The BCS, Not the Polls, Got It Right
by Jeff Anderson
(published December 2003)
Alexis de Tocqueville, a keen observer of American culture, wrote that,
in America, “[t]he majority lives in the perpetual utterance of self-applause.”
Seldom has this been more evident than in the aftermath of the release
of the final 2003 BCS standings. The response on ABC, ESPN, and throughout
much of the print media, has largely proceeded along these lines:
“Both polls have USC #1, while the BCS has USC #3; therefore, the BCS is
clearly wrong.”
This, of course, assumes—without argument—that the polls are right.
It is somewhat shocking to hear almost no one even attempt to make an argument
to support the polls’ conclusions. It appears that no argument is
thought necessary. Apparently, popular opinion, as registered by
the polls, does not need to be defended as being correct—it is self-evidently
correct. By this standard, the BCS standings will be “clearly wrong”
every time they disagree with the polls. But if that is true, then
why bother with the final three quadrants of the BCS standings—the three
more objective quadrants? Why not just use the first quadrant, the
subjective polls? Then the BCS can be “correct” every time.
The reason why it wouldn’t be good to do this, of course, is that the
subjective polls are not really very accurate at all. The polls prejudge
teams before the season starts. Then they combine these prejudgments
with a process of attrition, whereby you generally can’t move up unless
the team in front of you loses (so get in line early!). The whole
process works sort of like a tennis ladder, except that you don’t necessarily
move up when you win; you just drop when you lose. (You move up when
other teams lose.) In addition, the polls usually display an evident
bias against teams that play tough schedules and against West Coast teams.
(USC may seem like the exception to this rule, but not really. With
USC’s playing a very average schedule and being a West Coast team, these
two factors have more or less evened out.)
Maybe almost no one is even attempting to make reasonable arguments
to defend the polls’ choice because almost everyone realizes USC is #1
primarily because the Trojans lost two weeks before LSU. (And it
doesn’t hurt that USC started higher in the preseason, either.) If
LSU had lost first, then the Tigers would likely be #1. If Oklahoma
had lost first, then the Sooners would almost certainly be #1. I
suspect there are even poll voters out there who would favor Oklahoma in
a hypothetical matchup with USC and who would also say Oklahoma has had
a better season, yet who still voted USC #1 and Oklahoma #3. Such
is the way of the polls: it’s all about where you start and when
you lose.
One final point about the polls (which, one might add, were hardly unanimous
in their selection of USC as #1—LSU and Oklahoma combined for 41% of the
1st-place votes in the coaches’ poll and 35% in the writers’ poll, an unusually
high level of disagreement in the lock-step world of the polls):
Ask a random poll voter what he or she is even trying to measure, and it
is doubtful that you will get a particularly concrete answer. Is
the goal to measure which team has had the most season-long success, or
which team is playing the best football right now? These are not
the same things, yet most polls voters have not sufficiently sorted this
out in their minds. Instead, they apply a hodge-podge of these two
criteria in an ad-hoc manner, varying the weight of each from week-to-week
and season-to-season, based largely on which teams’ highlights look better
on ESPN and what names are stitched on teams’ jerseys.
The BCS was designed to provide a more accurate level of evaluation
than this. More specifically, it was designed to combine two essential
things: popular opinion and objective analysis. It has done
so quite successfully. Objectively, Oklahoma and LSU were the
two most accomplished teams this regular season. As such, they have
done the most to earn Sugar Bowl bids and are the most deserving of receiving
them. This was the conclusion of six of the seven BCS computer
rankings as well as of what you might call the BCS’s own internal rankings
(the combination of strength of schedule points and losses). The
only objective entity that disagreed with this conclusion—the New York
Times computer rankings—also ranked Oklahoma’s road-kill (Texas) ahead
of Oklahoma.
The truth is, the BCS got it right. The BCS has provided a
true national championship game whenever one has presented itself (Miami
vs. Ohio St. last season), just as it was designed to do. This
has prevented a recurrence of the events of 1997 (Michigan and Nebraska),
1995 (Penn St. and Nebraska), 1991 (Washington and Miami), and beyond,
when two teams finished the season undefeated but didn’t have the opportunity
to play each other. Whenever only one team has clearly emerged from
the pack, the BCS has always designated that team as #1 and provided it
with a berth in the title game (Miami in 2001, Oklahoma in 2000, Florida
St. in 1999, and Tennessee in 1998). Each time, that team has won
the national championship. (To be sure, there have been disagreements
about whether the BCS has selected the most deserving opponent to play
that #1 team, but the BCS-selected opponent has always at least been a
solid, defensible choice.) This year, when no clearly deserving team
has emerged—in fact, every team this season has a lower rating in the Anderson
& Hester Rankings than all past BCS title game participants apart from
Virginia Tech in 1999—the BCS is presenting a title game that matches the
top two teams according to a combination of subjective and objective standards
of evaluation: the best standards available. And now we get
to watch more than one bowl game with particular interest and then debate
the merits of the winning teams in barrooms and living rooms across America
just like in the old days—and that is enjoyable as well, and good for sustaining
interest in the game.
