There's a certain kind of "research" that consists of spending long hours in the library, or with EDGAR, or what have you, and digesting the available information to produce useful summaries of the current state of things. Aaron Swartz is looking for someone to do this sort of research on political issues, at <http://www.aaronsw.com/weblog/researcherjob>.
I think there's an unfortunate lack of this sort of "research" being produced and made public, and much of what *is* produced is being produced by partisan research institutes, which limits their credibility somewhat. (The job Aaron is offering is partisan, too; he cites a political orientation as one of the job requirements.) Supposedly this is one of the purposes of journalism, as well, but self-described journalists seem to be pretty terrible at carrying out this kind of research, by and large. So the following mechanism occurred to me as a way to aggregate demand for such research. A research institute sets up a web site where anyone can post a request for a report analyzing some issue, coupled with a pledge to pay an amount of money of their choosing if the institute produces such a report. Visitors to the site see a list of open requests and previously produced reports, and can pledge their own money to any open request. When someone at the institute finishes a report, they post it on the site for anyone to read, and the institute calls in the pledges on that report. Then that person chooses a new report to work on: the one with the largest total amount pledged, or perhaps the largest total amount pledged per hour that they estimate it will take. In a way, this is similar to Sourcexchange, CoSource, pubsoft.org (the Public Software Fund), eLance, and so on, except that there is only a single provider of what is being funded --- so all the issues of bidding, choice of service providers, and quality feedback are greatly simplified. It's a sort of auction: the next week (or whatever unit) of the researcher's time is "auctioned off" to the *issue* with the highest *total* bid on it. Contributors influence the priorities of the organization by submitting bids. The Dominant Assurance Contract Variant --------------------------------------- The above also bears some resemblance to an assurance contract: nobody's pledges are called in until there's enough money pledged to fund work at the institute's usual level of quality (whatever that may be). The incentives are slightly different, since whatever the amount currently pledged happens to be, you can increase the likelihood that your desired report will be the next one produced by putting in money; and there's no point at which any particular report definitely fails to get written, just a matter of being indefinitely postponed. But there's still an incentive to "free-ride": as long as the reports are made available to the public, you still get them at about the same time if you don't pay for them. On the other hand, only providing the reports to those who paid for them destroys the vast majority of their potential social value, and it also damages your institute's ability to market itself to potential new funders. Alex Tabarrok came up with a variant of an assurance contract in which, if the contract fails, everyone who pledged money gets a small amount back. This is supposed to give people an incentive to pledge money to any cause that they think will fail. He analyzes it in <http://mason.gmu.edu/~atabarro/PrivateProvision.pdf>. I've written about these before in <http://lists.canonical.org/pipermail/kragen-tol/2005-June/000783.html>. I think you can apply the same idea here: if the institute wants a particular report to be produced, it can offer an up-front payment of, say, 10% of your pledge, in exchange for you making the pledge --- sort of like buying a put option from you. This way, as long as the report hasn't been produced, you're ahead financially, so you have an incentive to pledge money to a report if you think it is unlikely to be produced. (Alex's paper envisions competing entrepreneurs funding different dominant assurance contracts. I'm not sure how that would work here.) -- To unsubscribe: http://lists.canonical.org/mailman/listinfo/kragen-tol