Odds are, not many LSU or Boise State fans were exactly riveted to Appalachian State's 42-14 blowout Saturday over Western Illinois in the first round of the FCS playoffs. And virtually no fan of any stripe would have been able to tell you what the Mountaineers' win could possibly have to do with the final Bowl Championship Series standings, released on Sunday night to kick off an entirely different postseason on the FBS level. But Wes Colley's computer could.
Somehow, the domino effect from the omission of the Mountaineers' win from Colley's computer rankings – one of the six computer polls that make up a third of the BCS formula – was enough to bridge a razor-thin margin keeping LSU behind Boise State at the bottom of the top 10. When the final standings came out on Sunday night, the Tigers ranked 10th, and Boise State 11th. After independent BCS guru Jerry Palm caught the error in Colley's data, the revised standings moved the Broncos to No. 10, giving them their fourth top-10 finish in the final standings in five years.
And that changes … what, exactly? This time, nothing. Neither team was selected as an at-large pick for a big-money bowl game. Whatever the order, LSU would still be scheduled to play Texas A&M in the Cotton Bowl, and Boise State to play Utah in the Las Vegas Bowl, matchups that were locked in before the BCS numbers were even released. It's hardly a matter of simple pride, either, since there are no BCS standings after the bowl games – final rankings are determined exclusively by the human polls, which certainly are not taking into account what's happening in a lower-division playoff game in Boone, N.C., or anywhere else.
What matters, as Palm writes, isn't that there was a mistake (of course there are occasional mistakes) but that it took an outsider to find it. And it was only possible to find it because Colley is the only one of the six BCS computer poll masters who makes his method available to the public. The other formulas are closely guarded as "proprietary" information. Monday's update was the first post facto revision of the standings in the 13-year existence of the BCS, meaning either a) There has never been a similar "glitch" in the computer polls; b) The glitches have never affected the final outcome, and have been revised without notice; or c) The glitches have never been caught, because no one is looking at the numbers.
The BCS says it was created solely to match the No. 1 and No. 2 teams in a championship game, but the rankings affect more than just the teams at the top. Any at-large team that finishes in the top four is guaranteed a spot in one of the big-money bowls; if the same team finishes fifth, it (and its conference, which receives the paycheck) can be left out in the cold. Two years ago, the Big 12 broke a controversial three-way tie in the South Division by deferring to the final BCS standings, sending Oklahoma to the conference championship game (and eventually the BCS Championship Game) ahead of enraged Texas. This year, the Big Ten is sending Wisconsin to the Rose Bowl ahead of co-champions Ohio State and Michigan State because the Badgers had the highest finish in the BCS – fifth, one spot ahead of the Buckeyes at No. 6. Some coaches have contract incentives for finishing in the top 10.
If the outcome of an obscure game most of the country wasn't aware was taking place can directly affect the standings with a single keystroke, how many times could an omission or typo have affected the numbers over the years in the computer polls that refuse to subject themselves to outside scrutiny? How easily could it affect who finishes No. 2 instead of No. 3, and therefore who moves on to the self-ordained national championship game? How easily could it be intentionally manipulated?
Computer polls are extremely useful for connecting more threads and bringing more relevant information to bear than any human ever possibly could – a handy tool when trying to compare dozens of teams with occasionally contradictory outcomes and relatively few overlaps in the data. But even the champions of wonky statistical analysis in sports have consistently panned the BCS for its lack of transparency and its lack of interest in computer polls as anything but a veneer of "objectivity" for doing whatever the human polls want to do in the first place. Even the guys who run the BCS computers admit they're forced to "sacrifice a bit of accuracy" because of restrictions like the prohibition on margin of victory. Most non-techno folk are inherently distrustful of the machines to begin with. If it takes an external watchdog to get the numbers right in the only poll he's actually allowed to watch, the credibility gap moves that much closer to irreparable.
- - -
Matt Hinton is on Twitter: Follow him @DrSaturday.
Somehow, the domino effect from the omission of the Mountaineers' win from Colley's computer rankings – one of the six computer polls that make up a third of the BCS formula – was enough to bridge a razor-thin margin keeping LSU behind Boise State at the bottom of the top 10. When the final standings came out on Sunday night, the Tigers ranked 10th, and Boise State 11th. After independent BCS guru Jerry Palm caught the error in Colley's data, the revised standings moved the Broncos to No. 10, giving them their fourth top-10 finish in the final standings in five years.
And that changes … what, exactly? This time, nothing. Neither team was selected as an at-large pick for a big-money bowl game. Whatever the order, LSU would still be scheduled to play Texas A&M in the Cotton Bowl, and Boise State to play Utah in the Las Vegas Bowl, matchups that were locked in before the BCS numbers were even released. It's hardly a matter of simple pride, either, since there are no BCS standings after the bowl games – final rankings are determined exclusively by the human polls, which certainly are not taking into account what's happening in a lower-division playoff game in Boone, N.C., or anywhere else.
What matters, as Palm writes, isn't that there was a mistake (of course there are occasional mistakes) but that it took an outsider to find it. And it was only possible to find it because Colley is the only one of the six BCS computer poll masters who makes his method available to the public. The other formulas are closely guarded as "proprietary" information. Monday's update was the first post facto revision of the standings in the 13-year existence of the BCS, meaning either a) There has never been a similar "glitch" in the computer polls; b) The glitches have never affected the final outcome, and have been revised without notice; or c) The glitches have never been caught, because no one is looking at the numbers.
The BCS says it was created solely to match the No. 1 and No. 2 teams in a championship game, but the rankings affect more than just the teams at the top. Any at-large team that finishes in the top four is guaranteed a spot in one of the big-money bowls; if the same team finishes fifth, it (and its conference, which receives the paycheck) can be left out in the cold. Two years ago, the Big 12 broke a controversial three-way tie in the South Division by deferring to the final BCS standings, sending Oklahoma to the conference championship game (and eventually the BCS Championship Game) ahead of enraged Texas. This year, the Big Ten is sending Wisconsin to the Rose Bowl ahead of co-champions Ohio State and Michigan State because the Badgers had the highest finish in the BCS – fifth, one spot ahead of the Buckeyes at No. 6. Some coaches have contract incentives for finishing in the top 10.
If the outcome of an obscure game most of the country wasn't aware was taking place can directly affect the standings with a single keystroke, how many times could an omission or typo have affected the numbers over the years in the computer polls that refuse to subject themselves to outside scrutiny? How easily could it affect who finishes No. 2 instead of No. 3, and therefore who moves on to the self-ordained national championship game? How easily could it be intentionally manipulated?
Computer polls are extremely useful for connecting more threads and bringing more relevant information to bear than any human ever possibly could – a handy tool when trying to compare dozens of teams with occasionally contradictory outcomes and relatively few overlaps in the data. But even the champions of wonky statistical analysis in sports have consistently panned the BCS for its lack of transparency and its lack of interest in computer polls as anything but a veneer of "objectivity" for doing whatever the human polls want to do in the first place. Even the guys who run the BCS computers admit they're forced to "sacrifice a bit of accuracy" because of restrictions like the prohibition on margin of victory. Most non-techno folk are inherently distrustful of the machines to begin with. If it takes an external watchdog to get the numbers right in the only poll he's actually allowed to watch, the credibility gap moves that much closer to irreparable.
- - -
Matt Hinton is on Twitter: Follow him @DrSaturday.
