by Michael Cain
It was recently announced that James Hanley’s Post-Zombie Apocalypse Constitutional Convention class is going to happen. I think this class is sufficiently unusual that it offers an outstanding opportunity to test the prognostication power of the people at Ordinary Times. Hey, I did predictions professionally for a number of years and am always on the lookout for a chance to encourage other people to look equally (in)competent. Winners get bragging rights, such as they are. The purpose of this post is to determine whether there’s sufficient interest and to solicit suggestions for how to structure a long-term slow-motion contest. I’ll provide a brief discussion to start things. Comments to the effect of “Mike, you need to get a life” are responsive.
The goal is to make accurate predictions about the behavior of the students and their final results. Predictions can be broad, as in “The students won’t even manage to get organized.” Predictions can be very specific, as in “LGBT rights will be mentioned specifically in the draft Constitution.” There ought to be multiple categories for winners and losers, such as best predictor, best prediction, worst predictor, worst prediction, and most interesting prediction whether it was right or wrong. Scoring could easily get out of hand. Should there be a bonus for a good prediction made early in the process rather than later? Should the first person to make a particular prediction have exclusive rights to it?
There are a number of ways things could be organized. At one extreme is a “closed envelope” arrangement. Participants submit predictions via a small web service, where they are held in secret. At the end of the semester, the judge(s) sort everything out and announce the winners. The disadvantage of this approach is that it does away with the whole community thing, the opportunity for ongoing public accolades and/or humiliation. I participate at Ordinary Times because it has an excellent community, so I’m not particularly thrilled about this approach. The advantage is that it’s straightforward; the whole input system could be coded up in an afternoon.
Another option is to try to run it the way questions such as “What’s the must-have album for a road trip?” used to be posed, through the normal comment mechanism. Someone goes through the comment stream, parsing out the things that are predictions and collating others’ +1 and -1 scoring. James has indicated that he will be providing at least irregular updates on the progress of the class, so there would be at least that many posts to go through. Absent considerable user discipline, predictions might appear at multiple levels in the comment structure. The current threading allows limited depth so matching scores to predictions might be hard. Trying to automate things like matching scores to predictions and identifying duplicate predictions might be an interesting AI problem.
Yet another possibility would be to parse out the predictions and put those up on another site that supports an appropriate community scoring system. I’m thinking broadly along the lines of Slashdot’s moderation points here. People could drop by and rate predictions (based on their interpretation of James’ posts) by correct, incorrect, too soon to tell, and interesting. Again, there’s a mixed set of advantages and disadvantages. Automated extraction of predictions from the Ordinary Times site is feasible, at least to the degree that one of the contest rules could be “Predictions not prefaced with the string ‘Prediction:’ won’t be entered.” OTOH, scoring is only partially tied into the community. Users have to go somewhere else to do it, and lose the opportunity to provide instant feedback to the person who made the prediction.
This is a harder problem on multiple levels than I thought it was when I suggested it as a throwaway remark a couple of weeks ago.