Relay-Version: version B 2.10 5/3/83; site utzoo.UUCP
Posting-Version: version B 2.10.2 9/5/84; site spp2.UUCP
Path: utzoo!linus!philabs!prls!amdimage!amdcad!amd!pesnta!pertec!scgvaxd!trwrb!trwspp!spp2!stassen
From: stassen@spp2.UUCP (Chris Stassen)
Newsgroups: net.rec.bridge
Subject: Bridge playing programs ("Bridge 4.0")
Message-ID: <601@spp2.UUCP>
Date: Wed, 15-May-85 13:14:10 EDT
Article-I.D.: spp2.601
Posted: Wed May 15 13:14:10 1985
Date-Received: Mon, 20-May-85 04:10:56 EDT
Reply-To: stassen@spp2.UUCP (Chris Stassen)
Distribution: na
Organization: TRW, Redondo Beach  CA
Lines: 58
Summary: Avoid "Bridge 4.0"


	A program called "Bridge 4.0" has just become available.  It
is the first program I have seen that will bid and play randomly
dealt hands (most other programs are tutorials - an electronic version
of "AutoBridge").  Bridge 4.0 is available for the IBM, Apple, and
Commodore computers.

	Review (of the IBM PC version):

	(1) The program is written in BASIC, and is interpreted.
	    It is SLOW, though compiling it might help.  It is also
	    "protected." (see "SAVE" entry in MS BASIC manual).

	(2) The program CHEATS.  When sorting hands, it counts the
	    total number of cards in each suit for each *partnership*.
	    It then decides what suit the final contract will be based
	    on maximizing (# cards) + (# points) per suit.

	    This leads to confusing auctions such as:

		Partner		You
		-------		-------
		  1D		  2C	(11 HCP, 5 clubs, 3 diamonds)
		  2D		  3D   
		  4C (!)          4D
		  5C (!!)

	    You partner has bid diamonds twice, you support him, and
	    he then suddenly decides that clubs is the best suit to be
	    in?  As it turns out, we had the same number of clubs and
	    diamonds between us, but one more high card point in clubs,
	    (AKQ, rather than AKJ).

	(3) Although I haven't been through the entire source yet, it
	    is relatively clear that the program cheats when it leads,
	    too.  It does too good a job of making the correct lead,
	    without any supporting information from the bidding.
	    It is such a small program (23K in "compressed" format),
	    that it couldn't do that well without cheating.

	Being rather disgruntled with the best the software market
has to offer (at the moment), I am setting off to write a program
that plays bridge.  I am working from the following requirements:

	(1) It must *NOT* cheat
	(2) It will support from 0 to 3 players (on extra serial ports)
	(3) It will understand and use a large variety of bidding conventions
	(4) At first, it will only run on the IBM PC (and don't expect it
	    to run in less than 264k).

	Any help that I can receive in the form of:

	(1) pointers to papers on strategy algorithms
	(2) suggestions for format/conventions/options

		Will be greatly appreciated.

				-- Chris