Today's entry is BZ's proposal for postseason play, which he outlines in great detail, and applies to recent history to assess its viability. Enjoy.
The Flex Playoff System
As I said previously, on a pure resume-ranking analytical level, there is nothing wrong with the old bowl system. You play 12-14 games and then we rank you on what you did. Simple as that. But for reasons that go beyond that and delve into what we want the system to mean or represent, that's not good enough. First of all, I want there to be a college football showcase - an EVENT - that we can all point to as the "system." I imagine that TV stations will never allow college football to go back to a million games on January 1 (the "event" that the old bowl system had) because it cuts into the viewership of each individual game. Additionally, it can be argued effectively that there is not always enough of a sample size to determine (even based on pure resume-ranking) the best team in any given year without them playing each other. Some years there is enough information from the regular season (2001 Miami comes to mind), but most years there is not (2005 Texas and USC comes to mind). Fleshing out "Underlying Problem #2," we don't want our system to end with confusion over who the true national champion is - see Predicate Assumption, above (this is a debatable assumption, one that BrooklynHorn has at least questioned in the past, but this whole conversation is pointless if we don't care about there being a single "national champion").
So there must be some way to reconcile these two underlying problems and develop a postseason system that is the most ideal solution to both underlying problems. Essentially, our primary goal should to design a system that crowns a "national champion" that is the most deserving based on our notion of what that means, while at the same time (though not to the detriment of the primary goal) designing the system that we want to be representative of college football. Based on my views regarding these "Underlying Problems" with the issue of how we determine a national champion, I have a proposal for the most ideal system for doing just that. I call it the "Flex System."
The basic idea is that in different years the regular season results lead to differing numbers of "legitimate contenders" for the national championship. Some years, notably 2005 with USC and Texas, there are two. Some years there are three. Some years there are four. Because we are resume-ranking, "legitimate contenders" for the national championship must have fewer than 2 losses. That's the primary rule and I'm not backing down on it (with two notable but I believe ultimately defensible and necessary exceptions).
[Please Note: I will refer constantly to the "BCS rankings" and "BCS bowls" in this proposal. By that I do not necessarily mean the BCS in its current incarnation. I mean some objective method (determined prior to the season, not by an ad hoc committee after the regular season) of ranking the teams. I personally like the combination of computers and humans - though I would prefer it if the computers could take into account margin of victory (up to a certain threshold - say 21 points) and the humans weren't such idiots. Improvements in that direction would be welcome.]
This is the outline of the system: there are 5 BCS bowls, the 4 big ones and a "national championship" game (basically just as it is now). All 6 BCS conference champions get in to one of these 5 bowls, just as it is now. The twist comes in that some years there will be 2 semi-final games (both BCS games) leading up to the championship game (basically, a four-team playoff). Some years there will be 1 semifinal game leading up to the championship game against a team with a bye to the championship game (a three-team playoff). Some years there will be no semifinal games at all (a two-team playoff). There is only one possible way that a BCS conference champion will not get into a BCS game: if there is a 4-team playoff and 3 of those 4 teams are not BCS conference champions, as that would leave 5 BCS teams with only 2 games. This is of course highly, highly unlikely. But in the event it happens, the team left out could still receive its BCS cut of the money in addition to its cut from the bowl game it does play in. It sucks, but really, how often is this going to happen? Alternatively, a sixth BCS Bowl could be added to ensure that this never occurs [the Cotton Bowl springs to mind as the ideal candidate if Dallas could ever get their shit together and adequately renovate the stadium].
So in a year with a 4-team playoff, there will be 2 at-large teams in the BCS - either in the playoff or in the regular BCS games (though note, that there could be 3 at-large teams if that many make the playoff, as mentioned above). With a 3-team playoff, there will be 3 at-large teams. With a 2-team playoff, there will be 4 at-large teams. I am not concerned with how the at-large teams are determined for the non-playoff BCS spots at this time. It can remain as it is now for all I care.
So here are the rules as to how we determine which type of tournament we will use in any given year:
Multiple undefeated teams
a. If #1 and #2 in the BCS standings are undefeated and there are no other undefeated teams in the top 5 of the BCS standings, then there will be a two-team playoff between #1 and #2.
b. If there are 3 undefeated teams in the top 5 of the BCS standings, then all undefeated teams will be in a 4-team playoff with the highest ranked 1-loss team. Seeding is made according to BCS ranking.
c. If there are 4 undefeated teams in the top 5 of the BCS standings, all undefeated teams will be in a 4-team tournament with seeding made by BCS ranking.
d. If there are 5 undefeated teams in the top 5 of the BCS standings, then the top 4 are in a 4-team playoff with seeding made by BCS ranking.
e. If there are 2 undefeated teams in the top 5 of the BCS standings, but they are not ranked #1 and #2, then the undefeated #1 team (if there is one) will be treated as undefeated, but the team(s) that is/are not ranked #1 will be treated as a "Top 4" 1-loss team according to the paragraphs below.
