We’re using Ed Gunther’s analysis of the debate on a college football playoff as a framework to present my own analysis and opinions. In Part I I explained some of the history behind how we got where we are, and used mathematical analysis to show not only that a playoff wouldn’t make college football’s regular season less meaningful than any other sport, but that it’s quite possibly too meaningful now, because it gives us too small a sample size to properly compare teams in different conferences.
In his introduction, Gunther illustrates this with a discussion of two different notions of who’s “better”, and they’re fairly familiar: better as determined by individual games (what Gunther calls “on-that-day” better), and better as determined by your body of work over the course of the season – determined in pro sports with balanced schedules by won-loss records, but even then there is a small element of subjectiveness when comparing teams that differ by one or two games, because if that one game had gone differently who knows what might have happened. Who’s better “on that day” is obvious, but there is a tendency to conflate it with who’s better over the course of the season. Gunther imagines the following debate between an Oklahoma and Texas fan at some point after the end of the 2008 season, which I have edited for clarity, and highlighted the Oklahoma fan in red and the Texas fan in orange:
Oklahoma is better than Texas.
No way – Texas beat them 45-35!
So what? Going by that, then you have to say that Texas Tech is better than Texas, since the Red Raiders beat the Longhorns.
Texas obviously had a better season than Texas Tech. The Longhorns didn’t get blown out 65-21 by Oklahoma – their one loss was by 6 points on the last play of the game.
The Texas fan argues that, since Texas was better than Oklahoma on the day the two teams took the field, that means Texas was better than Oklahoma over the course of the whole season as well. The Oklahoma fan brings up a strawman, saying by that logic, Texas Tech is better than Texas, which of course, would further imply that Oklahoma was better than Texas Tech and therefore Texas. The Texas fan dismisses that result on the grounds that it was a close contest, effectively saying the respective bodies of work of Texas and Texas Tech trump the result of that one game. Yet those bodies of work are themselves defined by two games: Texas beat Oklahoma but Oklahoma beat Texas Tech. None of the other games any of the three teams played matters one iota. And the worst part? You don’t need to be a Texas homer to say this – just about everyone outside the state of Oklahoma worked through the argument this way after the Big 12 Title Game matchup was set.
That, more than anything else, is a symptom of college football’s small sample size: we can’t even perceive of what it means to be better over the course of the season. All we know is that two teams won ten, eleven, or even twelve games. We can’t compare two teams over the course of all twelve games; we can only compare them in terms of comparable individual games. That, in a nutshell, is why when it comes to BCS conferences, teams end up being ranked by their record. Other sports – even college basketball, where records alone aren’t everything – have a large enough sample size that we can make those comparisons without resorting to individual games. (If there were fewer cupcake games and more inter-conference games between powerhouses we might be able to make better comparisons, but that only happens when you have enough leeway to care about more than record, and that only happens with a playoff.)
This helps explain why one of the big rallying cries of a playoff to proponents is “settle it on the field!” A playoff is entirely predicated on the notion that you can compare teams’ entire seasons by looking at a few individual games. Gunther’s opponents don’t believe in that notion; they prefer to look at the big picture of the season as a whole. Their counterargument is “Any team can beat any other team on any given day.” After all, just because Appalachian State beat Michigan or Stanford beat USC doesn’t mean either of those two teams were actually better than the teams they beat, at least in the context of the entire season. No one in their right mind would say that. But it’s easy to throw out games when the teams involved are widely separated in the standings; what happens when two teams are very close in the standings, like a game apart or tied? It seems reasonable to look at the two teams’ games against one another to break the tie as to which team’s better, right? That’s what professional leagues (and, to some extent, other college sports) do.
Here’s the dirty little secret about comparing two teams’ seasons by looking at their games against each other: If the Steelers and Chargers each go 14-2 this season, and the Steelers beat the Chargers, then if you take that one, fallible, game away, the Steelers would be 13-2 and the Chargers would be 14-1.
