You've described your hand, I've placed contract

[Bbradley62]

http://tinyurl.com/2a7bcp3

 

3♦ is explained as "3+D; 4+S; 6-9 total points". The description of 3NT includes "21 HCP; 22- total points".

 

(1) If North has at least 6 total points, and we bid game on 25 total points, why do I need 21HCP to bid 3NT?

(2) If GIB has described his hand to within a 3 total point range with his distribution reasonably well described, and partner then bids 3NT, GIB should consider that a signoff and pass.

[jamegumb]

GIB also makes a number of frustrating "4 or 6" decisions on minor suit hands; he raises your minor to 4, but when you place the contract in game he decides that you're better off in slam.

 

Thankfully, he doesn't get angry if you bid RKCB and pass his response.

[arigreen]

http://tinyurl.com/2a7bcp3

 

3♦ is explained as "3+D; 4+S; 6-9 total points".  The description of 3NT includes "21 HCP; 22- total points". 

 

(1) If North has at least 6 total points, and we bid game on 25 total points, why do I need 21HCP to bid 3NT?

(2) If GIB has described his hand to within a 3 total point range with his distribution reasonably well described, and partner then bids 3NT, GIB should consider that a signoff and pass.

One problem is that 6-9 total points is not really a playable range for 3D. 3D must be either NF or F. If NF, it should be limited to 7 total points. If F, it should be 8+ total points and forcing to 3N.

 

With that said, 3N shows 25 combined HCP. Partner has only promised 4HCP, so GIB thinks that its 3N bid promises 21 HCP.

 

A common pattern that we see is confusion between HCP and TP.

 

Perhaps a better way of looking at this would be:

 

- Responder could have up to 9 HCP for his 3♦ bid.

- With 25 combined HCP, opener will want to bid 3N

- Hence, responder should not play opener for more than 16HCP.

[calm01]

TOTAL POINTS appear to have little brudge meaning.

 

Before a fit is found total points are nearly pointless.

 

After a fit is found controls, HCP, or losing trick count are all good guides.

 

If no fit is found total points are entirely pointless.

 

If notrump is the selected final contract high card points are probably the best guide with playing strength a secondary consideration.

 

Perhaps the total points concept is a poor mans' playing strength estimate.

 

Ari, if you cannot determine any practical value for total points, just remove all reference from the descriptions and remove all coding references to total points.

 

This might improve the bidding judgement of GIB at one stride.

[arigreen]

It's not easy to drop total points altogether:

 

- Many rules rely on total points, so removing these rules will leave many gaps in the rule database.

- Even without a fit, total points can be useful for estimating playing strength of a hand.

 

It might be interesting, however, to convert all references to "total points" to HCP. I suspect that if we did this, however, GIB's judgement would be worse.

[junyi_zhu]

It's not easy to drop total points altogether:

 

- Many rules rely on total points, so removing these rules will leave many gaps in the rule database.

- Even without a fit, total points can be useful for estimating playing strength of a hand.

 

It might be interesting, however, to convert all references to "total points" to HCP. I suspect that if we did this, however, GIB's judgement would be worse.

Well, I think total playing tricks and defensive tricks are actually way more important than total points. Defensive tricks help to determine whether to sac or double opps. Total tricks help to determine how high you want to play. In slam biddings, a good count of total tricks is the key to good slam and grand slam bidding.

[arigreen]

I modified GIB to use HCP in place of total points and ran a 100-board match. As I had expected, the version with total points destroyed the version that used only HCP.

[calm01]

Ari,

 

Thanks for trying.

[barmar]

That result is hardly surprising, everyone knows that vanilla HCP is a very poor way to judge hand strength, barely good enough for beginners. You need adjustments based on shape, location and concentration of values, etc.

 

Human players look at many different things, and it's hard to quantify them all. The problem is that computers are hard to program if you can't quantify the data; representing all the vague patterns we recognize, and incorporating them into the bidding database, is difficult.

[calm01]

Barmar,

 

I agree about human pattern recognotion taking many aspects of hand evaluation into account. but if you were to limit GIB to just two hand evaluation techniques I for one would vote for:

 

- HCP for no-trump

- losing trick count (less one for 3 Aces) for suit contracts with a 8 card or longer fit.

 

You may have your favourite two techniques - why not apply them to GIB?

 

At lease two techniques are required because:

 

- HCP is very accurate for notrump as it distinguishes just one point difference for 3MT (25) and 2NT (23-24),

- just using HCP can be a poor guide for suit contracts with an 8 card or longer fir and even less relevant when no fit is found where a measure of playing strength can be more important.

 

I sispect Ari's test of 100 hands suffered from using HCP for those hands which wer notrump (little problem perhaps) and also for suit fir/non-fit hands where it would be a poor guide.

 

One size does not fit all -at least two techniques are required - one for each of notrump and suit-fit/ suit non-fit.

 

I suggest three techniques would pay even higher dividends as losing trick count for suit non-fits does not do a good job.

 

What do you feel is the best way forward to improve GIB bidding?

[Bbradley62]

Perhaps a better way of looking at this would be:

 

- Responder could have up to 9 HCP for his 3♦ bid.

- With 25 combined HCP, opener will want to bid 3N

- Hence, responder should not play opener for more than 16HCP.

No, this is backwards. If responder has no more than 9, and opener is willing to bid 3NT, showing a partnership total of 25, then opener has at least 16, not at most 16. (Opener also reversed.)

 

I generally don't like "3NT ends all auctions", but since GIB has these issues, how about adding a rule that 3NT ends all auctions unless GIB has at least 6 points more than the minimum he has shown, or 2 more cards more than he's shown in his longest suit? Note, I'm not saying GIB should always bid more if he meets this criteria, just that he should automatically pass if he doesn't.

[arigreen]

No, this is backwards.  If responder has no more than 9, and opener is willing to bid 3NT, showing a partnership total of 25, then opener has at least 16, not at most 16.  (Opener also reversed.)

 

 

I generally don't like "3NT ends all auctions", but since GIB has these issues, how about adding a rule that 3NT ends all auctions unless GIB has at least 6 points more than the minimum he has shown, or 2 more cards more than he's shown in his longest suit?  Note, I'm not saying GIB should always bid more if he meets this criteria, just that he should automatically pass if he doesn't.

I agree that opener has at least 16. But it is unsafe for responder to play opener for 21, and that is just what responder was doing before I made the last change.

 

Adding a rule along the lines of "3NT ends all auctions unless..." might be interesting.

[Bbradley62]

Hmm... "3NT ends all auctions unless..." has to have exceptions for constructive auctions like 1N-2H-2S-3N.

[Bbradley62]

Same concept...

 

[hv=d=w&v=n&s=sa93hj87dk954ckj9]133|100|Scoring: XIMP

This is the NORTH hand. No interference:

1♦-2♣

2♦-2♠

3♣-3NT

4♠

[/hv]

 

There's no reason to pull partner's 3NT.

[Bbradley62]

Another with the same concept, but not in 3N:

 

http://tinyurl.com/2cjvywe

 

After bidding 1N and 3♣, North's hand is described as:

2+C; 3-H; 3-S; 8-11HCP; 12- total points; likely stop in D.

 

That sounds like an accurate enough description to allow partner to place the contract without being over-ruled. Why does GIB bid 6♣?

[barmar]

Look at the description of your 5♣ bid, it says 6+C.

 

Minor suit bidding with GIB is really screwy. It seems like whenever you jump to 5, it thinks you're showing extras, so if it likes its hand it frequently raises to 6. But I can never tell whether it will take 4 as forcing to game. There doesn't seem to be anything in between game invitation and slam invitation.