Saturday, November 06, 2010

Book Review: Death to the BCS

In Death to the BCS: The Definitive Case against the Bowl Championship Series by Dan Wetzel, Josh Peter, and Jeff Passan, in case you weren't clear from the book's title, we learn that the BCS is responsible for virtually all of the problems that plague college football. These include:

1. Teams actually don't make very much money off going to bowl games, partly because teams are required to buy a lot of tickets they often have trouble selling. Partly because of this, athletic departments are not profitable.
2. Bowl games are profitable, partly because teams are required to buy a lot of tickets they often have trouble selling, and don't give money to charity even though they're non-profits.
3. Bowl games would too still continue to exist even after a playoff.
4. The BCS inevitably screws some team.
5. Teams schedule weak non-conference opponents (non-BCS teams and I-AA) because the BCS promotes winning over a competitive schedule and because they want to win 6 games to be bowl eligible.
6. The regular season would too still continue to be really important even after a playoff.
7. The computer rankings as used by the BCS are nonsense math.
8. Poll voters are bad because Harris Poll voters don't necessarily even watch college football and coaches vote in their own interest.
9. Lots more people would totally watch a playoff, partly because ESPN's College GameDay has promoted the sport as a whole.
10. The BCS was created and run by a cartel of the power conferences, which hates non-power conferences.
11. Going back to the old system of bowl tie-ins would be a bad idea.
12. A 16-team playoff, with early round games at home sites, would be a huge money-earner and is obviously the right way to go.

The book is full of righteous indignation, and seems to be addressed to those who have an immediate negative reaction to the BCS. There are a couple chapters, like one on "superfans" and the one on "GameDay" that don't really feel like they add to the argument.

The big issue is, is their argument convincing? Do they make a good case that (a) the BCS should die, and (b) should be replaced by a 16-team playoff? To give you an idea of where I'm coming from, I think the BCS was created and continues as a reasonable compromise designed to address the problems created by bowl lock-in relationships preventing more desirable matchups but still keeping the historic and valued bowl system.

I. The Problem of a "Playoff"
The BCS, as currently constituted, is essentially a two-team playoff with a bunch of trappings. The trappings (the other BCS games) exist because they were needed as inducements to bring together the parties needed to create and maintain without defections a two-team playoff.

Wetzel et al. prefer a 16-team playoff including the conference champions of the 11 I-A conferences (the 6 "Cartel" conferences and the 5 others), plus 5 other teams chosen by a selection committee. They don't explain their proposed playoff in huge detail, but seem to prefer the first three rounds at home sites for the higher-ranked team with a national championship game at the Rose Bowl.*

*-Just to give a flavor of Death, they mention "Even an old Ohio State fan ... sees the allure of Pasadena hosting the national-title game" (184). Correct me if I'm wrong, but Ohio State is a Big 10 member, and the Rose Bowl in Pasadena is where the Big 10 has historically sent its conference champion. If you asked me to pick where a fan of any B10 team would prefer a national championship game, I'd guess at least 95% would say the Rose Bowl. A better, more interesting question is where an SEC fan would prefer the national championship game; my guess is New Orleans, home of the Sugar Bowl, would be the majority call, but Wetzel et al. are content to simply point out our Ohio State fan prefers Pasadena and leave it at that.

The ideal number of teams in a playoff is a question without an obvious answer. Obviously, Division I-AA has a 16-team playoff similar to what Wetzel contemplates. If you look at other leagues, most of them have had different playoff formats over time.

  • The NHL has had a 16-team playoff for a while, but previously conducted four four-team playoffs for each of its divisions before switching to 8 teams per conference with guaranteed top-3 seeds for the division winners.
  • Major league baseball previously had two divisions per league, a league championship series between the two division winners (even if one division winner had a worse record than several teams in the other division), and the World Series between the two LCS winners before switchng to its current two leagues, three divisions per league plus one wild card format.
  • The NFL since the AFL-NFL merger has gone from 4 playoff teams per conference with one wild card to 5 and then to 6, and for 1970-89 had a rule in place that a division winner could not face a wild card from the same division before the conference championship game.
  • The NCAA basketball tournament has gone from 16 to 65 teams and from conference champions and at large teams not in a conference only to guaranteed berths for conference champions and at large teams from a general pool.