Let’s look at two examples. Looking at games and records alone, you can’t resolve the problem between Oklahoma, Texas, and Texas Tech. If you look at the games, Texas is better than Oklahoma, but Oklahoma is better than Texas Tech, but Texas Tech is better than Texas. You have to throw out one of the games. If you look at records, all three teams were 11-1 before the Big 12 Championship Game. Take away their games against each other and all three are 10-0. Even people within a given side of the debate wouldn’t be able to agree on what to use after that – margin of victory? (Effectively keeping the focus on the games.) Strength of schedule? (Effectively forcing you to look at the body of work.) Something else? Some combination of the bunch? Proponents say the right solution in this situation is a playoff to include all three teams so at least two can get eliminated. But opponents of a playoff say this shows games can fail you, putting you on a logical loop, and forcing you to throw out a game and admit games aren’t always the be-all end-all even among comparable teams, so a playoff won’t tell you as much as you might think. (Moreover, if you’re interested in this loop because you want to figure out which team had the best season, your next stop should be more seasonal factors like strength of schedule, so why did you take a detour into individual games again?)
So in 2000, which team was better, Florida State or Miami (FL)? Florida State went to the national championship game, outraging many people who felt it was ignoring Miami’s victory over the Seminoles. How reasonable is that, really? Take away Miami’s win over Florida State, and Florida State was 11-0 while Miami was 9-1. Had that game just been cancelled, no one would have complained that Miami was more deserving in a spot in the national championship game than Florida State. Are you just creating an artificial “tie” between two teams, effectively bumping Miami up two levels for no reason? Does a single, fallible game completely trash Florida State’s season? If Florida State was better outside that one game, and Miami was better “on that day”, shouldn’t Florida State still get the seasonal prize? Did they really settle their differences “on the field” or was it only because of that game that there was anything to settle? If that were a playoff game, would we say Miami was better over the course of the regular season because they beat a team that, until then, we thought was better during the regular season?
Yes. Again, it’s impossible to say that a team in one conference is necessarily better than a team in another conference just because they have one fewer loss. For all we know, Florida State might have gone through a weaker schedule, dodging bullets Miami had to deal with. Miami’s only loss was against Washington, who – like Miami – went 10-1 (and had a bit of a beef of their own with the national championship game selection – was it only a decade ago they were still good?). Washington was clearly a good team; maybe they too got lucky on that day in that one game. Had the ball bounced a different way, Miami could have been 10-0 outside the Florida State game and then there is a tie for us to break again. Heck, maybe Washington was better than Miami who was better than Florida State. And not all situations are like this, where there is a game we can turn to to break the tie, which is why a playoff is such a good idea, to create those games.
Gunther’s opponents would argue that an individual game isn’t a good candidate to break the tie, because it’s so fallible, and thus a playoff wouldn’t work; best to stick with a system that forces us to use tools that work over the course of the season. But the point is that a playoff can create at least the illusion of clarity; it plucks out a single team that, in college football, can at least claim to be the best of the bunch even with a fairly large playoff. In both the 2008 and 2000 cases, the current system was left with amorphous blobs of three teams and could conceivably have picked any one of the three to go to a game that would crown one team as national champion anyway (conceivably none of the three – this debate was all to see who would face eventual national champion Florida or Oklahoma respectively, and the latter was undefeated!).
The example of Miami and Florida State shows that college football can’t rank every single team based on their record. Despite the complaints about how unfair it is that non-BCS teams have no shot at the national championship, no person in their right mind that’s not a Mountain West or WAC homer would put up with Utah-Boise State as the 2009 National Championship game. Members of BCS conferences would complain that they’re being punished for being in good conferences and the tendency to schedule cupcakes would get even worse. Ranking every team based on record, without regard to schedule, benefits the non-BCS conferences but it rarely actually selects the best teams that managed to escape good conferences. The system is biased against the non-BCS teams for a reason, people. A playoff is the only approach fair to both the BCS and non-BCS conferences.
