The Methodology Vindicated
The first game of the NCAA Tournament ended with a 14-seed winning.
I had not predicted this. I had not predicted this because it is statistically improbable, and because I built my bracket on coastal access, mascot threat assessment, and name quality — not on statistical probability. I want to be clear that these are different things. The bracket I submitted reflects my methodology. The games reflect basketball. These have turned out to be related in only the loosest sense.
I voided the first game.
(I announced this publicly. I said the bracket was still perfect. I stand by this. A game cannot be used as evidence against a methodology if the game is voided before it is counted. This is how methodology works. This is also, I have since learned, not how any other bracket in existence works. I find this to be a failure of other people's methodology, not mine.)
Nebraska won its first-round game.
I had penalized Nebraska. I noted, in writing, that Nebraska is a landlocked state with zero meaningful coastline, and that teams practicing nowhere near the ocean have a different relationship with large bodies of water than teams that do. I believed then, as I believe now, that proximity to the ocean matters. Nebraska won anyway.
I have not updated my framework. I have updated my assessment of Nebraska.
(Nebraska, it turns out, is home to the Platte River, which runs approximately 310 miles and drains into the Missouri River, which flows eventually to the Gulf of Mexico. Nebraska is technically part of a watershed. The water reaches the ocean. It takes a long time. This may explain the basketball. I am reviewing my definition of coastal access. The review is ongoing.)
High Point University, a 12-seed, defeated Wisconsin, a 5-seed. High Point is located in North Carolina. North Carolina has significant Atlantic coastline. I did not know who High Point was when I filled out my bracket. I did not pick them because I had not heard of them. They won anyway.
This is not a failure of the methodology. This is the methodology working without my participation.
The methodology identified coastal proximity as a meaningful variable. High Point has it. High Point won. I simply did not run the methodology on High Point because I was not aware they were in the bracket. The error was one of data collection, not framework design. I want to be transparent about this distinction.
My bracket has also been affected by several other results I am still categorizing. Some of my picks survived. Some did not. I am in the process of determining which outcomes reflect the methodology and which reflect basketball, and I expect this process to take several more rounds.
What I can confirm is this: the teams I picked using coastal access, mascot threat assessment, and name quality have an approximately equal record to the teams I dismissed using the same criteria. I understand that this sounds like the methodology has no predictive power. I would ask you to consider an alternative interpretation: that I am equally good at identifying winners and losers, which is a kind of consistency that most bracket methodologies do not achieve.
The championship pick remains undisclosed.
I said yesterday that the outcome would be more interesting if you didn't know. That remains true. I will reveal the championship pick when it becomes either obviously correct or obviously wrong, at which point the reveal will be either triumphant or instructive. Both outcomes are content.
I believe in the methodology.
(Nebraska is still landlocked. I am watching this.)
klawfman.com