[Notes: This is mostly straightforward. I used the "top 5" arbitrarily. It could legitimately be 4, 6 or maybe even 7. The goal is to catch mid-major undefeated teams that have played a tough enough schedule to warrant inclusion with the big boys (i.e. Utah in 2004 and BSU in 2006), but exclude those that didn't (i.e. BSU in 2004). Whatever the number is though, voters would likely put the undefeated team high enough to reach that number if they believe them deserving. Paragraph (d) sucks, but really, when is that ever going to happen? Paragraph (e) exists on the assumption that if an undefeated team is ranked behind 1-loss teams, it is a mid-major undefeated that is on par with a 1-loss major program and does not deserve to automatically be put in a 2-team playoff game under paragraph (a). See "additional hypothetical," below.]
One undefeated team
f. If the only undefeated team in the top 5 of the BCS standings is not ranked #1, then it will be treated like a "Top 4" 1-loss team under paragraphs (k) through (n) below.
g. If the only undefeated team in the top 5 of the BCS standings is ranked #1, and there is only one 1-loss team in the top 4 of the BCS standings, then those two teams will advance to the national championship game in a 2-game playoff.
h. If the only undefeated team in the top 5 of the BCS standings is ranked #1, and there are two 1-loss teams in the top 4 of the BCS standings, then the undefeated team will receive a bye into the national championship game and the top two 1-loss teams will play for the right to go to the national championship game in a 3-team playoff.
i. If the only undefeated team in the top 5 of the BCS standings is ranked #1, and there are three (or more - see paragraph (e)) 1-loss teams in the top 4 of the BCS standings, then there will be a 4-team playoff. Seeding will be made according to BCS ranking except where an undefeated team being treated as a 1-loss team under paragraph (e) would not be included, in which case that team is seeded #4 and the #4 team in the standings is excluded.
j. If the only undefeated team in the top 5 of the BCS standings is ranked #1, and there are no 1-loss teams in the top 4 of the BCS standings, then the "national championship game" will be played between the top two teams in the BCS standings but the winner will not automatically be crowned "national champion." A final poll will be taken after the bowl games to determine the national champion.
[Notes: The reason for (g), (h), and (j) is that a 2-loss team does not deserve to be automatically named national champion based on a victory over an undefeated team. Thus in (g), the sole 1-loss team advances automatically to the national championship game. In (h), there is not a 4-team playoff because the 2-loss team ranked #4 shouldn't be in the national championship conversation when there are also two 1-loss teams and an undefeated team. In (j), it is debatable whether the 2-loss team that won the "national championship game" against the previously undefeated team is more deserving of the national championship than the 1-loss team that lost that game. So a poll is taken. But this will rarely if ever happen. Paragraph (f) exists on the assumption that if the only undefeated team is not ranked #1, it is a mid-major undefeated that is on par with a 1-loss major program and it doesn't deserve an bye into the national championship game any more than the non-undefeated #1 team.]
No undefeated teams
k. If there is one 1-loss team in the top 4 of the BCS standings and no undefeated teams in the top 5, then the 1-loss team will play the highest ranked team other than itself in the national championship game.
l. If there are two 1-loss teams in the top 4 of the BCS standings and no undefeated teams in the top 5, then those two teams will play a 2-team playoff in the national championship game.
m. If there are three 1-loss teams in the top 4 of the BCS standings and no undefeated teams in the top 5, then the top 4 teams in the BCS standings will play a 4-team tournament. The team with more than 1 loss will be seeded #4 and the three 1-loss teams will be seeded based on BCS rankings.
n. If there are four (or more - see paragraphs (e) and (f)) 1-loss teams in the top 4 of the BCS standings and no undefeated teams in the top 5, then there will be a 4-team tournament. Seeding will be made by BCS rankings except where an undefeated team being treated as a 1-loss team under paragraph (e) or (f) would not be included, in which case that team is seeded #4 and the #4 team in the standings is excluded.
[Notes: Paragraphs (k) and (m) are the notable exceptions where a 2-loss team is allowed into the playoff and can be named the national champion if it wins. This is defensible because the end result is unlike in paragraph (j) where at the end of the day, there is still a team with a better record than the 2-loss "winner." All the teams will have the same record, with one team having beat the others. I hate this, but I think it's necessary.]
