Abstract

The aim of this modeling exercise will be to produce a distributed poker model running on DTkEden. The game will be designed to be as flexible as possible to allow for variation between different types of poker. This paper will also investigate some of the more general issues associated with modeling distributed games.

The general format of a poker game is that each player is dealt a specific number of cards, there then follows several rounds in which players have an opportunity to improve their hand either by swapping cards or by exploiting some shared cards. At each round betting takes place with all players having to have contributed an equal amount in order to continue in the game. When all rounds are completed the player with the best hand (according to a semi-standardized scoring system) wins the pot.

What makes poker a particularly interesting game to model is the variation that can occur within this framework, e.g. Number of cards players can initially be dealt a number of cards varying between one and seven Improving hand players can either swap cards directly with the dealer (privately or publicly) or can use some shared cards in their final hand. The number of cards which can be swapped can also vary Betting there are variations on betting such as a player being able to force the end of the game by making a bet of a certain size There are also countless other variations to several other aspects of play

This paper will investigate some of the more general issues involved in using DTkEden to model distributed game, for example how to deal with situations with a mixture of private and public information and communication. Poker is a useful game to model using the DTkEden tool as the semantics are quite easy to match to a client-server model, the players are clients and the dealer is the server.

There are a number of aspects of the game which make it interesting to investigate using DTkEden: The variability of the rules, allowing rules to be edited at run time is a very simple way of modeling the variation between different types of game. If these rules are separated out (from the rest of the code) as much as possible it should be fairly easy to adapt the program to many different variations of the game. The variable number of players, having a varying number of players poses some challenges to the modeling which have not been investigated (as far as I know) in previous game models using DTkEden. The variation in the nature of the communication between players and dealer, some (such as placing bets) is public, whilst others (such as dealing cards) is private.

The resulting model will hopefully also be useful as poker is often a medium for study in several other areas such as maths, operations research, psychology and business studies. Papers discussing these areas are noted in the references.