[gszes]

a charming bidding sequence (no op int) 1c p 1h p 1s PASS OUT anyway gib hand was x QJxx Qxxxxxx xx. So not only did gib bid 1h instead of 1d it left me in a ridiculous 41 (known) trump suit. A side thought might be to have gib bid its longest suit with weak hands so they have some semblance of a rebid if things are looking grim. 1c 1d 1s 2d would have been light years better than the path chosen even if playing some form of xyz a 3d rebid (weak) would have been light years better than the chosen sequence.

[smerriman]

Taking this to the extreme, if you give GIB ♥5432 ♦AJT987654, it will bid the same way - 1♥ then pass. This doesn't seem right.

[cloa513]

 smerriman, on 2017-October-23, 18:18, said:

Taking this to the extreme, if you give GIB ♥5432 ♦AJT987654, it will bid the same way - 1♥ then pass. This doesn't seem right.

Taking what to the extreme. You mean the normal GIB sequence or the alternative. The alternative say　bid 1D in response to 1C. 　

[smerriman]

As in, the extremal case - it won't even bid 1♦ when holding a 9 card suit to the AJ and 5432 in hearts (as opposed to, say, it only doing it when it had heart honors, or a 6 card diamond suit and not 7, etc).

[The_Badger]

�� Duh!

[steve2005]

Hate to tell you this but a lot of humans will bid 1M over a long minor with a weak hand.
Look at all the people who check the frequentlybypass 4+♦box.
You have to live with the 1M bid.
Can Gib later bid the long minor to play? I doubt it is possible.

[johnu]

 steve2005, on 2017-October-24, 12:52, said:

Hate to tell you this but a lot of humans will bid 1M over a long minor with a weak hand.
Look at all the people who check the frequentlybypass 4+♦box.
You have to live with the 1M bid.
Can Gib later bid the long minor to play? I doubt it is possible.

How many humans would pass opener's second suit bid when they have a singleton (or void)?

[helene_t]

 steve2005, on 2017-October-24, 12:52, said:

Hate to tell you this but a lot of humans will bid 1M over a long minor with a weak hand.
Look at all the people who check the frequentlybypass 4+♦box.
You have to live with the 1M bid.
Can Gib later bid the long minor to play? I doubt it is possible.

GIB does not play Walsh. There's no reason to respond 1♥ since opener will bid 1♥ if he has four hearts.

[steve2005]

 johnu, on 2017-October-24, 13:35, said:

How many humans would pass opener's second suit bid when they have a singleton (or void)?

How many humans have been programmed to pass when they don't know what to do?

[virgosrock]

 gszes, on 2017-October-23, 15:35, said:

a charming bidding sequence (no op int) 1c p 1h p 1s PASS OUT anyway gib hand was x QJxx Qxxxxxx xx. So not only did gib bid 1h instead of 1d it left me in a ridiculous 41 (known) trump suit. A side thought might be to have gib bid its longest suit with weak hands so they have some semblance of a rebid if things are looking grim. 1c 1d 1s 2d would have been light years better than the path chosen even if playing some form of xyz a 3d rebid (weak) would have been light years better than the chosen sequence.

I read all the other responses to this thread.
1. I don't think a Monte Carlo simulation methodology applies to bidding. A different approach is needed for bidding. Monte Carlo is Lazy Programming for bidding.
2. Did the opps miss anything? Usually when GIB does "stupid" things the opps had something on.
3. Bidding is so much easier to program than play or defence.
4. I believe Bidding could be programmed better but undoing and redoing will open up a BIG can of worms or be VERY expensive hence BBO won't do it.

virgosrock

[johnu]

 steve2005, on 2017-October-24, 19:46, said:

How many humans have been programmed to pass when they don't know what to do?

Too many

[smerriman]

 virgosrock, on 2017-October-25, 12:25, said:

3. Bidding is so much easier to program than play or defence.

This isn't true at all. The logic behind the play and defense algorithm is actually reasonably straightforward; coming up with bidding rules to cover every possible situation is incredibly complicated.