So, obviously, there is no obvious ideal and universal playoff format. These formats do have one thing in common: at the end of the playoff, they crown a champion.

The issue with a 2-team playoff as opposed to a 10, or 12, or 65 team playoff is that the team is guaranteed to be among the N-best teams as determined by the playoff method. I'll get to the method of determining the "best" team later, but the pretty inevitable result of a 16-team playoff is a team not among the 5 "best" teams in the country will win the national championship. A good example of this has been the NFL-the last team rated #1 in DVOA to win the Super Bowl was the 2002 Tampa Bay Buccaneers, and it's not particularly unusual for a team to make or win the Super Bowl over a team that was clearly far superior to it-the most recent example of this is Super Bowl XLII, when the Giants beat the Patriots and were awarded the championship, even though the Patriots (a) won 6 more games than the Giants did in the regular season, and (b) defeated the Giants on the road in the regular season.

A probably even better example has been the NCAA men's basketball tournament. Despite the last three years, when arguably the #1 team in the country won the championship every year, you only have to go back to 2006 when a team not ranked by the Selection Committee among the country's top 8 won the championship.

In that way, college football's 2-team playoff has been, at least in my eyes, an interesting and pleasant exception: the champion is guaranteed to be among the two best teams in the country according to the selection criteria. Yes, that eliminates the possibility of the enthralling upset, but as I've described it's a trade-off.

Beyond the inevitability of an mediocre champion, the other problem Wetzel's proposed 16-team playoff has is the inability to include every team that could win the championship. W/r/t the NCAA basketball championship, a 65-team tournament and the ability to play 6 games in 3 weeks and 34 at-large teams means every team capable of winning the tournament is in the tournament; I'd estimate that the top 40 teams in the country are guaranteed to all be in the tournament. A 16-team playoff's smaller size, only 5 at-large teams, and conference championship games that may knock out a conference's best team, likely means that between 8 and 12 of the 12 "best" teams in the country would be included in a 16-team playoff.

Wetzel of course uses the 2009 season as proof this is a parade of horribles, and the actual highest-ranked (by AP) team they exclude is #14-ranked BYU. That depends, though, on 8 auto-bids in the top 10, which is not inevitable. A better example may be 2002, when only 5 auto-bids went to teams in the top 12.

The other thing a larger playoff does that Wetzel et al. don't acknowledge is it doesn't stop the debate about which teams should make the playoffs, it merely shifts the debate. The marginal team, instead of Oklahoma over USC in 2003, becomes LSU over BYU in 2009, but the same debates still apply. Unless I missed it, at no point do Wetzel et al. ever acknowledge this point.

II. Bowl Economics
Bowls occupy their privileged place in the college football postseason primarily because of historical inertia. They were initially created as essentially exhibition games, and were typically ignored in the polls until the late 1960's. Because they existed, and provided a valuable data point in comparing teams that were generally tough to compare, pollsters started incorporating them in their rankings. But because the bowls existed, there was never a strong push to create an NCAA-sanctioned postseason the way there was with other sports, simply because the most important teams wouldn't have been better off.

The most effective part of Death was just how much of the money from the bowl games stays with the bowl organizing committees. This is partially the result of a weird economic hybrid-some of the value of the bowl games, especially the more prominent ones, derives from their historical value, but some of it is obviously attributable to the historical ties of the games. Take, for instance, the Rose Bowl: how much of the value of the Rose Bowl brand belongs to the Rose Bowl organizers, and how much belongs to the Pac-10 and Big 10? The right answer here isn't obvious to me, and I think part of their outrage is over the top-a good bit of the Rose Bowl brand value, which belongs to the Rose Bowl and not the competing teams, is a result of the historical value accrued by the Rose Bowl organizing committee, and so it makes sense that the bowl keeps a portion of its provided value. Not that I need to tell you, but this sort of non-obvious nuance isn't to be found in Death.

Actually, probably more effective than how much money stays with the bowl organizing committee is the way payouts really work. This information isn't entirely novel, but essentially a bowl payout of $X includes $Y in expenses for the school. It's possible in some circumstances, depending on the bowl, for Y to be greater than X, especially if you look at it on a team-specific case.* In addition to $Y that's part of $X, schools also $Z in related expenses, like coaching bonuses.

