Yes, there are many flaws in my calculations at this point. I too noticed that the Hersey values seemed most suspect, for the reasons you mentioned. I've been able to predict local races now for several years (check out my predictions for 3A and 4A State Championships for Washington State), but I haven't gathered enough data yet to apply the same level of rigor to the nation-wide calculations. I'm a bit premature in releasing this, but I threw caution to the wind and thought I'd give it a shot anyway. It takes a lot of time to do these types of calculations, but I'd like to partner with Dyestat or Athletic.net to make the calculations available to the world throughout the season. This will require the services of a computer programmer to harvest the data from Athletic.net so that I can run the calculations.
Once the calculations are done, runners can compare their performances across the nation without many of the drawbacks of comparing raw times.