NCAA Football, and the Science of SubjectivityS

With true-to-life fidelity, my most recent season simulation in NCAA Football 10 found Boise State losing a trap game late in the season and, as the token BCS Buster from a minor conference, paying for it dearly in the polls.

Having gone undefeated through 10 games, the Broncos (not a user-controlled team in this dynasty) reached the BCS Top 4, striking distance of Florida, Oklahoma (with an uninjured Sam Bradford) and Alabama. The week 12 standings were strongly analogous to present day standings, absent TCU and Cincinnati, both undefeated in the real world.

And then Boise fell at home to Nevada, tumbling far out of both voting polls' top 10, and to 12th in the BCS. The machine held the lesser-conference team to the same double-standard as the human voters, who have matched one-loss teams from the major conferences in the previous two title games. Further, Boise State quarterback Kellen Moore (sorry, "QB #11") who had led the Heisman voting to that point, bottomed out to third in the final tally. Finally, a two-loss Oregon (to Boise and to Utah) leapfrogged the Broncos, as many expect the real-life one-loss Ducks will do once pollsters realize their votes will affect on a national title and major bowl bids.

The plausibility of all this is not just dumb luck. Games in November always bring into sharp focus just how American college football's poll-driven, playoff-less season and postseason is the most meritless selection of a team champion in the entire world. And yet NCAA Football 10, unlike any other sports simulation, has the responsibility to simulate the same purely subjective conditions, which aren't just the subplots of a season, they are the season itself.

"With sports there is always going to be controversy," said Kendall Boyd, a senior product manager for NCAA Football 10. "We do our best to make sure we have a little drama, but also ensure the integrity of the current system."

As college football, with seven unbeaten teams, lurches toward another inevitable matchmaking controversy, this week I tugged on EA Sports' shirttail, asking how they build a game that, if it mirrors reality, should also screw some deserving team out of a title shot every year. Boyd didn't answer that question head on, and I really can't blame him - the hypercriticized Bowl Championship Series is one of the game's biggest licensors, after all, and it's not doing so to be held up to ridicule or split polls in virtual reality, too. But he did shed light on how NCAA Football incorporates reputation into its team and individual performances.

"The biggest factor of our ‘human element' is leveraged against your conference's prestige," Boyd said. "If you play in a BCS conference, you're going to move up the rankings a lot easier than a smaller conference school would."

Conference prestige - this is different from the six-star rating each program has in NCAA 10 - comes most into play in the game's simulated coaches' poll, the human factor most driving the game's BCS rating. The coaches' poll routinely favors programs from the major conferences, as 33 of 59 voters in this year's poll represent, and still others have previous experience with them.

Boyd said that once the season gets moving, "our media and coaches' poll are very similar." NCAA 10's media poll is, of course, analagous to the AP Top 25, which asked out of the BCS formula five years ago but remains an influential measure for judging the biggest matchups week to week.

NCAA Football, and the Science of Subjectivity

"The previous week's rating is evaluated," Boyd said, "and then the following factors are brought in: score versus opponent that week; was it a game on the road? What was the ranking on our ‘Toughest Places to Play' poll versus the opponent you played, if it was road game? And then finally, the separation between the two is simple percentages, so we have a disparity between them."

The game's BCS computation comes into play in Week 8, the same as it does in real life, but it is not a strict replication of the actual matrix. For one thing, the Harris Interactive Poll, which serves as the second human poll in the BCS formula, isn't a factor all that distinct from the game's media poll. And the six indices - with names like Colley, Sagarin, Wolfe and Massey - that form the ranking's computer average are not used in NCAA 10, Boyd said, even though their formulae are public. "We do make it equally interesting," Boyd said. "Without giving too much away, we combine the media and coaches' poll and then add in other variables, such as strength of schedule."

Such as? "Quality wins and losses are a big factor. Losing to a bad team will definitely have a severe impact on the rankings in our game." Also, timing is a key factor, just like real life. "In our game, it's better to lose early than lose late," Boyd said. "If you were to lose in the first few weeks of the season to a strong opponent, you will naturally move up the rankings as long as you continue to win."

The biggest question I had is whether NCAA 10 internally gooses the polling to help out a user-controlled team, in the name of a more fun video game for the person who bought it. Because in more than six years of playing console sports sims, few experiences have been more gratifying than taking over a two-star doormat, storming the Top 10, and getting that "Where'd They Come From?!" headline in the next week's NCAA news.

Answer: No. "We want it to be an even playing field," Boyd said. If you manage to take Temple to the Orange Bowl, you came by it honest. "I believe most of the ways we evaluate the teams would be affected if we skewed it toward the human-controlled teams," he added.

Nor is the voting skewed toward user-controlled players in the game's Heisman Trophy simulation. However, "We do have a special circumstance for potential upsets in the voting to keep it dynamic, for a twist," Boyd said, "but we don't want to disclose the formula, to help keep the intrigue. But this is equal among human controlled and CPU teams."

NCAA 10's Heisman voting likewise reflects the values of its real-world counterpart. It typically goes to quarterbacks, running backs and wide receivers, although I have seen offensive tackles and defensive backs get mentioned week-to-week, as they sometimes are in real life.

Significantly, Boyd said that the stats or results of a simulated game in a dynasty carry no additional weight, positive or negative, in the game's Heisman voting. And while it's easier to load up arcade numbers against creampuffs, he said a surer path is to take on tough teams on the road and log credible performances that contribute to a win there.

And no, Boyd said, there is no East Coast Media Bias helping players or teams from that region, in either the polls or the Heisman sims.

After our conversation, I went back into NCAA 10 to try to test out what Boyd had to say. I ran another simulation pitting Boise State versus a much tougher nonconference schedule this year. The Broncos went 8-3, losing to Oregon at home and Alabama and Texas in Tuscaloosa and Austin. Boise still ended the season at No. 11 - remarkably, the highest-rated three loss team in the nation, although all of the defeats came early. In fact, 11 is an uncommonly high rating for any three-loss team, let alone one from the WAC. Strength of schedule, with two Top-5 games on the road early in the season, clearly was in play here.

NCAA Football, and the Science of SubjectivityS

But it was impossible not to notice that a lesser team, North Carolina - whose football ranking I've long said is propped up by the school's basketball reputation and the votes of people who wish they went there - had hit No. 2 by the end of the regular season on a schedule as weak as the last swallow in a 2-liter of Cheerwine. And that literally raised up the old State alum anger in me, seeing the despised Tar Heels exalted by a system that would never ever give the Wolfpack the same benefit of the doubt, which is pretty much how we think about things in real life, too.

But then in the conference title game, Carolina suffered the kind of crushing loss that is so common to arriviste college football programs - 28-13 to Clemson, the ACC's original football power, booting UNC back to a lesser bowl, the Gator. And I threw a fist and roared with delight at, again, the true-to-life fidelity of NCAA 10.

Stick Jockey is Kotaku's column on sports video games. It appears Saturdays at 10 a.m. U.S. Mountain time.