*-Individual teams divide their bowl payout at the conference level, rather than keeping all of it. Wetzel "shows" that Florida's appearance in the 2009 BCS Championship Game only earned the school $47,000, despite the advertised $17.5 million payout. This isn't entirely untrue, but isn't entirely forthright either:
1. All bowl revenues are divided among the entire conference, so UF's revenue isn't bumped by as much as you'd expect from going to the BCSCG.
2. The BCSCG doesn't pay as much as it should based on how valuable it is compared to even the other BCS bowls.
Both revenue distributions are exactly what you'd expect if you were looking at a system designed not to for a single team to perform like gangbusters in its most successful year but rather to be successful for more teams over a bigger number of conferences.

A related point is that Wetzel et al. seem very confused by not-for-profits. To a corporate lawyer, the key difference between not-for-profits and regular for-profit entities is that for-profit entities are under an obligation to maximize shareholder value, while not-for-profits are instead obligated to pursue their mission. There are two points of contention here:
1. The bowls, by talking about their not-for-profit status, make people think they're good and wonderful charitable organizations, but don't give as much money to charity as you'd expect from the impression they give. It's a useful talking point for the book, but I rolled my eyes. Approximately nobody, including all charities, gives as much money to charity and charitable causes, as they talk about.
2. Athletic departments don't maximize profits. Instead, they tend to value more highly the prestige of their sports teams. Among other things, this favors bringing as many people, whether team members, band members, athletic department employees, sponsors, donors, and the like, as your bowl game permits. I'm sure part of the reason Florida only made $47,000 in profit from their BCSCG appearance was because the traveling party was larger and spent more lavishly than would have been the case if they'd gone to, say, the Music City Bowl.

If athletic departments really cared about maximizing revenue, they'd have as few non-revenue sports as they could, lobby the NCAA and the conferences to decrease the number of non-revenue sports they have to have*, and fund sports, especially the non-revenue ones, as meanly as possible. They comprehensively don't do this, in some cases quite the opposite. I'm pretty sure I know some of the reasons why, but Wetzel et al. don't acknowledge any of the unclear priorities. The not-for-profit status of athletic departments also explains why schools aren't completely eager to embrace a higher-revenue 16-team playoff, and why only 12 athletic departments turn a profit, and many others are reimbursed by a school's general fund.**

*-See this WSJ article (PDF) on the differing approaches between Texas and Ohio State.
**-Because they can.

III. Non-Conference Scheduling
Teams schedule non-power conference teams and I-AA teams, we are told, because (i) the BCS rewards teams for winning a lot of games and (ii) getting to 6 wins is important for bowl eligibility. Oddly for a book focused on teams being stupid and money-grubbing, what I see as the most important reason teams do that is never mentioned in the chapter: it's revenue-enhancing for the teams involved and college football as a whole.

Here's a hypothetical example. The dollar amounts are approximates, but the details are generally right and that's all that matters for this example.
Scenario 1: Ohio State plays Toledo, and Texas plays Wyoming.
Ohio State pays, e.g., Toledo, e.g., $750,000 to play one game at Ohio Stadium. Ohio State earns, let's say, $2.75 million off a home game. They pay $750k to Toledo and walk away with $2 million. Texas schedules Wyoming to play one game in Austin, pays them $750k for a home game, earns the same $2.75 million, and walks away with the same $2 million net.
Scenario 2: Ohio State plays Texas, and Toledo plays Wyoming.
Ohio State and Texas decide to play a home-and-away. That's how this normally works. I'm less sure of this works, but let's say to help even out revenue from year-to-year, Texas and Ohio State each give the other $1 million when they go to the other team. Let's say Texas and Ohio State earn a little more in TV revenue, so instead of $2.75 million for the home game, they earn $3 million. Meanwhile, Toledo and Wyoming play each other; I'm not sure what the numbers here quite look like, but let's say it's $750k in revenue and a $250k payment to the other team.

