I like football. Not-so-much college football because of the constant bickering involved in "who is the best" type discussions. Although, I wish I had paid more attention in my statistics classes in school. I’m trying to understand this whole college football snafu in determining who is actually the number one college team.
I’ve come to the conclusion based on several years of programming experience that there is no mathematical algorithm that can define who is the national champion in college football.
The BCS is quite simply broken beyond repair. It is therefore quite an interesting study in corporate application development.
Looking at the BCS standings from a mathematical standpoint has me just stumped. I tried reading through Colley’s Matrix info but my eyes glazed over. Colley's rankings make up essentially 1/7 of the BCS calculation. In retrospect, I probably started with the wrong part of the BCS calculation, as there seem to be a few simpler methods out there being used.
But Colley makes a good point before he gets to all the mathematical jargon. There are 117 division I-A college football teams that each play about 11 games. Some of them play more than 11. This upsets the rankings. Some of these games may even be against division I-AA teams. This also upsets the rankings. Therefore, the opponents of each team for a season cannot be a statistically viable sample representation of the teams competing since teams compete on multiple skill levels (I-A, I-AA) and for multiple durations. Put another way:
There’s too many teams in college football with too much disparity between them and they don’t all play the same number of games.
The solution? Well, a lot of people want to get rid of the BCS computerized rankings. Some people want to change them to add in margin of victory. Some people want to add an 8 team playoff like in I-AA. (The problem there is how to decide which 8 teams though, so the BCS will again be involved).
This becomes an interesting business problem because this is not a problem that can be solved by throwing more data at it. This is a classic (but wrong) business system approach:
“We don’t think this data is always right. So, we’re going to fix it by throwing more data into it.”
It won’t help if they add certain items like “margin of victory” into the calculations. Any programmer can almost immediately come to the underlying conclusion that there is failure built into the system. The architecture, if you will, for determining the college football national champion will almost never provide a definitive champion. Quite simply, the BCS system is broken and needs to be re-built from the ground up. Much like a malfunctioning application, sometimes you have to know when continual debugging isn’t helping, and be confident enough to design and build a new system.
The solution to all of this BCS nonsense is three-fold.
First, they need to get rid of half the teams in division I-A football. Half of those teams are terrible anyway. Army. Vanderbilt. Western Michigan. Good gravy, my mom and her sewing circle could beat those teams.
Second, every team in the country needs to be in a conference and that conference needs to have a playoff system to determine the winner. Yes, this means you Navy and Notre Dame (if you are ever good again).
Third, you can’t let division I-A teams play division I-AA teams. It screws everything up. The math goes out the window and it is like comparing apples to pineapples.
Or of course, we could just say “who cares” and let those boys just earn a decent education instead of spending all this time, money and effort on trying to weed out the winners from the losers.
But who the hell wants to do that? It’s so un-American.