Rette Sich Wer Kann is a strategic and challenging board game
where the players, through a series of elections, try to rescue their
own crewmembers from a sinking ship and win the game. In each
election, the players will need to negotiate with their competitors
to find enough support to ensure a favourable outcome. In the end
though, only one player can win the game. Hence the players are
forced to cooperate in order to compete.
Another aspect of Rette Sich Wer Kann is the possibility of deceit
as a promise made in the negotiations is a promise that may be broken.
Hence the game deals with issues of deceit and trust.
These interesting aspects of the negotiation in Rette Sich Wer
Kann makes it an interesting game to study for Multi-Agent Systems.
The first part of this report deals with the implementation of Rette
Sich Wer Kann in software in order to create a platform on which
rtificially intelligent agents may play the game and be the subject for
Multi-Agent Systems research.
The second part of the report details the agents that has been
created to run on the aforementioned platform and evaluates their
performance. The most interesting of those agents is the I-Owe-You
Negotiator which trades favours with other Negotiators in order to
win the game.