[CarlRitner]

barmar, on 2012-July-25, 23:45, said:

It bumps up the reputation of the author of the post.

I would like to keep the input / output very simple.
KISS, and it is good to be back!

[Grizz1y]

CarlRitner, on 2012-July-29, 13:01, said:

...While bidding can be helped by simulation, it is pretty much rule based, since every bid is supposed to be bounded, and needs to have a definition and an explanation for the opponents. Otherwise it just is not bridge.

Well, you only need to give explanation to agreements -- not if your bid is not based on an agreement...

Just thinking that simulation in bidding would usually be applied in the final stage -- for example, when deciding whether to accept partners game-invitation or not, or when competing with opps abt the final contract (should we pass or bid once more).
And, incidentally, there is not much explanation to give, for the last bid in sequences like:

1H-3H-4H
1H-3H-PASS

or
1S-(2H)-2S-(3H)-3S
1S-(2H)-2S-(3H)-PASS

Its hard to see why a properly written simulation would not work well in such cases...

[barmar]

CarlRitner, on 2013-January-13, 19:48, said:

I would like to keep the input / output very simple.
KISS, and it is good to be back!

What does that have to do with what I wrote about using the + button to "like" a post?

[barmar]

Grizz1y, on 2013-January-17, 19:16, said:

Well, you only need to give explanation to agreements -- not if your bid is not based on an agreement...

That was the problem with the original GIB implementation. Everything was based on simulations, and it made crazy bids that partner couldn't understand. Or it would make unilateral decisions like jumping to game or slam. Bidding is a conversation, you need agreements.

I suggest you search the Google Groups archive of rec.games.bridge for discussions of GIB around 2000 to read about the crazy stuff it did.

[Grizz1y]

barmar, on 2013-January-18, 10:29, said:

That was the problem with the original GIB implementation. Everything was based on simulations, and it made crazy bids that partner couldn't understand. Or it would make unilateral decisions like jumping to game or slam. Bidding is a conversation, you need agreements.

I suggest you search the Google Groups archive of rec.games.bridge for discussions of GIB around 2000 to read about the crazy stuff it did.

Yes, I know this history.... Thats why I wrote "properly written" simulation

Seems to me it was the implementation that was kind of broken at the time and never fixed and/or used in the wrong situations, etc...
Not that simulation as such is bad, when used in the appropriate situations (like, whether to compete once more, to raise to game or not, to explore/bid slam or not, etc).

Wasnt it just that the bidding-database had too many holes in it?
And simulation maybe used as soon as it could no longer find any matching sequence/bid in the database, or similar?

[CarlRitner]

So, the bidding database is awful because there's not much in it. BUT, if we took every red bell, analyzed it as a group, made decisions, documented any changes to our system, recompile, that's one less of these you'll see.

Tons of work? Yep. However, your are solving the most frequent situations, and each one of these foundational bids will help the database (a little now, a lot later), and the database rules ALOT right now, and ALOT later too.