[virgosrock]

 smerriman, on 2017-October-25, 13:27, said:

This isn't true at all. The logic behind the play and defense algorithm is actually reasonably straightforward; coming up with bidding rules to cover every possible situation is incredibly complicated.

interesting thanks smerriman. will need to think about this. are you into software?

vrock

[Vampyr]

 smerriman, on 2017-October-25, 13:27, said:

This isn't true at all. The logic behind the play and defense algorithm is actually reasonably straightforward; coming up with bidding rules to cover every possible situation is incredibly complicated.

Well. When I was a child, I had this Goren bidding wheel. If Gib played a simple system with few or no conventions, a reasonably simple lookup table would, I think, suffice.

Gib could probably get pretty far with a rule like "bid the longest suit you haven't shown, or extra length in a previously bid suit if you have enough values to bid at the level required". A slightly different rule could be used for fit auctions.

[virgosrock]

 Vampyr, on 2017-October-25, 18:26, said:

Well. When I was a child, I had this Goren bidding wheel. If Gib played a simple system with few or no conventions, a reasonably simple lookup table would, I think, suffice.

Gib could probably get pretty far with a rule like "bid the longest suit you haven't shown, or extra length in a previously bid suit if you have enough values to bid at the level required". A slightly different rule could be used for fit auctions.

Well said Vampyr. i was trying to say the same thing but could not articulate it.

Bidding based on Monte Carlo simulations is a waste of time. How many scenarios could GIB do in the what looks like 1/10 th of a second when the number of possible hands is 1 followed by 30 0's or something like that and the number of bidding sequences are say a 1000th of that.

Pshaw simulation for bidding !

vrock

[smerriman]

 Vampyr, on 2017-October-25, 18:26, said:

Well. When I was a child, I had this Goren bidding wheel. If Gib played a simple system with few or no conventions, a reasonably simple lookup table would, I think, suffice.

Gib could probably get pretty far with a rule like "bid the longest suit you haven't shown, or extra length in a previously bid suit if you have enough values to bid at the level required". A slightly different rule could be used for fit auctions.

Fair enough. I suspect it would still be considerably more complex than you expect, especially when you consider opposition interference, but getting rid of all conventions altogether would definitely make for a simpler robot, albeit not one that I'd ever be able to play with.

 virgosrock, on 2017-October-25, 19:05, said:

Bidding based on Monte Carlo simulations is a waste of time. How many scenarios could GIB do in the what looks like 1/10 th of a second when the number of possible hands is 1 followed by 30 0's or something like that and the number of bidding sequences are say a 1000th of that.

Pshaw simulation for bidding !

vrock

Simulations are only used by advanced robots, only for a handful of the bids, and even then, only rarely change the resulting bid (it still uses rules to come up with its first choice of bid). It's what causes the biggest improvement over the basic bots, so not something you'd want to get rid of.

[virgosrock]

 smerriman, on 2017-October-25, 21:21, said:

Fair enough. I suspect it would still be considerably more complex than you expect, especially when you consider opposition interference, but getting rid of all conventions altogether would definitely make for a simpler robot, albeit not one that I'd ever be able to play with.


Simulations are only used by advanced robots, only for a handful of the bids, and even then, only rarely change the resulting bid (it still uses rules to come up with its first choice of bid). It's what causes the biggest improvement over the basic bots, so not something you'd want to get rid of.

Valuable info smerriman. Does MB(money bridge) use advanced Bots? Based on what you say the Rules suck. I like what Vampyr is saying. I just have this dreadful suspicion GIB goes for low probability end cases rather than high probability choices. These "bug fixes" reported are ad infinitum and ad nauseum because of the gargantuan number of bidding possibilities. I still think the number of bidding sequence possibilities are not that vast and a look-up table like someone suggested is the way to go. and a simple rule "GIB not sure, bid a supported suit" instead of passing will work better. We have seen the abysmal PASS many times.

vrock