The Flex System Applied to All BCS Years
[I realize that looking at these years in retrospect is a bit of a fool's errand because the formulas varied over time and in my ideal system the formula would probably change some more. But the goal of this exercise is to show how well this system would mitigate the complaints that teams and college football fans generally had in each and every year, while still being the most ideal system for determining the national champion according to what we want the national champion to be and what we want the national championship system to mean, or represent.]
1998 A four-team playoff according to paragraph (i) above. Undefeated #1 Tennessee plays 1-loss #4 Ohio St. and 1-loss teams #2 Florida St. and #3 Kansas St. play each other. Possible complaints: UCLA and Arizona, both 1-loss teams. Response: look, we can't let every 1-loss team in. It will get too big. The other 3 are lucky they got in. Next time don't lose a game. Other Notes: this will mitigate the complaint of KSU, whose sole loss happened to come at the end of the year and was thus bumped down to the Alamo Bowl.
1999 Undefeated #1 Florida State plays undefeated #2 Virginia Tech in the national championship game, under paragraph (a). Complaints: no one. Nebraska played a tougher schedule than VT but lost to Texas, which wasn't that good in `99 (woohoo Texas!).
2000 A four-team playoff according to paragraph (i) above. Undefeated #1 Oklahoma plays 1-loss #4 Washington and 1-loss teams #2 Florida St. and #3 Miami play each other. Complaints: Virginia Tech, and Oregon State, both 1-loss teams. Response: see 1998. Other notes: this would solve the main complaint from this year which is that FSU's only loss was to Miami whose only loss was to Washington, who only lost once as well. Well, here you go. Merry Christmas.
2001 A three-team playoff according to paragraph (h). Undefeated #1 Miami gets a bye into the national championship game. One-loss #2 Nebraska plays one-loss #4 Oregon for the right to play Miami in the national championship game. Complaints: #3 Colorado. Response: You lose TWICE! Once to Colorado State! The other time by THIRTY-FOUR POINTS to Texas! Shut the hell up. Other Notes: In all honesty, Miami was the only real team this year. But Oregon's legitimate complaint that it wasn't in the national championship game over Big XII North runner-up Nebraska is rectified.
2002 Undefeated #1 Miami plays undefeated #2 Ohio State in the national championship game, under paragraph (a). Complaints: no one. Georgia and Iowa were 1-loss teams in the top 5, but they have no legitimate complaints. Georgia lost to 8-5 Florida and Iowa lost to Iowa State and then USC in their bowl game so screw them.
2003 Paragraph (m) applies in this 4-team playoff. One-loss #1 Oklahoma vs. two-loss #4 Michigan and one-loss #2 LSU vs. one-loss #3 USC. Complaints: A million other 2-loss teams. Response: Screw you. Lose less than twice next time. Other notes: In retrospect, USC vs. LSU should have been the national championship game. And with the BCS formula changes made after this year, that may have happened with USC ranked #1 in the BCS, playing Michigan in, I don't know, let's say the Rose Bowl and then playing LSU after LSU beats OU in, I don't know, let's say the Sugar Bowl. That would've been nice, right?
2004 Two words: Giant Clusterfuck. There were 5 undefeated teams this year. Anyway, under the rules above, paragraph (b) applies in a 4-team playoff. Undefeated #1 USC plays one-loss #4 Texas and undefeated #2 Oklahoma plays undefeated #3 Auburn. Complaints: #6 Utah, #9 Boise State, and #5 Cal. Cal's complaint is that they were barely edged out for #4 by Texas. Yeah, yeah, we know. Texas Tech anyone? Boise State was weak that year. It wasn't the same as 2006. They played no one. Utah was another story. They could play with the big boys. As I wrote in my notes to paragraphs (a) - (e) above, the "top 5" criterion is completely arbitrary. Top 6 is equally credible, though I would argue that Top 8 is too much because it would include teams like BSU circa 2004. If you extend the criterion to "top 6" then replace Texas with Utah. Totally reasonable, and perhaps even preferable, as sad as that is to say as a Texas fan.
2005 Undefeated #1 USC plays undefeated #2 Texas in the greatest national championship game ever, under paragraph (a). Complaints: no one. Other Notes: This is where a playoff would destroy everything I hold dear about college football. The regular season CLEARLY separated these two teams from the rest of the field and if we have to play superfluous playoff games to get this game (or god forbid, an upset occurs and this game never happens), then what's the point of the regular season? p.s.. I Heart VY.
