As CommerceNet’s latest hire, and Principal Investigator for a new project (“Zocalo”) within the lab, I wanted to introduce myself to the readers of this blog.
I have been interested in Prediction Markets (AKA Idea Futures) for quite a while. Over the last few years, I have been working with a handful of other people to find ways to promote this technology. As a software developer, I have built a couple of prototypes on my own to explore different directions.
When I heard late last year that CommerceNet was starting a lab to explore decentralization and ecommerce, I was very excited to find that they might be interested in including Prediction Markets as a part of the lab. After some preliminary discussions, CommerceNet agreed to fund the development of a more specific proposal and its presentation at the DIMACS workshop on Markets as Predictive Devices in February at Rutgers. The proposal has now been issued as CN-TR-05-02: Zocalo: An Open-Source Platform for Deploying Prediction Markets. The presentation I gave at the workshop is up on the DIMACS web site. And my job is to make it all happen.
The goal of the project is the development of an open source toolkit for creating Prediction Markets and of a community of interested users. Prediction Markets are markets based on financial securities that pay the holder an amount that depends on the outcome of particular events. Markets in these artificial commodities generate forecasts that have been shown to improve on other prediction methods (pundits, expert panels, polls, focus groups) whenever participants have relevant information about the likelihood of the events. These markets decentralize decision-making by creating reliable predictions based on wide-spread expertise from people who wouldn’t be consulted in any conventional forecasting procedure.
There is an active body of research into the structure and effects of these markets. Applications include forecasting (project schedules, level of sales, election outcomes) and making decisions (which pharmaceuticals to take to trial, which flu vaccine to manufacture this season, what mix of products to build). Trials of prediction markets are demonstrating their usefulness, but people who want to make use of them are handicapped by not having access to tools that make them straightforward to deploy. Zocalo is intended to fix that problem.
Our initial users will be experimental economists, because that is a forgiving environment, but one that will push us to build the market infrastructure on a principled foundation. Economists seldom have access to professional developers, and usually develop new software for each experiment. We can help by providing a reusable framework that can be adapted to many uses. Later applications will include trials in industrial settings and eventually public markets for prizes or real money.
We have started working with economists at George Mason University to develop software that will support an experiment they are planning to run in April. Later projects are not yet scheduled; I’d appreciate contacts from economists who would like help setting up experimental prediction markets, or others who are interested in exploring deployment of prediction markets to answer important questions within your organization. As we proceed, I expect to lay out more details of what tools we want to build next, and the kinds of pilots and experiments we hope to participate in.
It’s great to be working in this exciting environment with the quality people I find around me. CommerceNet seems ideally situated to contribute to development of the next stage of Internet commerce. I hope Zocalo can play a big part in that.