MGoBlog has moved. The new site can be found at

Monday, October 23, 2006

Look: I know with the BCS standings coming out and so forth and suchlike that it's the time of year cranky people who don't know an algorithm from their ass (Attention Dennis Dodd: an "algorithm" is the one you don't write your columns with) to go "blah blah blah computers blah blah hate blah blah Injustice In One Particular Poll blah blah prune juice blah where's my adult diaper." It is traditional column fodder for the lazy, an opportunity to get some easy whacks in on a much-loathed institution. It's like a Miami brawl that comes only once a year.

But I know computers. I whisper in their ear and get them to whinny out poll results every week. I feed them sugar cubes and they make graphs for me. I have a vague idea of the difficulty in measuring teams when given all the nuances my pretty little head can handle and can't imagine trying to sort things out if all I was given was a set of wins and losses. And I say this to you, crabby BCS-bashers of the world: the seemingly nonsensical arrangements of numbers that purport to be polls from Sagarin or Massey or Colley and his matrix are not the fault of computers. Computers are good people.

Who is at fault then? As any good Bloom County fan knows: this is... THE MEDIA'S FAULT!

And so it is. Back in the old days when the BCS formula was a hilariously overwrought thing that included two human polls, computer polls, strength of schedule, quality victories, a loss column, and everything short of an "artistic merit" score for the uniforms*, the computers were allowed to take margin of victory into account. This caused no end of bitching. Coaches whose teams ended up on the wrong side of late touchdowns complained that opponents would run up the score despite the fact that computer margin of victory was capped at 21. Sensing a good opportunity to rattle off 600 research-free words, columnists everywhere leapt on the bandwagon. Think of the children. Kill MOV. Nonsensically, this was done, and the yearly columns pointing out the various OUTRAGES perpetrated by the computers got a little more ammunition as the rankings got a little less reasonable.

This is, of course, completely mad. The BCS found itself faced with a problem: the computers could not distinguish meaningful scores from cosmetic points scored after the game was decided. The powers that be decided that the solution to this problem was to take information away from the computers instead of providing enough information to judge which scores had an effect on the game. Now we have the half-measure of relying on crippled computer rankings that aren't allowed to take even the simplest facts about a game into account. This is bound to lead to ludicrous results. Could you rank the top 25 if you had no idea that Clemson waxed Georgia Tech and ND scraped by? Or that Tennessee stomped Cal? I can't make sense of my weekly BlogPoll ballot and I know not only the final scores of all the game but how accurately they reflect the flow of the game. If I were given only a sheet of wins and losses, my ballots would be as inane as Sagarin, et al.

I must reluctantly agree with the general opinion of computer rankings as a communist plot designed to undermine college football. As they stand they are useless and should be excised. But put the blame in the right place: with the people who decided to give computers the bare minimum amount of information. There is another way. Feed them as much information as they can handle. Let them take into account scoring margin, the time at which the scores occurred, and how they came about. Give them reams of historical data and let them judge what attributes a winning team has and let them project the past into the future without having to resort to the logo on the helmet.

Then when they rank six different I-AA teams over Miami you can take them seriously and bet on Duke to cover. And lo, it will be good.

*(though the secret inclusion of such might help explain that whole Nebraska-over-Oregon thing.)