May 23, 2008

Fixing the BCS: You should start with the coaches poll

The Men of Scarlet and Gray point out the elephant in the room leading to the push for a playoff in college football: the current poll system is broken and simply not getting it done.

But in all the arguments for and against, nobody has yet stated what the actual root problem in CFB is:

It’s the poll system that’s the problem, people.

It’s the poll system that’s broken. Unless the polls are fixed, any changes to CFB will be an exercise in futility.

The primary issue with the BCS is not that it’s unable to match up nos. 1 and 2 - it does that just fine. The problem is that it contains no institution to ensure that the teams ranked number 1 and 2 are in fact the two best teams in the country, as the BCS rankings themselves are compiled using a flawed poll system.

Playoff advocates bleat on about “settling it on the field,” but their own systems are just variations of the same flawed concept: “we’ll just seed the top X teams…”

Top ‘X?’

Whatever the ‘X’ is, what institution decides who those teams are? What decides exactly where those teams relatively lie in that list of four/six/eight/etc.?

I couldn't agree more - the polls are broken. And none more so than the Coaches Poll.

Texas coach Mack Brown a different take on the Coaches Poll: don't let the coaches vote at all.

“I would like to see coaches not have a vote,” Texas coach Mack Brown said. “We would have a panel like the one that chooses the 65 teams in basketball.”

That’s what ails the BCS — credibility. Every coach participating in the American Football Coaches Association poll has a vested interest. The Harris poll has some credible voters but not enough. As Brown said, “For someone to say, ‘Mack, can you honestly vote without saying how it affects your team?’ That’s a hard thing to do.”

Mack's right - it's hard not to be aware of your own team's fate as you put pen to paper.

And any system that allows a coach to vote Duke into the poll at the start of every season is broken beyond all repair.

Well, maybe not broken beyond all repair.... Senator Blutarsky suggests implementing mechanism design into the coaches poll:

But I did find an interview with Roger Myerson, one of the three economists who won a Nobel Prize for their work in this area, who had this to say about applying the concepts of mechanism design to voting:

… If there are 10 candidates on the ballot running, say, in a Democratic primary, you can vote for several of them, but you can’t give more than one vote to any one of them. And the person who gets the most votes, most approval votes, was approved by the largest majority, gets the bonus of winning the votes of the state.

And with that, the light bulb went off over my head. Instead of letting each coach rank in order the top 25 schools and then add up the ranked totals, why not simply have them each turn in his list of the ten best schools in the country - unranked? The teams would then be ranked in order of those which received the most votes. With 600 some odd votes for teams, you’d still be able to construct a top 25 fairly easily, I would think. (And if not, you could always let the coaches vote on a bigger group, such as a top twelve or fifteen.)

The beauty of this approach is that it minimizes, if not eliminates, the effect of any individual coach’s bias or conflict.

Blutarsky's revolutionary approach -- in which coaches would simply name ten elite teams without ranking them, per se -- would eliminate some of the gaming of the coaches poll system that has taken place in recent years.

Ranking a team in your elite ten would carry the same weight because the coaches wouldn't be able to assign a mathematical significance to any team. Under such a system, coaches wouldn't be able to rig their top ten in an attempt to influence the BCS (ie Wyoming's Joe Glenn ranking Georgia #10 last year).

Once the votes are sent in to the poll administrator, then the statistical rankings would be compiled, producing the poll results.

But would a system in which none of the pollsters can rank the teams even work?? Umm... yeah.

I decided to kill an hour last night by going back and retallying the 2007 final regular season coaches ballots on the basis of top five and top ten votes in order to see how my “fix” might have worked. I wasn’t expecting a big change in the final rankings - after all, I was still using Hal Mumme’s first place vote for Hawai’i - but I was curious to see how the mechanics played out. Would there be some massive tie for sixth place that would be problemmatic? And how many schools received top ten votes?

The tally was simple to organize. The schools were ranked on the basis of the number of top ten votes each received. The votes for the top five were only used to break a tie between schools which placed in the top five in the balloting, in order to make it more difficult for coaches like Mumme to game a tiebreaker for Hawai’i (i.e., a school unlikely to finish in the top five in the voting).

...

Anyway, here are the final results based on approval voting (top five ballot votes in parentheses, in case of tie):

T1. Ohio State (59)

T1. LSU (59)

3. Oklahoma (51)

4. Georgia (40)

5. Virginia Tech

6. Southern California

7. Missouri

8. Kansas

9. West Virginia

10. Hawai’i

11. Florida

12. Arizona State

13. Illinois

T14. Boston College

T14. Boise State

As you can see, the final order didn’t change much. LSU moved into a first place tie, Florida changed places with Arizona State and Boise State rose eight slots.

One thing that DIDN'T happen in the results: a lot of ties. I had actaully expected to see a significant number of ties, but that's the genius of mechanical design, I suppose. (Maybe those Nobel Prize folks really know what they're talking about.)

It shouldn't be a surprise that a mechanical design approach produced much the same results as the ranked system when 2007 is reviewed.

After all, when the Coaches Poll ballots were compiled, the "game the system" method was still in effect. The seven voting SEC coaches all had LSU and Georgia in their top five while the seven Big Ten coaches all had Ohio State at #1.

If the Coaches could go to a mechanical design, however, I'm inclined to think that irregularities would be diminished. Again, a vote for Ohio State as an elite team would be equally significant on the ballot as a vote for LSU. Removing the ability to significantly game the system would add much-needed credibility to the coaches poll.

Then again... assigning credibility in the BCS system has never been a priority, so maybe all this mechanical design is just hooey.

But, then again, it's clear that we aren't going to have a playoff in college football until AT LEAST 2015, so maybe we should start working to improve the BCS now.

Posted by Kevin Donahue

Share|