ESPN Insider general editor Chris Sprow looked at the first two rounds of the NCAA tournament last week using Insider's Bracket Predictor tool. This week, he uses the matchup-based Game Predictor to simulate the Sweet 16 through the final.
When the first try of Bracket/Game Predictor picked the Kansas Jayhawks to win it all -- and I picked my bracket in lockstep with the machine -- you can understand why I felt it necessary to perform Michael Bolton's Revenge on the computer after my bracket got Farokhmanesh'd. For the record, I forgave Bracket Predictor, but my Louisville Slugger couldn't. In the race against the machine, call it creative destruction. I'll innovate elsewhere.
But maybe I was wrong.
A closer look at the work of Bracket Predictor shows that, despite the trouble with Kansas, the computer is sure as heck smarter than most of you. Consider:
• The Predictor's Sweet 16 would still be in the 83rd to 85th percentile on ESPN's Bracket Challenge -- even with the Kansas loss. BP would be winning more than 90 percent of pools.
• It correctly called a trio of 10-seeds winning their first games.
• It sensed some Big East weakness, saying the Pittsburgh Panthers wouldn't make it to the Sweet 16, even as a favorite, and -- as I wrote last week -- it trusted Washington in a sim over Marquette, even when there was little statistical reason to.
• It had Cornell as basically a pick 'em in Round 1 and superior on other levels. Because it was so close, I felt compelled to take the higher seed just to be consistent.
• Lastly, in my favorite category -- genetics -- the computer had Kentucky with a 100 percent seed difference over ETSU, and was vindicated when Kentucky got off to a 100-71 run to start the game and didn't look back.
That said, it's time to play out the Sweet 16.