2018-04-17 Andrei Dubovik gave a talk at Tilburg University about the project (slides).
2017-09-08 Our first tournament was held at Erasmus University Rotterdam. Vladimir Karamychev won the tournament.
Many real industry settings remain unsolved in economics. Consider a typical industry:
- firms vary in sizes,
- there are capacity constraints,
- firms can invest in capacity and divest capacity,
- firms compete in prices in multiple periods,
- financing costs matter,
- demand is uncertain.
Should a given firm invest in extra capacity? What prices should it set? Should the antitrust authority allow or block a merger? This is just one example out of many, where modern economic theory does not give a satisfactory answer. Often there are no theoretical predictions, in part because no polynomial algorithms are known for numerically computing Nash equilibria. And when theoretical predictions exist they often involve complex mixed strategy equilibria that are sensitive to the specifics of a model.
Eliciting behavioral patterns is not possible as such settings are too complex for lab experiments. Real life data is noisy, sometimes non-existent, and mostly unstructured. So why not take the middle ground by organizing an AI challenge with applications to economic games? Broad AI challenges are common today: some run by hobbyists, some by universities, some sponsored by tech companies. Then why not an AI challenge specialized on economic games? One possible answer: economists cannot program. But maybe some of us can?
With some colleagues we are nourishing an idea to start playing economic AI games. We all have better things to do, so we are taking it slow and lazy. However, the plan is to have a first few tournaments between ourselves during the course of 2017. The programs and the resulting data will be posted on this website.
You are free to browse through the ongoing work on this project, but editing is restricted to registered users. If you would like to join the competition, please email Andrei Dubovik at firstname.lastname@example.org.
Andrei Dubovik and co.
- Technical Organization (all technical aspects, e.g. the programming language, version control, are discussed here)
- Suggestions / Considerations
- Technical Issues (i.e., bugs, feature requests; I have just installed this wiki and surely there are / there will be some)
- Literature Overview
- Mailing Lists