You keep trying to draw a distinction between what I want it to capture and what it should be capturing that doesn’t compute for me.
I guess I disagree that is what KenPom and Sagarin are doing (injuries being a major counterpoint).
To expand a bit, KenPom and Sagarin (and as we’ve discussed) 3MW treat every game like it’s equivalent to every other game. But humans know that the most important games are in the tourney, and teams and coaches and fans (and all of Sumner Redstone’s crazy kids and whoever owns TBS/TNT these days) all know that, too, so those games are higher stakes and efficiency margins aren’t necessarily the best predictor we have for that. I guess I’d trust the computers to know who the best 200 teams are, but I still trust the humans to know who the best 4 teams are.
I’m saying there is no objective ‘what it should be capturing’ lol you want it to capture who is most likely to win the national title but that isn’t objectively what it needs to be or should be. That’s a preference. I prefer a weekly power ranking to see who is playing the best at the moment
Right. There’s an alternative score-keeping system the voters are implicitly using to re-grade their votes each week that sort of parallels an actual power-ranking of the current best team and people’s predictions on winning the title, but has some really obvious differences. Everyone understands this is the case and people discuss it all the time, but never actually sit down and explain it because its considered obvious.