In that case, here's your bottom line financial picture for one year of each series:
Scenario 1: Ohio State earns $2 million, Texas earns $2 million, Toledo earns $750k, and Wyoming earns $750k. Total revenue: $5.5 million.
Scenario 2: Ohio State earns $2 million, Texas earns $1 million, Toledo earns $500k, and Wyoming earns $250k. Total revenue: $3.75 million.
That is your bottom line: it simply does not make financial sense for individual marquee teams to play other marquee teams on a regular basis in non-conference play. The only explanations I can come up with for Wetzel et al.'s failure to mention that in their chapter on non-conference scheduling are: (a) they don't know that and are therefore incompetent, or (b) they do know that and are deliberately withholding that extraordinarily important (to me) fact and are therefore mendacious.

The other problem with this chapter is, well, it's actually wrong. Non-conference scheduling HAS in fact mattered a great deal in making BCS decisions. Oklahoma making the Big XII championship game over Texas (and Texas Tech) in 2008 because they had a higher BCS ranking was partially the result of Oklahoma's superior non-conference scheduling (and actually winning those games). Ditto the Sooners making the title game over Auburn in 2004.

It was at this point in the book (Chap. 9, pp. 91-100) that I threw down Death for perhaps the sixth time and view to stop reading it. I decided, however, that before I declared a trio of national journalists to be either incompetent or mendacious, I should finish reading the book.

IV. Determining the Best Team
One of the recurring problems that Wetzel et al. consistently deal with, but never address directly is the difficulty of determining the best team. The closest they get here is in their chapter on how awful the computer rankings are and the chapter on the Harris Poll voters, but the real problem is something else:

JOURNALISTS ARE BAD AT PICKING THE BEST TEAM, BUT DON'T CARE BECAUSE THEY'VE BEEN THE ONES WHO'VE DONE IT.

Okay, that's probably a little strong. Here's the more nuanced story:
1. The national champion in college football has historically been the team ranked #1 in the AP poll.
2. The voters in the AP poll are journalists, most of whom cover a single college football team's games every Saturday. They watch more college football than most people, but less than many hardcore college football fans.
3. When the BCS was created, the powers-that-be (SEC Commissioner Roy Kramer being the main driving force) created a hybrid formula composed of the most famous computer rankings and the human polls (AP and coaches).
4. Any time the BCS formula produced a BCS Championship Game with teams other than AP #1 and #2, the AP pollsters threw a tremendous hissy fit and the BCS formula was adjusted so that in future years that exact case would produce a BCSCG with AP #1 and #2, even though the instant result was never clearly wrong.
5. The flaws of the Harris Poll Wetzel et al. point out aren't really a problem, because of the anchoring effect created by the AP poll. Ditto also the non-problem not solved by not doing the Harris Poll until October; there will still be preseason rankings and in-season rankings done by various people, and until those are outlawed (or fixed, to the extent such a thing may be possible), the Harris Poll cannot be truly fixed.
6. At no point do Wetzel et al. mention all of the well-known problems with the results of polling, most notably poll inertia where teams that win rise and teams that lose fall. An individual weekly poll is almost always the result of poll momentum, not a comprehensive rating of the relative strength of the teams in the poll.

The "solution" they advocate is the creation of a selection committee. This is something I'd advocated myself in the past, because it's a back-door way of creating legitimacy with journalists any time you have a result other than What The AP Poll Says. I'm skeptical this will actually work, given that, as noted above, journalists are incompetent whiners who've been important historically. Why exactly Wetzel et al. want a selection committee and what sort of criteria this selection committee would use is left clear as mud; I have no clue if they recognize just how bad the human polls are, or if they skate around the topic without mentioning it directly because they don't see or believe that the human polls are bad. Without a clear statement that the selection committee should completely ignore the polls the way the men's basketball selection committee does, a selection committee is not in my mind much of an improvement.

UPDATE #2: I realized that I neglected to write in more detail about the BCS's use of the computer rankings. There are two problems here. First, Richard Billingsley's rankings are included because Billingsley's been ranking college football teams since the late 1960's even though Billingsley's ranking system is insane. Second, the computer rankings have been neutered by removing margin of victory because including margin of victory was blamed for producing results other than AP#1 v. AP#2. The BCS isn't interested in good computer rankings because the pollsters are only interested in good rankings if the good rankings agree with the pollster rankings.

V. The "Cartel"
The Cartel is how Wetzel et al. refer to the power conferences that were responsible for the creation of the BCS and are the primary beneficiaries of the BCS.

