WikiSim: simulating knowledge collection and curation in structured wikis.

The aim of this work is to model quantitatively one of the main properties
of wikis: how high quality knowledge can emerge from the individual
work of independent volunteers. The approach chosen is to simulate
knowledge collection and curation in wikis. The basic model represents
the wiki as a set of of true/false values, added and edited at each
simulation round by software agents (users) following a fixed set
of rules. The resulting WikiSim simulations already manage to reach
distributions of edits and user contributions very close to those
reported for Wikipedia. WikiSim can also span conditions not easily
measurable in real-life wikis, such as the impact of various amounts
of user mistakes. WikiSim could be extended to model wiki software
features, such as discussion pages and watch lists, while monitoring
the impact they have on user actions and consensus, and their effect
on knowledge quality. The method could also be used to compare wikis
with other curation scenarios based on centralised editing by experts.
The future challenges for WikiSim will be to find appropriate ways
to evaluate and validate the models and to keep them simple while
still capturing relevant properties of wiki systems.