The names on the project that I recognize are Hal Varian, an economist at Berkeley, who has written about prediction markets before, and Patri Friedman, son and grandson of David and Milton respectively, and a regular on the Catallarchy blog, who’s currently working for Google.
They didn’t say anything about the form of the market, or what software they’re using (homebrew, or borrowed from one of the existing providers.) Patri says that his part of the project was analysing the data in order to show that it has some predictive utility, and (gasp!) the predictive power gets better as the deadline approaches.
I will have to talk to some of the people I know at Google (everyone in Silicon Valley knows a few people at Google, I’m not claiming to have any special connections) and find out what they’re doing, and how it’s working for them.
There are no hints that they are considering setting up public markets. There are plenty of reasons for large companies to run internal markets, and software companies in particular can benefit from better predictability of their software development efforts, and better tools for aggregating opinions about what is likely to succeed in the marketplace.