@MongooseMan Very interesting analysis, mirrors my own quest to understand the system. First of all, the seeding system is magical, it has some frustrations for the individual players, but for what it's intended to do it's incredible and I certainly couldn't do better. I'm going to add a couple of my own theories for your consideration. Sorry if you've mentioned them, I've honestly not read your opening post enough times to understand it perfectly.
My theory for the key frustration is that the system assumes that the average penalty for all riders is 0. By the time we get to the season opening, and when the average rider has generally spun up to being fit again, the average penalty for all riders is probably nowhere near 0, so your best fit straight line winningtimeXbeta line is artificially shifted for the average riders current index to be a good number of units higher than a fair reflection, and this provides the biggest frustrating feeling of "I feel like I'm riding like _ rider, but I'm only getting _.
Then, the base race. I think that setting CTCT as the base race gives it one important distinction, which I picked up from David's interview. The compounding penalty problem would continue to drive the theoretical 0 index time away from the actual winning time, but once a year we have our base race. If you look back at indexes, it's the only race where the winning time is the actual winning time, not the theoretical 0 index time. I think this effectively cancels out the compounding error. It's the global reset that removes the average error from the average index and fixes the system.
I wouldn't stand by any of this particularly steadfastly though. I have theories for why some rides just seem to work better than others. Thats a different discussion.