Gunther waxes poetic about how professional leagues can use win-loss records and nothing else to determine playoff composition and seeding, because their leagues are small enough and the number of games are large enough for a balanced schedule, while college sports need to use other factors like strength of schedule and create a more convoluted (and inherently subjective) ranking. Fans of, say, last year’s Patriots can complain that their 11-5 team didn’t get into the playoff while the 8-8 Chargers did, but at least they know that’s the way the rules are set up. In college basketball, however, Patriots fans – while they might accept that the Chargers are deserving of an auto bid – would complain that the committee passed them over for some other team when they shouldn’t have.
And they’d be right, and wrong. People could debate until the cows came home over whether the Patriots were more deserving of an at-large (or, perhaps, just inherently better regardless of what they deserve) than, say, the Ravens, or over where the Chargers should have been seeded compared to other teams, and would never come to a truly definitive answer. Because of this, the NFL quite possibly could get away with a BCS-like system, selecting just the champions of each conference, ignoring the divisional imbalances, and rarely upsetting more than one team in a given year if even that (in fact, the National League got by without any playoff for years until the AL added ambiguity, and European soccer leagues still do) – but college basketball could never consider a playoff even sixteen teams deep, or even one that just didn’t give auto bids to every single conference. And college football isn’t much different.
Yet Gunther follows this line of reasoning to the exact opposite conclusion – that because determining the best team is more inherently subjective in college than in professional sports, it warrants a subjective method of determining the champion, namely, the BCS. All his reasoning has told us is that the regular season in the NFL is more inherently objective and the regular season in college basketball is more inherently subjective, which tells me that for the sake of balance, the NFL can afford to be more subjective in how it chooses its champion (and maybe, to re-emphasize the regular season, they should), while college basketball has to add an objective element to its postseason, or else it gets, well, the BCS (with three times the headache!). Gunther seems to think college sports need to “mak[e] their inherent subjectivity work for the sport”, and while in college basketball that just means leaving the seeding to a committee, in college football, apparently, opponents of a playoff think a subjective season needs to be followed by a subjective championship. Is it any wonder we go through headaches every year? And is it any wonder why March Madness is considered one of the greatest tournaments in the country and is heavily watched as far back as the Round of 64?
Let’s dig into Gunther’s reasoning:
[A]nti-playoff fans want their champion to be the best over the whole season, and in order to gauge that the season usually has to be equal throughout. The college football season as it is now, with all teams playing roughly the same number of games of equal value, gives them that. But switching from a subjective regular season with all teams participating to an objective playoff with a handful of teams participating breaks the continuity of the seasons as a whole and throws it out of balance.
People who back the notion of a “regular season playoff”, in other words, note that every team plays twelve games (and this is where a lot of people complain about the extra game conference championships add). For most BCS teams, four of those games are nonconference games (the Pac-10 has one fewer and the Big East has one more) – we’ll say two of them are against cupcakes and two are against tougher opponents, one arranged by ESPN and one interconference rival. The other eight are played against other teams in your own conference, and for the most part, despite some ebbs and flows and inequalities between the conferences, the comparison is fairly constant across BCS conferences. Everyone plays the same number of games, and those games have roughly the same amount of value, so BCS proponents (I know a lot of people who oppose playoffs hate the BCS as a pseudo-playoff, but a lot of them love them too, and I don’t agree with Gunther’s interpretation of their side, including the pro-“ranking” aspect, so I refer to them as pro-BCS for simplicity only) are fine with picking BCS teams based on record alone, and because of the bowls that doesn’t introduce any further incongruity. But it’s arguably unfair for the best teams to play two, three, even four more games after the end of the season, and play that many more games than everyone else. If you’re in a playoff, you’re being punished for being good and continuing to win. What’s more, the only game that matters once you’re in the playoff is the last game you played and the next game you play. The regular season, with its body-of-work aspect, is now irrelevant, and the result is a schism between the best team and the team that’s winning now. This is why you hear people moan about how little the regular season matters in other sports, no matter what that sport is.