The power conferences also have the primary allegiance of almost all fans of most college football, had all of the most important bowl tie-ins, perennially have most to almost all of the best 10, 20, and 40 teams, and in short provide substantially all of the value. Death mentions the memorable 2007 Fiesta Bowl game between Oklahoma and Boise State; what they don't mention is that game got horrible television ratings. The TCU-Boise State Fiesta Bowl this year actually had non-terrible ratings (that was the Iowa-Georgia Tech Orange Bowl), but both those teams were undefeated and have had a decent run of recent success. That's the same reason the BCS conferences control the BCS: they've had historical success and the large alumni bases that can deliver television ratings points. That's where the value is, and that's why the control the BCS. Weird how control follows money, though that's maybe the corporate lawyer in me again.

VI. Quick Hits

  • Wetzel et al. treat the 16-team playoff as though it's a novel concept, even though it's the only feasible playoff if you abandon the bowl system.
  • The best argument for a 16-team playoff is one they don't make, namely the lack of connectedness in college football and concomitant difficulty in comparing teams across conferences makes it impossible to identify the best teams, so we'll done as well as we can and hope for the best, and that's a 16-team playoff.
  • Minor bowls would become completely meaningless in a playoff, and sponsors and television would be less willing to pay enough to keep them viable. Some of them, probably 10-25 of the extant 36 or so, would die if a 16-team playoff were implemented.
  • Individual regular season games would, I believe, see their ratings decline. This LSU-Alabama game I'm watching is a de facto BCSCG elimination game for both teams. If it's not even a playoff eliminating game, which would be quite possible under a 16-team playoff, I'm probably watching stuff off my DVR instead. This underrating of the casual fan most interested in the most important games is persistent throughout Death.
  • Even the traditional bowl system was less popular than the NFL; this can be seen by comparing even pre-BCS ratings to NFL games played the previous or subsequent day. A college football playoff will not match the ratings of the NFL's playoffs.
  • Nobody aside from former Pac-10 commissioner Tom Hansen thinks going back to the old bowl tie-in system would be better than the BCS, and Hansen probably would get rid of the internet and maybe even the telephone as new-fangled inventions of dubious value.
  • Housekeeping notes: I call them I-A and I-AA and the Big 10 because I want to, and know and don't care that they're officially otherwise ("FBS", "FCS", and "Big Ten" respectively). I-AA is also consistent with Death, not that that's a positive indicator. I also got tired of typing "Wetzel et al.", so read that any time it just says "Wetzel."
  • The revenue split from a 16-team playoff is not addressed.


Conclusion
Anyway, I've ranted enough. I've hit some topics Wetzel et al. didn't cover quite the same way I did, and haven't hit every topic they did, but I think I've given you a brief overview of my thoughts on why Wetzel et al. have written a bad book you shouldn't read. Emphatically not recommended.

See also the book's website and the book's twitter feed if you want to see more of what they have to say, not that I recommend doing so. See also Joe Posnanski's review, the least bad that I saw. Posnanski also gave BCS executive director Bill Hancock space to respond, not that he uses it particularly well (or could, for political reasons).

UPDATE (11/6 1742 CT): Doing some minor cleanup and proofreading; one point of order: it was pointed out to me that the normal reason for CFB not playing on Jan. 1 was because it was a Sunday, not because of the NFL conflict. Fair enough, so I've revised that bullet point to reflect CFB's lower ratings compared to NFL games the next or previous day.
UPDATE #2 (11/13 0015 CT): I finally made it through and re-read the review with clear eyes, and cleared up a few times where I'd had the wrong word and some grammar/readability issues. I also added a paragraph on the computer rankings in determining the best team (marked UPDATE #2) which I neglected to include in the initial post.
UPDATE #3 (9/1/11 1116 CT): Some numbers courtesy of MGoBlog suggesting the revenue cost of giving up a home came is in the neighborhood of $3.9 million.

1 comment:

sunpar said...

"They comprehensively don't do this, in some cases quite the opposite. I'm pretty sure I know some of the reasons why, but Wetzel et al. don't acknowledge any of the unclear priorities."

I researched the book that you linked in this sentence but it's still unclear exactly what the reasons are. Would you mind briefly explaining exactly what you meant by this sentence?