Perhaps most importantly, for six years the BCS selection process
has accentuated week-to-week regular season drama in a manner beyond that
which anyone could have predicted. Move to a playoff, with conference
champions guaranteed to receive most of the slots, and we will be right
back to where no one pays much attention to non-conference games or to
games played in other conferences. And we will have lost the
most unique and distinctive ending to a season in all of sports:
the bowls. (Then again, as Tocqueville writes, Americans love to
make everything equal. We love sameness.) The impassioned fans
who attend the bowls will be replaced by corporate suits—the real fans
being unable to travel from site to site on such short notice. The
guy sitting on his couch in his undershirt, munching on potato chips, might
be happy with this arrangement, but college football will have lost much
of its traditional richness. And will the best team truly emerge
from this playoff? Does the best team always truly emerge from the
NCAA basketball tournament? Is Kansas St. truly the best team in
the Big 12?
The BCS has done its job, and it should be celebrated, not maligned.
Oklahoma and LSU should be playing in the national title game. The
polls should not be assumed to be right without debate, and in truth they
are wrong. USC has beaten only one team in anybody’s (the BCS,
the AP, the coaches, etc.) top-25, and the Trojans have lost to (7-6) Cal.
How is that the resume of a true national championship contender?
Not only has the BCS clearly gotten it right, but the polls have clearly
gotten it wrong. But don’t expect most people to challenge the perceived
authority of the polls, however arbitrary or capricious they are, for they
largely reflect popular opinion, and Americans are loathe to challenge
prevailing popular views. As Tocqueville writes, “In America, the
majority raises formidable barriers around the liberty of opinion:
within these barriers, an author may write what he pleases; but woe to
him if he goes beyond them.”
I would love to see college football fans prove Tocqueville wrong, at
least in this instance, by demonstrating a willingness to look fairly and
impartially at the true accomplishments of the three teams in question.
If they do so, I am confident they will see the true merits of the BCS’s
selections—whether or not they ultimately come to agree with them.
Good work, BCS! Keep it up!
Postscript:
While the BCS has done its job very well, it is not perfect, and
not all of its components are created equal.
Here are the top-10 teams according to the BCS computer rankings:
1. Oklahoma 1.17
2. LSU 1.83
3. USC 2.67
4. Michigan 4.67
5. Ohio St. 5.50
6. Miami, Ohio 6.00
7. (tie) Florida St. 6.83
7. (tie) Texas 6.83
9. Miami, Fla. 8.17
10. Tennessee 9.50
*Note that the largest gap, by far, is between #3 USC and #4 Michigan
(a 2.00 gap), as should be the case.
Comparatively, here are the top-10 teams according to the somewhat-less-accurate
BCS internal rankings (the combination of strength of schedule points and
losses)—which make up half of the BCS standings (before the quality
wins adjustment) and thus count for twice as much as the computer rankings
or the polls, or for the same amount as the computer rankings and polls
combined:
1. Oklahoma 1.44
2. LSU 2.16
3. Ohio St. 2.28
4. USC 2.48
5. Miami, Fla. 2.52
6. Michigan 2.56
7. Florida St. 2.60
8. Texas 2.80
9. Kansas St. 3.40
10. Iowa 3.64
*Note that the largest gap is between #1 Oklahoma and #2 LSU (0.72),
which is not a convincing place for a large gap. And USC is ranked
#4—behind 2-loss Ohio State. In fact, USC is notably closer to #7
Florida St. (0.12 ahead), #6 Michigan (0.08 ahead), and #5 Miami, Fla.
(0.04 ahead) than to #2 LSU (0.32 behind)—and the Trojans are closer to
#9 Kansas St. (0.92 ahead) than to #1 Oklahoma (1.04 behind). This
relatively low placement of USC is very hard to justify.
In sum, the BCS computer rankings are notably more accurate than
the BCS internal rankings.
|