This project was done for the Master course Multi-Agent Systems. The application
models the Da Vinci Code game.
People can play this game, using one computer. The disadvantage of this is that
to be able to play the game correctly, the players have to switch a lot between
who is sitting behind the computer. It was our intention to make an application
with which people can play this game via an internet connection.
However, making this game was already more work than this project entailed.
Therefore we recommend adjusting the game for multi-computer use as a future
(research) project.
When modelling this game, we found out that actually less knowledge is necessary
than we first thought. During this game, one can reason about what other agents
know. However, this is not necessary to be able to find the codes and it will
not improve performance of an agent. Therefore this was not implemented in the
game.
One can argue whether common knowledge is useful. In this game the common
knowledge is represented by all tiles that lie open on the table. We implemented
common knowledge in prolog (see
common_knowledge.pl). However this did not improve the performance of agents
(it did not decrease the number of possible codes) and we decided not to use
it in the application.
The knowledge in our system can best be found in the hints we give, while playing
the game. These hints actually show what a true logician would be able to deduce.
We would have liked to implement more knowledge into this game, but as it turned out
that this was not necessary for good gameplay and as the amount of time for this project
was limited, we were not able to do so. However, we think that the application as it is,
already shows that people barely are true logicians and in order to be a real logician you need,
besides good reasoning abilities, also quite a big working memory, to store all possible
worlds, all possible values for all tiles. The true logicians in our society, should not
need the hints the application gives. However, most people probably do.