adding some signalling in defence will help a lot.

 

So there is much to go for.

 

In summary it seemed to me like a "wouldn't start from here" situation but Asi has gven me hope for a better future.

 

What do you feel?

[junyi_zhu]

junyi_zhu,

 

What you say about the lack of few a basic bridge guidelines in GIB and its tendency not to no double for penalty very often, its unwillingness to bid NT to limit its hand and its poor slam bidding jugdment etcetera all resonates with my experience of GIB as a bidding partner or bidding opponent.

 

This all makes very grim reading.

 

I could add to your comments on GIB mis-bidding two suiters to its frequent mis-responding to a partner (GIB or human) showing a two-suiter. What seems to be missing is being able to communicate or recogne distributional information.

 

GIB also does not always support with support (especially for hearts) losing much of the advantage of a 5 card major system.

 

This sounds like a recipe for abandoning all hope.

 

However, from what Ari has written in many recent posts, it seems that correcting a few of the more commonly occurring bidding descriptions and ensuring GIB understanding corresponds to its descriptions will make a very significant improvement in the GIB bidding experienced by human partners.

 

Also its database or similar situations may help to improve its bidding over time.

After all it is not a bad declarer. And adding some signalling in defence will help a lot.

 

So there is much to go for.

 

In summary it seemed to me like a "wouldn't start from here" situation but Asi has gven me hope for a better future.

 

What do you feel?

Of course solving all those problems case by case is possible. Still, it will be a very tedious work and a general guideline is still very important, IMO, to avoid major disasters like passing a forcing bid, failure to pull to the right suit contract when opp makes penalty doubles ( it does happen sometimes, I sometimes showed 5-5 and opp doubled me in one suit, gib failed to pull to the right 5-3 fit and left me playing in the 5-1 fit). Also, a subroutine to evaluate the number of tricks is certainly very important IMO.

[hotShot]

Because there is no strict rule that Gib has to bid its longest suit first....

 

A strict rule to bid the longest suit would make GIB bid the 4 card suit of 4333 shapes, bid ♣ if you have 5♣ and 4♦ or bid a 6 card minor prior to a 5 card major. Are you sure you want that rule?

 

I guess that GIB also has a rule that reverse bidding promises extra strength and

perhaps it knows that AKQJT might be a better suit than 765432.

 

If you ever tried to make a set of rules for bidding, you would have noticed that often rules overlap/conflict. We can assume that GIB is allowed to ignore (some) rules to resolve a conflict.

[junyi_zhu]

Because there is no strict rule that Gib has to bid its longest suit first....

 

A strict rule to bid the longest suit would make GIB bid the 4 card suit of 4333 shapes, bid ♣ if you have 5♣ and 4♦ or bid a 6 card minor prior to a 5 card major. Are you sure you want that rule?

 

I guess that GIB also has a rule that reverse bidding promises extra strength and

perhaps it knows that AKQJT might be a better suit than 765432.

 

If you ever tried to make a set of rules for bidding, you would have noticed that often rules overlap/conflict. We can assume that GIB is allowed to ignore (some) rules to resolve a conflict.

Well, if you like the way gib bids now, you don't have to join this thread. Of course we are not in the same school of bidding. I really like those who usually open 1D with 4D-5C or open 1M with 5M 6m as my opponent. Still, that's even not the point of this discussion. I am talking about general bidding principles, not exceptions.