My favorite family legend involves my dad's baby sister—my "Тетка," in Macedonian—Bonnie. As the story goes, Bonnie's first-grade class organized a Brownie troop, and their first act was to elect a troop leader. When they counted the votes, Bonnie was the only girl who hadn't voted for herself. Upset at this outrageous display of self-interest, she walked out and in tears announced to my баба she'd quit the Brownies.
This week, Oklahoma State fans and media members have voiced outrage at Nick Saban's "attempt to manipulate the system" in the final, deciding USA Today coaches' poll, in which he ranked the Cowboys fourth, behind Stanford. The narrow margin Alabama kept over OSU in the final BCS rankings—determining LSU's opponent in the BCS championship game—was influenced to a very small degree by Saban's choice (which was shared by five other coaches). Upset at this outrageous display of self-interest, SEC radio legend Paul Finebaum went on Twitter and declared Saban's reputation "cheapened."
Of course, you could reasonably argue that Saban's reputation is for doing whatever is best for Nick Saban. But that's the interesting thing about his ballot, and about the coaches' ballots in general: Strictly speaking, he didn't vote in his own self-interest—not entirely, at least. A coach in Saban's position, acting rationally, would vote his own team No. 1 and leave unranked any team that might threaten his access to the championship. That would be selfish. From an economics standpoint, Saban acted irrationally. Given the enormous stakes of postseason play—for coaches, for their programs, for their conferences—it's worth asking why.
* * *
According to the American Football Coaches' Association, with whom USA Today administers the poll, UPI first developed the idea of using coaches to rank the nation's top teams. This was mostly to create a competitor to the Associated Press's noted national poll of the times. UPI tapped five coaches from each of seven national regions, increasing the pool to 50 in 1986. The board currently includes 59 coaches; each conference is permitted to empanel half its coaches, which is done randomly according to USA Today. Some coaches beg out of the process, but statistics would suggest that overall turnover in the board would be high, especially given the rate of head coach turnover in FBS overall.
Despite this, only 16 active coaches who voted in 2010 lost their vote in 2011—and some coaches, like Steve Spurrier, have held onto their vote on a yearly basis (he's been on the board every year since he returned to college football in 2005). If the coaches truly are selected at random, the OBC should hit up a casino. The odds of his being randomly selected as one of the SEC's voters that many years in a row is .8 percent. (Nick Saban's been a voter since 2009, while Oklahoma State coach Mike Gundy has miraculously been passed over for board selection since 2005.)
How coaches gain entry to the AFCA Board of Coaches is a process long shrouded in mystery, and the organization itself only promised to use a fully random selection scheme in 2009. Meanwhile, an article in Contemporary Economic Policy (publication forthcoming) claims the AFCA leaves board nominations up to the conferences themselves. (USA Today denies that, according to Pollspeak.com.)
Of course, none of this mattered much when the poll was for entertainment purposes only. Since 2004, however, it's been 33 percent of the formula that determines which teams compete in the BCS championship game and earn the lucrative payday that comes with it.
* * *
Coaches have had several years to learn how to vote optimally and actually manipulate the system, but none of them do. Nick Saban isn't the only one. Every other SEC coach on the board (Spurrier, Auburn's Gene Chizik, Vanderbilt's James Franklin, LSU's Les Miles, and Georgia's Mark Richt) chose not to vote with the optimal strategy. Oklahoma State coach Mike Gundy isn't on the board, but Baylor's Art Briles, Iowa State's Paul Rhoads, Oklahoma's Bob Stoops, and Texas Tech's Tommy Tuberville all should have voted OSU No. 1 and left Alabama unranked if they had been truly acting in their own best interest.
The real ball-droppers, however, are the 26 coaches of non-automatic-qualifying teams. If all 26 had entered first-place votes for TCU and left every team ranked above it in the BCS standings unranked, they could have pushed the Horned Frogs into the BCS top eight. This would have guaranteed TCU a BCS at-large invite; since non-AQ teams share their BCS bowl payout with all non-qualifiers, it would have maintained a revenue stream they've had since 2006 (the last year without a non-qualifier in a BCS bowl).
