As we approach the final hashing out of college football's first playoff in 143 years, the debate hinges on how to determine the four-team field.
The battle lines are clear.
The SEC and Big 12 want the four-team playoff (the 1-4 model) to feature the top four ranked teams.
Alabama still would have made the final four in either of the proposed playoff plans. …
The Big 10, Pac 12, ACC and Big East have expressed interest in another format, one that gives weight to conference champions. While there is still some discussion of a foursome comprised of only conference champions, there is also broad recognition that there needs to be a hybrid model.
The Big Ten has said a conference champion needs to be "ranked in the top six." Others have expressed support for having the top three conference champions and one wild card (non-champ) if that team is "ranked" in the top four. That so-called "3-and-1" plan would also allow access for independents, most notably Notre Dame and Brigham Young.
Yet as long as a selection committee is used and polls and formulas are discarded, it doesn't appear there will be any appreciable difference between having a field that features the so-called top four teams versus one that includes three conference champions and one at-large spot.
It's likely the only difference would come from multiple upsets in conference title games that, in the case of the 3-and-1 model, could put a three- or four-loss "champion" into the playoff.
That alone is why the 1-4 playoff makes more sense.
Mostly, however, this looks like the current leadership's longstanding inability to comprehend math, logic and probability while clinging to "traditions" that make no sense.
Confused? It seems everyone is.
I'm going to do my best to explain my own theory, culled from speaking with numerous experts. (Warning: there is math involved).
One subject appears to have decent support – the use of a selection committee to choose the field. While it will, no doubt, be a difficult and controversial job, it appears to have broader backing than a single public computer system (the current BCS formula that is riddled with issues) or the use of the final standings of one or more weekly opinion polls.
It's imperative to solving this argument because a selection committee essentially makes the debate moot. Why? Because, without a BCS-style formula, no longer will there be a method to determine what constitutes the top four or top six in the first place.
[Related: Pat Forde: Forty things to know about the 2012 college football schedule]
Barring a season that based on my research has never before happened in college football, there will be no appreciable difference between a 1-4 formula or a 3-and-1 formula. More importantly, with a committee there will be no way for anyone to even know there was a difference.
Take last year.
Under the 1-4 model – and using for this example BCS rankings – the playoff would have featured LSU (1), Alabama (2), Oklahoma State (3) and Stanford (4). Under the 3-and-1 model, it would have been LSU (1), Alabama (2), Oklahoma State (3) and Oregon (5).
While these two don't appear to exactly line up, they actually would under either of the proposed playoff models because there would be no BCS standings to tell us that Oregon was ranked behind Stanford. Instead, the selection committee would rightly chosen Oregon over Stanford based on their 53-30 win over the Cardinal in Palo Alto.
No one would've argued. No one would've even thought to argue. In the end, three conference champions would be in the top four.
The selection committee and only the selection committee would determine the top four (or the top three-plus-one). No one can argue that they took the fifth-best team or the sixth-best team or the ninth-best team because there is no fifth-, sixth- or ninth-best team. There are just teams.
Once you eliminate the ranking system, there is no officially sanctioned controversy. The AP poll, various power rankings or even computer formulas can make their claims and can be used as a tool by the committee (the way the RPI is in basketball), but none of them truly matter. None of them count.
Rankings can (and will) be cited in arguments for and against each team, but they don't carry the official, binding weight that the BCS standings did. You'll never get rid of the arguments, but this offers important cover for the selectors.
The top four is whatever the committee says it is. And it is just as arbitrary as the old way.
The BCS was an insult to math, science, logic and reality. No one ever thought it through. The BCS leaders believed they needed a "math" component for public relations purposes, so they used computers.
These leaders appear to have no understanding of math or even an interest in having an understanding of math.
The formula consisted of six computer equations. Five of the equations were secret: Not even the BCS knew the formulas or whether they were being properly computed. They had no way to check. In fact, the one public formula was calculated inaccurately on multiple occasions, a mistake discovered not by the BCS but by independent analyst Jerry Palm of CollegeBCS.com.
In recent years, the computers were forced to exclude margin of victory, which caused a number of the people who run the computers to complain they were producing less accurate numbers. It even sparked a boycott of the BCS by some quantitative analysts, led by UC Irvine professor Hal Stern, who were aghast that credibility of numbers was being debased.
“It's just nonsense math,” said Bill James, the baseball statistician and driving force behind sabermetrics.
BCS officials didn't care. They didn't want to know what they didn't understand.
Then there were the opinion polls, one by the coaches, one by a group of people with some ties to the game called the Harris Interactive Poll. They voted every week. The coaches even voted in the preseason, which meant expectations had a role in the formula.
Why? Because college football has almost always had weekly polls and, at least it appears, no one ever bothered to question if they were effective at picking a title game match up.
[Related: Mike Huguenin: Traveling to play a non-Big Six team could be dangerous]
The answer: They aren't.
The most famous national poll – the Associated Press – began in the 1930s as a way to promote the sport. A bunch of sportswriters were rounded up to rank the best teams each week. This was a fun and harmless exercise. It was also ridiculous. There was no television then and writers had no ability to watch teams from across the country. They essentially just voted off the newspaper agate page.
There is no way anyone thought this was accurate or legitimate.
