Can't trust computer Bg
Omar S Mamoun
In a way such a thing is inevitable, because even the world's most powerful
computers have difficulty solving backgammon problems. I have this game called
Hyper-Gammon, a backgammon derivative in which one plays 3 pieces against the
opponent's three. Its a CD game, and the author purports to have hooked up his PC
for two years to solve every single possible position involved with three
backgammon pieces. 32 million positions he claims his computer solved, and there
is this massive 172 megabyte data file on the CD, yet the damn thing still
cheats!! For some odd reason, the only doubles it has ever rolled were double-6's.
I'm thinking, what in the HELL is this?
I also bought a mini-backgammon computer called the "Micro Backgammon
Challenger," made by Fidelity. At level seven, when the computer rolled the dice,
it won by a landslide. When I manually entered the dice, the computer still
played fairly solidly yet collapsed like a deck of cards in the long run. Which
just goes to show that the sheer statistical dynamics of backgammon make it
impossible to trust a serial-processing backgammon computer.
ZZyzx
Mamoun) wrote:
>Yep: there's nothing more annoying than a computer backgammon game that cheats.
>In a way such a thing is inevitable, because even the world's most powerful
>computers have difficulty solving backgammon problems. I have this game called
>Hyper-Gammon, a backgammon derivative in which one plays 3 pieces against the
>opponent's three. Its a CD game, and the author purports to have hooked up his PC
>for two years to solve every single possible position involved with three
>backgammon pieces. 32 million positions he claims his computer solved, and there
>is this massive 172 megabyte data file on the CD, yet the damn thing still
>cheats!! For some odd reason, the only doubles it has ever rolled were double-6's.
>I'm thinking, what in the HELL is this?
>
If you knew Hugh Sconyers, the author of Hyper-Gammon, you would know
that cheating was not a part of his game.
Raccoon
ZZy...@ix.netcom.com (ZZyzx) writes:
>On 29 Jun 1996 00:20:41 GMT, osma...@mail1.sas.upenn.edu (Omar S
>Mamoun) wrote:
>>[alleges that the Hypergammon program cheats]
>If you knew Hugh Sconyers, the author of Hyper-Gammon, you would know
>that cheating was not a part of his game.
Agreed, Zzyzx. And this is this the first time I've ever heard anyone claim
that the Hypergammon computer-player doesn't play (roll) fair. I
seriously doubt it.
I might be persuaded that in an average game of Hypergammon, double 6
comes up more often than other double -- not because the program cheats,
but because in a 3-checker per side backgammon game, double 6 is more
likely to be a game ending roll than any other double.
Logicians, say it ain't so!
------------
Raccoon
Bob Koca
rac...@netcom.com (Raccoon) wrote:
>ZZy...@ix.netcom.com (ZZyzx) writes:
>------------
>Raccoon
>
It is true that 66 is most likely double given that was last roll of
the game. This does not mean that it occurs more than other doubles
however. There are various ways to explain this but one which attacks
the intuition used in th false argument is: since 66 is often likely
to win the game it is less likely than other doubles to be the second
to last or third to last roll... A more complete argument could show
that this balances out exactly.
,Bob Koca
bobk on FIBS
Stephen Turner
Raccoon wrote:
>
> I might be persuaded that in an average game of Hypergammon, double 6
> comes up more often than other double -- not because the program cheats,
> but because in a 3-checker per side backgammon game, double 6 is more
> likely to be a game ending roll than any other double.
>
> Logicians, say it ain't so!
>
Sorry, it ain't so. But the reasoning is quite subtle. (Proof which
should
satisfy any probabilists out there: the end of the game is a stopping
time).
--
Stephen R. E. Turner
Stochastic Networks Group, Statistical Laboratory, University of
Cambridge
e-mail: sr...@cam.ac.uk WWW:
http://www.statslab.cam.ac.uk/~sret1/home.html
"You may notice that your Customer Reference Number has changed"
British Gas
Anna Zueva
rac...@netcom.com (Raccoon) writes:
>I might be persuaded that in an average game of Hypergammon, double 6
>comes up more often than other double -- not because the program cheats,
>but because in a 3-checker per side backgammon game, double 6 is more
>likely to be a game ending roll than any other double.
>Logicians, say it ain't so!
It surely isn't so. There is no way the consequences of a particular
pair of dice can affect its probability of being rolled. Unless of
course, you are trying to prove Omar's original hypothesis...
...Zoiks
Stephen Turner
Actually, it's subtler than you think. If you count all of the rolls
except
the last roll of a game, 66 is rolled less often than 1/36 of the
time.