Your rebuttal might be that these coaches would all suffer some form of punishment for their manipulation of the polls, including a permanent ban from future voting. But this makes two assumptions: First, that the coach will be selected to vote in an upcoming year (uncertain for those not named Spurrier); and second, that a coach will still be working at an FBS school the next season. Even if you accept both as probable, the return in value is unlikely to rise significantly, nor are the modest benefits of being a poll voter comparable to the benefits gained from manipulating the system (especially if you are Nick Saban).
So why didn't Saban pull his Grim Trigger? Why, indeed, has every voter since the creation of the BCS acted irrationally and chosen a suboptimal voting strategy? Some possible reasons:
1. Altruism. This is how most economists explain situations in which people in the real world diverge from what should be their optimal strategy, but I don't think that applies here for two reasons: First, altruism usually comes into play when discussing sacrificing one's own benefit for the common good (i.e. charity), whereas college football is a zero-sum game akin to war; second, Nick "I'm not going to be the Alabama coach" Saban is not an altruistic person. (He does have a charity fund, but unless its name is the Society for the Preservation of the Big 12, it wouldn't apply here.)
2. A fear the optimal strategy will lead to the death of the coaches' poll. This, too, is unlikely, given how some coaches themselves say it shouldn't be a factor in the BCS, others consider filling out the ballot a nuisance, and many delegate the duty to someone else.
3. Ignorance. Saban doesn't recognize the losses at stake, nor does he see his optimal strategy to avoid those losses. Nor, for that matter, does he recognize the opportunity for other coaches to collude and influence the poll in their favor (this is, in game theory terms, a game of Deadlock in which all parties benefit by defecting regardless of what competitors do). This is probably the most likely, though I'm still skeptical because college football coaches are very smart men, many of whom hold degrees in math, philosophy, psychology, or theology, courses of study that would have exposed them to the very economic theories discussed here. But, then, football coaches are terrible at choosing optimal strategies on the field, too.
* * *
Push a little, and every aspect of conventional balloting logic begins to wobble a little. Let's go back to Saban: As you know, he ranked his team behind LSU. In other words, he stated that he believed LSU was the best team in the country and that Alabama was not—yet he maintains Alabama ought to have the opportunity to show they are better than LSU. If he thinks his team will beat LSU, then he must believe his team is better than LSU—but his ballot suggests otherwise. If a coach doesn't think his team will win, why are they even allowed to play? And what will Saban say if the Crimson Tide win the championship? "I guess we're better than I thought we were"?
It's ridiculous given the impact the coaches' poll has on the money and prestige of BCS bowls that any coaches are excluded from voting, let alone more than half of them. (AFCA's response: "It's hard to collect 59 ballots.") The idea of sampling coaches comes from an era during which games were not even televised, and to use such an antiquated approach to determine a national champion is simply ridiculous. If the poll and its influence must be kept, every team needs a vote.
But polls themselves have somehow mutated from a measured approximation of each team's ability (relative to other teams) to a ranking of accomplishments, like college football is a Modern Warfare 3 leaderboard. Yes, game results are evidence suggesting the ability of a team, but they have an alarming lack of reliability from a statistical standpoint. Boise State coach Chris Petersen has every right to believe his team is the best in the country at this moment, yet he ranked them fifth because of a loss that happened weeks ago. The coaches' poll is full of weird niceties and curtsies to some sort of convention, but all anyone wants to talk about is the cold-eyed self-interest that is neither as cold-eyed nor as self-interested as critics would have you believe. Coaches vote in ways that are irrational, illogical, and contrary to the interests of themselves and their universities. If we're going to force them to rank teams, after all, shouldn't we let them claim their own team is good?