Yet somehow the weekly poll became so entrenched in the college football psyche it became part of the official system even though it was never designed for such a purpose.
Mathematicians, political scientists, statisticians and psephologists mock the use of such ranking systems. They are wholly flawed. Some of it is obvious, such as the prevalence of groupthink, political pressures and the proven inability of voters to jump teams in the rankings. Some of it is more advanced, such as the concept of independence of irrelevant alternatives, Arrow's Impossibility Theorem (which merely won a Nobel Prize) and the Pareto efficiency principle.
There's more. A lot more. You can read up on those on your own, but to crudely summarize, the best way to chose the top two or four (or any segment) of teams is not to have a large group rank them with 120 or so possible choices. It doesn't work. It never was designed to work. No one in his or her right mind would expect it to work, which is why no other sport attempts it.
So counting on the BCS formula, or almost any formula that features compromised math and/or a ranking-based opinion poll, is completely illogical. It's a security blanket for the ignorant.
Unfortunately, it also is the basis for the arguments the commissioners are fighting over.
They are fighting over what is 1-4 when 1-4 isn't real. They're trying to create a standard of a top 6 when the way you choose the top 6 is scientifically bankrupt.
The selection committee will have its challenges and controversies, but it isn't pretending to be anything based in science. Only people who fear science would rail against a committee to instead fall back on failed science.
The simplicity of the selection committee makes it honest. This is the field. Why? Because we said so. If we're all adults here, we can admit there is no good way to pick four out of 125 teams. So deal with it.
The commissioners appear to be planning for tomorrow based on yesterday's pointless examples.
There are only three scenarios where a 1-4 and a 3-and-1 can yield different results.
1. If two conferences each produce powerhouse runners-up and every other conference has a relatively weak champion, likely with two or more losses.
2. If one conference has three powerhouse teams and every other conference has relatively weak champions, likely with two or more losses.
3. If one conference has four powerhouse teams and every other conference has a relatively weak champion, likely with two or more losses.
In each case, four teams from just one or two conferences would have to significantly and obviously separate from the rest of college football.
Looking back through college football history, there does not appear to be a single case where choosing the top four champions or three champions and one at-large team are obviously or even notably different. There is nothing even close to the second and third examples. (Yes the SEC had three strong teams last year, but the Big 12 and Pac 12 had strong champions). So while it's possible, it's extremely improbable.
As for the first example, during the BCS era the closest (and only) situation would be 2006, when the top five according to the BCS formula were Ohio State, Florida, Michigan, LSU and USC. The debate would likely have come down to the final spot between LSU and USC, both of which were 10-2. Also in the running, Louisville (11-1), Wisconsin (11-1) and Boise State (12-0).
USC, Louisville and Boise won at least a share of their league titles. LSU and Wisconsin didn't. Would there be a debate over these five? Absolutely. Yet three of the choices would yield the same result – three conference champs and one at large (Michigan).
And, once again, since there would be a selection committee and not a public, weekly ranking system, this would require separation that was considerable, a truly undeniable debate. There was a reasonable debate here.
By killing off the BCS formula, the problem is effectively solved.
The selection committee could simply have said they chose USC over LSU because they thought the Trojans were better. There would have been no pesky “ratings” to explain.
Plus, with conference expansion consolidating quality programs, Boise and Louisville, now in the same league, would knock each other off in 2014. Michigan and Ohio State, meanwhile, would have had a rematch in the Big 10 title game, which could have ended one of their candidacies.
This isn't too bad or probable. And again, the 2006 example is the most controversial we could find. Not sure it's worth killing each other over.
There is one caveat history doesn't account for – upsets in the conference championship game.
Conference title games didn't begin until the SEC in 1992. By 2014, when the playoff kicks in, every major league except the Big 12 will play a title game. The leagues love title games because they make millions of dollars. It isn't necessarily the best way to determine a champion, although as leagues continue to expand, true round robins aren't possible.
Mostly the issue is with divisional champions, rather than purely the top two teams in a league, reaching the title game. (It's essentially the same argument on a conference level, which is why the pratfalls of division champs should serve as a cautionary tale).
If one division is weaker, a flawed team could score one upset and still be crowned “champion.” Just last year, in part due to NCAA sanctions against USC (another variable), a 6-6 UCLA team played for the Pac 12 title.
The fear here is that you could, conceivably, have across the board upsets in all the title games that would produce weak champions. Under the three-and-one plan, you might be forced to take a three- or four-loss “champion” because it is technically the third best conference champ. A slew of teams with better resumes that didn't win their own division would be excluded.
This doesn't really help anyone. For the good of the game, the new playoff should try to avoid scenarios where a poor field is created.
There is no perfect way to do this. So it becomes a numbers game. No matter the system, it is extremely likely that at least three conference champions will be in the four-team field. And while the odds are long for a true clunker making the playoff, that scenario is easier to envision.
So going 1-4 should, in practice, give what both sides want while guarding against the embarrassment of a weak champion sneaking in, all while keeping the process simple for the public to follow.
It's the right choice.
We'll let the commissioners have their death fight, but other than an entertaining battle of egos, the fans shouldn't really care.
As long as there is a selection committee instead of polls and rigged computers, the final product will probably be the same.
No comments:
Post a Comment