From: Brett Paatsch (paatschb@ocean.com.au)
Date: Tue Feb 11 2003 - 01:39:16 MST
Lee Danial Crocker wrote:
> Poker is the next challenge. The currently best programs
> are the University of Alberta's and a commercial simulator,
> neither of which is able to keep up with a moderately skilled
> human player. There is now, however, a financial incentive
> for creating better programs: internet casinos. There are
> probably bots playing them already, and they can make
> money because the vast majority of players aren't even
> moderately skilled. But it won't be long before some of
> them raise the bar and start beating the pros.
Interesting. I don't know much about poker but imagine real
good poker players would combine a sort of built in set of
probability tables, a good memory for what has already been
seen, and perhaps (and this is what I am not sure about in
terms of how important it is) an ability to read the facial
expressions, "tells" and body language of less skilled players
for additional hints.
To the extent this last - face and body language counts -
it is likely to add significant complexity to the design of
any "artificial player".
One thing seems likely, though, the "artificial player" would
have a pretty unreadable "poker face" and would hold up
well in the long sessions.
Anyone know to what extend the mix is numbers and
memory and to what extent its the capacity to read the
cues as opposed to the cards?
Brett
This archive was generated by hypermail 2.1.5 : Tue Feb 11 2003 - 01:16:17 MST