Mumme Poll

“I always tell people that if they don’t like the way I vote, they can rank teams how they please. Thanks to you, they can.”Sports Illustrated’s Andy Staples

INTRODUCTION

The Mumme Poll is an ongoing experiment to construct a viable method to rank the best Division-1/FBS college football teams, absent the bias and potential conflicts of interest that affect other polls, in particular the USA Today Coaches Poll.  The poll is jointly sponsored by two college football blogs, Get The Picture and 3rd Saturday in Blogtober.

We attempt to accomplish this by conducting the voting in two very different ways from other football polls:

  1. The first vote does not take place until after the games of Week Six have been played.  Voters are not asked to evaluate teams based on preseason expectations and are not expected to use those as a baseline from which to rank teams for the rest of the year.
  2. Rather than being required to rank twenty five D-1 teams in order of preference, Mumme Poll voters submit ballots which consist of their top ten teams in the country, without ranking (other than to designate a single best from those ten, for use as a tiebreaker).  The poll rankings are then compiled by means of approval voting; that is, the teams are ranked in the order of the total number of times they appear on voters’ ballots.  (For more on approval voting as a means of addressing bias in the Coaches’ Poll, see this post.)

BACKGROUND/HISTORY

The inspiration for the Mumme Poll came from the final 2007 regular season Coaches Poll and Tony Barnhart’s post breaking down some of the more curious ballots cast.  (You can read that post here.)  The poll is named in honor of perhaps the most questionable vote of that Coaches Poll – Hal Mumme’s ballot listing Hawaii as the number one team in the country.

In response, the initial Mumme Poll ran at Get The Picture for the 2008 season.  (Florida finished first at season’s end.)  With the 2009 season, the poll moved to a new home at this site.  The move allowed for a web interface that made the voting process much more user friendly.  For 2010, the site was tweaked in order to allow for greater feedback from voters.

PARTICIPATION

We’re (small “d”) democrats here.  Mumme Poll voting is open to any knowledgeable, enthusiastic college football fan.  That’s part of the fun of it.

RULES

There aren’t many, but what few there are matter.

  1. Registration. To cast a ballot, you’ll need to register at the Mumme Poll site.  Registration each season is limited to the week leading up to and through the games of week five.
  2. Voting period. In most weeks, balloting takes place at any time between 9:00 A.M. on the Sunday following the games and 9:00 P.M. on the following Monday.  On those rare weeks when a Sunday game is played, the voting is pushed back 24 hours.
  3. Ballot format. A ballot must contain votes for ten schools and a top team selected from those schools.
  4. The one commandment. Don’t try to game the system. Don’t subvert the vote by submitting a ballot with your favorite school and the nine worst teams in D-1, for example.  The ballots are monitored and if something questionable comes up, a voter will be given a chance to explain.  If we’re convinced there’s a deliberate effort to muck things up, we’ll toss the ballot and the voter.
About these ads

55 responses to “Mumme Poll

  1. Unbiased?

    Wasn’t sure if you read this on you last post.

    Just one question/suggestion…

    Have you thought of asking others on the blogroll to participate in a similar Mumme Poll with their own readers?

    I think even fans from the Big 12, PAC 10, etc., wouldn’t mind putting in their 12* cents on Sunday towards such a foward thinking idea.

    It might help to have a broader scale and some relativity in this little experiment.

  2. I did see it, and it’s a good suggestion.

    I’m not sure how many others want to go to the trouble of managing a poll. If you watch what Brian Cook goes through in running the BlogPoll, it’s a little intimidating. Although I do think the Mumme Poll will be a good bit easier to manage due to its structure.

    In one sense I agree with your point about conducting the voting on a broader scale. The fact that the vast majority of folks voting here are Georgia fans is something I’ll want to monitor carefully. On the other hand, if the point of approval voting is to reduce the element of bias in the results, maybe it’s a good thing to see how the voting plays out with that sort of a group.

    It’s definitely food for thought. Thanks.

  3. Munson's_call

    Is it Week 6 yet? I am ready to go!

  4. dean

    I would love to participate in the Mumme Poll. However I have one potential conflict with voting after week 10. The wife is making a strong push to take the kids to Jacksonville and “swing” by Disney Land/World (which ever is in Orlando) the Sunday and Monday after the game. And you know what they say – Happy Wife = Happy Life. Anyway I generally don’t like taking laptops or blackberry’s on trips because work never fails to interfere somehow. So it’s very doubtful I will be able to submit a vote by the deadline that week. If this disqualifies me from participating that’s fine, I’ll understand and somehow move on.

  5. dean – we’ll figure something out. It could be that you could send a quick e-mail from the hotel you’d be staying at.

    BTW, Week 10 features a Sunday night game, so the balloting will be moved back one day as a result. Does that help?

  6. Unbiased?

    Another question.

    How will conference championships play into the polling?

    Will voting stop at the end of the regular season?

