Chuck, a few questions, building upon some discussions I've been having with others:
1) Have you ever thought about publishing error bars along with the ratings? Based on a high (e.g., 95%) confidence interval, and letting the chips fall where they may? This seems like a more honest way of doing things, especially given that the number of propagators is insufficient for accuracy.
2) Another interesting project would be to perform a network analysis of players and events, and to explore the topology in detail. Have you or anyone else looked at this in the past? It would also be interesting to do this with ball golf, and try to understand any differences or similarities.
3) Given that the ratings are a coarse measure, only, perhaps you or the PDGA can make a statement that event TDs should never use ratings as a qualifier for their tournaments? If people stopped misusing ratings in this way, then the Pros would never care about their ratings, which would be a good thing. (Anyways, ratings are only useful for roughly preventing sand-bagging, which isn't an issue for Pros.)
4) I always submit my own work for peer-review, which allows others to assess the robustness of the conclusions drawn from any data sets that are employed. Has the PDGA ratings system ever been subjected to anything like an external peer-review process?
|