Sunday, August 9, 2009

computers, polls and why everything you know is wrong

Another college football season is on the horizon. Seriously.

And, while there are sooo many things about college football season we've grown to love — tailgating, rivalries, pretending we're still in college, etc. — there's at least one meme regarding college football that won't die: college football's postseason. Big-money conferences and their bowl tie-ins have for (at least) the last 20 years stood in between us as fans and the only satisfying end to any sports season: with the two best teams standing toe-to-toe in the arena, battling for the championship. We simply can't get that in big-time college football, and it grows more frustrating with each passing season for those of us who love the sport — we're tired of defending it from its detractors (mostly hateful Northern writers who thrive in the lifeless world of the NFL).
Everyone has his own theory about how to make a better postseason (even I've unveiled 2 separate theories). Most people who write about this for a living say a plus-one is the best model (except, of course, Mike Leach, who's completely freaking nuts).

Nearly everyone, however, has one principal complaint about the BCS: "Computers shouldn't determine our national champion." And they're absolutely right.

The truth of the BCS formula — a complicated mathematical equation that only a handful of people in the world can fully understand — is that it does incorporate computer-based rankings, along with human polls, to determine the top 2 teams in college football. Just about everybody thinks it's stupid that computer geeks should have more of a say in determining that top 2 than, say, a room full of dorky sportswriters or a group of coaches, most of whom don't actually vote anyway.

That's where they're wrong. The computers are the most reliable and objective methods of analysis we have at this point.
(Sidebar: I turned on Comedy Central's Roast of Joan Rivers — right now, Tom Arnold is bombing terribly. It's painful, frankly. Barely a chuckle in the entire room.)

Here's what I mean by "reliable and objective": assuming the formula is entered correctly, a computer is a better selector of the two best teams in college football than any human being. Why? Because human beings see uniforms, tradition, fan bases and dollar signs — computers only see the relevant criteria.

It all comes back to the concept (trademarked by the NCAA Basketball Tournament) of a "blind resume." Never, never, never judge a team by its history — judge it solely by its body of work. For example, the 2008 Utah Utes, who gained steam as one of the BCS casualties after drubbing 12-1 Alabama in the Sugar Bowl. The lone undefeated team in the country, Utah's merits as a potential national champ — or, at the very least, a title-game participant — were shouted from the rafters all across the country.
But here's the truth about Utah: they didn't get a shot at the national title, because they didn't deserve one.
Seriously. Look at the Utes' 2008 schedule. Their biggest victory prior to the Sugar Bowl was a 3-point win over Oregon St. in Week 6. They sweated out a 2-point win over a Michigan team that finished 3-8, and needed TCU's kicker to turn into Ryan Pflugner to survive a 3-point win over the Horned Frogs in Week 10. And the rest of their schedule was spare: Weber St., Air Force, BYU, etc.
So you would them in the national title game? Really? Over who, exactly? Florida, whose only blemish was a 1-point loss to Ole Miss (who would also have drilled Utah)? Oklahoma, who only lost to third-ranked Texas at a neutral site? Or Texas, the team that lost on the road at Texas Tech on a last-second touchdown? What about Alabama, the team that lost once to (yes) Florida, in a game they led by 3 in the fourth quarter, also at a neutral site?
All four of those teams played tougher schedules than Utah. All four of them were more deserving candidates for a national title shot.
(Sour grapes from a 'Bama fan who was forced to congratulate a bunch of gel-haired idiots on Bourbon Street on Jan. 2? Maybe. But the facts are still the facts.)

Which is why the computers exist. Computers don't get caught up in nice stories, the way the voters did back in 1984 with BYU, as chronicled here by the guys at Braves & Birds. Computers aren't bogged down by shiny uniforms or East-Coast bias or any of the other garbage that affects our judgment. Computers deal only in cold, hard data. They're better at it than we are, and deep down, we know it.

Of course, we could get rid of all of it if we settled in the most object way possible. But that's probably asking too much.

No comments: