Inner Machinations

Since it's been asked a few times recently, and since the website that used to host the page mysteriously vanished one day, here's a PDF of the final 2005 Chaos Rankings (which didn't include bowl data) and the formula explanation. Having watched two Terminator movies today, it reminded me of how uppity people get when computers output data that doesn't seem to mesh with their preconceptions. Usually, it's just a difference of opinion, but sometimes there are bugs. I've found and corrected a couple in the software I wrote to compute Chaos Rankings, but usually they were human-induced data-entry errors. Fortunately, I spent some time in the offseason writing a parser for ESPN's scoreboard, so that's much less likely to cause problems in the future. At any rate, the objectivity of computers has been discussed at length by other people, so I won't go into it (again), but the point is that computers (unless programmed to do so) don't care about the name of the school on a jersey (like so many voters) or preseason perceptions (again, voters). But if I'm going to make a stink about poll voters entries being made public, I also have to believe that computer ranking formulas should be made public. So here you go. Other than front end GUI stuff, data source (ESPN vs NCAA), and including bowl game data, the formula and weightings used in Chaos Rankings are unchanged from those listed in the 2005 information. The only other thing to note is that I utilize the previous season's final rankings as the new season's preseason ranks and weight them 66% and 33% over the first two weeks, respectively. This is mostly to prevent huge ranking swings as a result of early losses and cupcake scheduling. But rest assured, everything tends to settle out eventually, and that "preseason" data has no bearing on anything once week 3 data is available.