2006 A three-team playoff according to paragraph (h). Undefeated #1 Ohio State gets a bye into the national championship game. One-loss #2 Florida plays one-loss #3 Michigan for the right to play OSU in the national championship game. Complaints: #8 Boise State. Response: I think "top 8" is too big a spread for inclusion of undefeated teams, and top 7 is borderline. Top 6 is reasonable as is top 5. But whatever the threshold is, I'm convinced that if voters think a team deserves to be in the playoff, they'll put that team where it need to be to get there.
Additional Hypothetical: Paragraphs (e) and (i) need explaining because they are confusing. Say the BCS rankings look like this:
#1: Team A: 0 losses
#2: Team B: 1 loss
#3: Team C: 1 loss
#4: Team D: 1 loss
#5: Team E: 0 losses
There are 2 undefeated teams so we look at paragraph (a), but we shouldn't have a 2-team playoff between A and E because if E is undefeated and ranked that low, it's more like one of the 1-loss teams than A. There should be a bigger playoff than just 2 teams. So E is treated like one of the "Top 4" 1-loss teams under paragraph (e). Now we only have one undefeated team so we look at paragraph (i). There is a 4-team playoff, but we effectively have one undefeated and four "top 4" 1-loss teams. By virtue of being undefeated and in the Top 5, E gets in, seeded #4. This means that D is left out. This is reasonable, I believe. Most people would say that an undefeated #5 deserves to get in more than a 1-loss #4, but not necessarily over a 1-loss #2.
A few notes on criticisms of the Flex System
I've brought this system up before (though not in such excruciating detail) on this site, and there were people who shot it down for various reasons (i.e. "Flexable playoff = horrible idea" - Thanks Wells!). I'll attempt to head off those criticisms by addressing them here.
First, there was the criticism that the TV stations wouldn't like it because there wasn't enough certainty about how many games there would be, etc. Let me make this clear: there is the same number of Bowl games and BCS games every year. The only difference is whether the winner of one or 2 of those games goes on to play in another. And remember, if you have an 8-team playoff, you're effectively replacing the BCS games with the playoff and if you have a 16-team playoff, you probably need to scrap the whole Bowl System altogether. Maybe the TV stations would be more excited about those than what we have now. But they certainly wouldn't be LESS excited about the flex system than they are now. And should we really design our system for determining the national champion based on what TV stations think? Isn't catering to them how we got Rule 3-2-5(e)?
Second, and heavily related to the first, is that the Bowls themselves wouldn't like the system because it was too uncertain. This is a minor concern, in my opinion. First, Bowls don't have to sign on to the new system if they don't want to. They didn't have to sign onto the BCS contract, and the Rose Bowl considered not doing it because they wanted to preserve the Pac-10/Big-10 bowl tie-ins. But of course, all that would have accomplished was leaving them outside of the big money games. Bowls will sign on for that reason. There can be a rotation of which bowls get the play-in games depending on who has the championship game. Something will get worked out and the Bowls will sign on.
Third, the flexible system will encourage more cupcake scheduling. Well, it will not encourage MORE cupcake scheduling, and may encourage even less since you're more likely to get in with 1 loss than you are now, but I understand the idea. A large playoff will probably encourage harder scheduling, generally speaking (though not as much as people think since seeding is based on regular season results). And while I agree that harder scheduling is something that should be strived for, I think that's a secondary consideration in relation to making sure the "best" team is made national champion. (My notion of "best" is based on a resume ranker's perspective, of course.) There has to be some balancing. Hence, the Flex System. Additionally, with a large playoff system, while you would probably get stronger OOC scheduling, your interest in the results of those games is lessened because you can still get into the large playoff with 2 or even 3 losses.
Fourth, yes, it's a complicated system. It's not actually as complicated as it looks because, as you can see in its application to all the BCS years so far, it usually boils down to 3 or 4 different circumstances. The complexity is merely a result of there being exigencies that have to be accounted for, no matter how unlikely they are to occur. And yes, I'm familiar with Occam's Razor. But I am aiming at very specific goals here and I think that if you just go with the most simplistic system, you're completely ignoring the underlying problems in the debate over the best postseason system and, in turn, doing a disservice to college football. So I chose something more complicated. We've got a pretty simple system right now as it is (#1 vs. #2 = MNC), but that's not exactly working out how you want it to, is it? Complicated does not equal bad and simplistic does not equal good, just as the simplest argument isn't always the most convincing argument.
Thoughts, insults, profanities aimed in my direction for putting you through such excruciating minutiae? I want it all.