KING JAMES BIBLE: Human Polls, Human Error
Just as in 2003, the Crimson head into this weekend’s matchup with regional powerhouse Northeastern undefeated, but still remain unranked in two of the most widely cited I-AA polls.
In fact, those polls—the ones provided by Sports Network and ESPN/USA Today—are the only I-AA rankings indexes to leave Harvard out of the top 25, placing them behind seven teams with two losses and one with three.
To illustrate how out of touch the voters in these two polls are, let’s surf on over to I-AA.org, a site which has compiled a rating formula know as the Gridiron Power Index (GPI). The GPI is the cheap, generic knockoff of the Bowl Championship Series (BCS) formula, as it uses the rankings from eight computers and three human polls—the two mentioned above and the one from AnyGivenSaturday.com, which currently has Harvard sitting 24th.
While the GPI may be the CVS brand Minoxidil to the BCS’s Rogaine, the I-AA version is very still instructive and is one of the top predictors of which teams will ultimately qualify for the playoffs—which, incidentally, the Ivy League is still barred from participating in by its “athletic-savvy” presidents.
In 2002, it forecasted seven of the eight at-large bids correctly—only missing Wofford, which itself was so stunned that it hung effigies of NCAA I-AA playoff selection committee members. Last year, it went seven-for-eight again, as it incorrectly predicted Lehigh would get the final at-large spot over eventual recipient Bethune-Cookman.
Four of the eight computers used in the GPI have Harvard ranked in the top 10 in I-AA. If the GPI used the Sagarin rating that the BCS employs—the Elo-Chess category, which doesn’t include the ever contentious margin of victory—Harvard would be ranked third in I-AA and 48th in all of Division I. (Obviously, I don’t intend to defend the latter statistic, but rather I included it to illustrate the ridiculous nature of some of the constraints of the BCS system, such as removing margin of victory from the calculations. That’s a topic for another day though).
The Sagarin ranking that the GPI actually uses places Harvard 16th—the same ranking the Crimson received in the Matthews index. Throw in a 15th and an 18th in the final two computer indexes, and it’s pretty clear that the average ranking (12.38) spit out by one’s trusty IBM lifts the Crimson well out of the dubious Other’s Receiving Votes column.
The GPI, however, averages the computer and human rankings, dropping the highest and lowest number from among the eight machines, adding in the three human polls and dividing through by nine.
Currently, the result of that excursion into mathematics leaves Harvard as the 16th best team in I-AA. Northeastern, its opponent this weekend, sits at No. 10 in the GPI and enjoys a top four ranking in three different computer polls, as well as a slot in the top 20 in each of the human polls.
And that’s why this weekend is about respect.
After defeating the Huskies 28-20 last season, the Crimson moved into the two major human polls for the first time since a 36-35 loss to Lehigh in 2002 and opened up at eighth in the GPI.
With seven of the teams in front of Harvard taking on other ranked opponents and four of those going head-to-head, a win over Northeastern could once again propel Harvard into the top five of a few computer ratings, the top 10 of the GPI and the top 20 of at least one human poll.
From there, if the Crimson were to survive trips to Princeton, Hanover and Philadelphia as well as home dates with lowly Columbia and the roller-coaster ride known as Yale, it could challenge the Ivy record for highest finish in the GPI—fourth, when Penn finished 9-1 two years ago.
Harvard could even take home the top spot in one of the eight computer polls—the Quakers took home a second in 2002.
But the human poll plague that befell Penn last season has already taken its toll on the Crimson. No matter what Harvard does, there’s a ceiling on how high it can rise in the estimation of Sports Network and ESPN Coaches’ Poll voters. Last year, the Quakers went 10-0, but could only rise as high as eighth with one week to go in the season. Penn actually fell to 12th in the final poll, because, presumably, its 59-7 win over Cornell in a meaningless game wasn’t convincing enough. And Penn began its climb up the human poll ladder much earlier than this year’s Crimson squad.
I like rewarding outstanding accomplishments, so while everyone drools over next week’s I-AA rankings, I’ll be paying my respects to the one index that is excellent at what it does. Who knows, maybe if Harvard drills Northeastern this weekend, those human polls might begin to do the same.
—Staff writer Michael R. James can be reached at email@example.com. His column appears every Tuesday.