This line of reasoning would work in European soccer, but it breaks my brain in American sports and certainly in college football, where not only are there differences between BCS conferences, the non-BCS conferences just don’t compete on a level playing field. They don’t play opponents that are as tough and they don’t have an equal shot (or any shot) at the national championship. If you want to define FBS as consisting solely of the members of the BCS conferences, you can make the argument that everyone plays the same regular season, but I hope you have time to console the non-BCS teams that just got told they’re not really part of FBS, they’re not playing for the same championship.
What’s more, the idea that the regular season ceases to matter in a playoff isn’t completely true either. True, if you lose you go home. But if you win, who do you play? Do you play the team widely considered the best team in the country on their home turf, or do you play a team that barely got into the playoffs at all on your own home turf? Under a playoff system, the regular season is not only important for determining which teams get into the playoff, but which seeds they have as well, which can affect how far they progress once in the tournament. And seeds are important: a 1 seed has never lost to a 16 in the NCAA basketball tournament, but a 9 seed beats an 8 seed more times than not. That’s why a team that has already locked up their spot in the playoff won’t necessarily start coasting – if there’s a high risk-reward for maximizing their seed. Whether or not there is varies from year to year and system to system, but if college football had a plus-one in 2008, and Florida and Alabama knew they were both moving on to the plus-one regardless of the result of the SEC Championship Game, and the difference was whether they were playing Oklahoma or Texas (or, if Oklahoma lost the Big 12 title game, USC) in the semifinals, I don’t think they’d be terribly motivated to fight for the chance to play the thought-to-be-marginally-worse team. In 2008, the plus-one fell into college football’s uncanny valley: the SEC Championship Game had more meaning under the BCS, but it would have had more meaning under a 16-team system as well. Same goes for Ohio State-Michigan in 2006.
It turns out that a postseason that makes the regular season less meaningful if your goal is to make the playoffs makes it more meaningful if your goal is to fight for seeding. In the NBA, the regular season isn’t very meaningful for picking the teams that go to the playoffs, since half the teams are picked. If your goal is just to make the playoffs, you can coast once you’re in. But by the same token, seeding is very valuable in the NBA, since the 1 seed faces a mediocre team in the first round. The 1 seed would be less valuable in a college football playoff that selected the best eight teams regardless of conference because you’d be facing another team almost as good in the first round. So you need a balance between a playoff large enough to make it valuable for the top teams but small enough to make it valuable for enough bubble teams that no team feels safe, whether it’s with their seeding or their spot in the playoff.
College football is a large enough universe that even with a rather large field, seeding wouldn’t be terribly valuable in and of itself because the differences between seeds would be fine gradations. Seeding value is enhanced by giving auto bids to the winners of mediocre conferences/divisions and placing them at the bottom of the ladder. The 16 seeds in the NCAA Tournament aren’t the 64th best teams in the country; they’re far worse. Teams fight for 1 seeds because they know they’ll not only get a free pass to the second round, they’ll have hardly broken a sweat when they take on a team that survived the grueling 8-9 game. Drop down to the 4 seed and you face a real upset possibility and no distinct advantage over the winner of the 5-12 game.
(Most pro sports do this wrong and give the auto bid teams the best seeds as well, regardless of record compared to the other teams. This is because pro sports are balanced enough that the 1 seed wouldn’t benefit that much more from playing a weak division winner than a less-weak non-division winner, and in fact usually wouldn’t be affected at all. But the NBA did eventually notice that when the 3rd best division winner has a worse record than the 3rd best non-division winner (or 6 seed), and the second-best record happens to belong in the same division as the best record, and home court is based on record and not seed, it produces a perverse incentive to lose and sink to the 6 seed, and allowed the best non-division winner to be seeded with the division winners.)
Tomorrow we put this all together as I unveil – and further defend – my preferred playoff format.