    How will that affect non-championship

  7. Unbiased?

    oops…non-championship game conferences (once again) if voting does continue?

    That’s one of the things I hate about the polls now, You are punished for losing every week, but a loss later in the season ends up hurting your team more in the end, even if your first (or second) loss is in a hard-fought championship game.

  8. Unbiased?

    oops.

    …non-championship game conferences?

  9. Will voting stop at the end of the regular season?

    I would think we’ll track the other polls in that regard. The last regular season polls come out after the conference championship games.

  10. Unbiased?

    sorry about those bogus posts, that shows how well I can operate my cell phone walking through an airport.

    Feel free to delete them Senator.

  11. I agree with your premise and I volunteer.

  12. picturemerollin

    I would love to participate in this poll

  13. picturemerollin

    Let me try this again – with contact info this time:
    I would love to participate in this poll
    http://picturemerollin.wordpress.com
    picmerollin(at)gmail(dot)com

  14. Pingback: Slow down, you move too fast. « Get The Picture

  15. Wolfman

    I know its been too long since the post, but I want in.

  16. Don’t worry, it’s not too late. The first ballot won’t be taken until after week six games. I’ll take participants up until then, unless the numbers get too unwieldy.

  17. BigDawgEatsAlot

    I’ll give it a go Bluto. Count me in. Anything to debunk the BCS propaganda by having another poll. :)

  18. I would be honored to have your participation, sir.

  19. a redcoat

    just bringing it up again in this line of comments, but i totally want in.

  20. Auditdawg

    Just sent in my poll.

  21. Pingback: Mumme Poll #1 « Thinking Bulldog

  22. Pingback: Mumme Poll; Week Six Ballot « Picture Me Rollin

  23. Pingback: Mumme Poll « Bulldoginexile’s Weblog

  24. Joel Rusche

    I’d like to vote, if possible. Your site is pretty cool.

  25. Joel, thanks for the offer and the praise. Unfortunately, I can’t add people at this point, but keep it mind for next season.

  26. Alex Stroud

    That’s hilarious. I knew Hal Mumme when he was the coach at Valdosta State University (I was in high school). Not too many brain cells accounted for on that guy.

  27. Pingback: Mumme Poll; Final Regular Season Ballot « Picture Me Rollin

  28. Hal

    Are you doing this again in 2009? Need a Pac Ten guy?

  29. I would love to participate in the Mumme Poll

  30. I would like to volunteer for the Mumme Poll.

    Seems like an interesting idea, I would like to see how it comes out compared to the other polls

  31. Nate Dawg

    I would love to participate in this poll. I’m an avid college football fan and from what I read about this poll or experiment, it seams very fair. I’m also a HUGE Dawg fan but I promise not to let that get in the way of my judgement. I also enjoy reading “Get the Picture” on a fairly regular basis. Great name for a blog!!

  32. Pingback: Ballot business, Week 1 | The War Eagle Reader

  33. ecdawg

    I would like to participate again this year.

  34. Check out whobeat.net, a win-based seeding process for D-1A football. All 120 teams start out at #1. The only data utilized for positioning is the “W”. Teams create the data/outcomes and thusly position themselves based on each team’s success against their respective schedules . This process is purely about the “W” and establishing values for head-to-head victories. The info at whobeat.net is the kind of info needed by anyone participating in a vote-based football poll such as the Mumme–I want to play too! OK?

  35. Macallanlover

    Senator, I participated in last year’s poll. What is the registration process for this year? How will the new “structure” change the way ballots are submitted, or timing? Thanks.

    (PS The way the game results have obliterated the top ranked teams so far this season, this method of waiting until after Game 6 is even further validated!)

  36. I’ll have a post up in the next day or so about registration. In my opinion, the voting process is greatly improved over last year.

    Timing won’t change.

  37. kidchocolate

    Found a link to this on David Hale’s blog (http://dawg-extra.blogspot.com/)

    Great idea! Would love to participate if possible.

  38. Pingback: A Hipster’s Worst Nightmare | Slow States

  39. I am in, how do we register?

  40. There appears to be no link to the actual Mumme Poll in the body of the main article.

    Could this be added?

  41. Pingback: Another season, another Mumme Poll | Get The Picture

  42. Pingback: The Mumme Poll returneth! | Get The Picture

  43. If you’re limiting the number of teams people can vote for (apparently to 10) then that’s not Approval Voting.

  44. Pingback: Friday morning buffet | Get The Picture

  45. Pingback: Missouri hoping to race past injured Georgia | Campus Union - SI.com

  46. Pingback: Earning Their Stripes : One Caribbean Radio | The Global Mix

  47. Clay Shentrup

    “Mumme Poll voters submit ballots which consist of their top ten teams in the country,”

    That is certainly not Approval Voting. The point of Approval Voting is that you can vote for any number of options. That’s what makes it so resistant to tactical voting.

    http://www.electology.org/systems-confused-approval-voting

    Clay Shentrup
    Co-founder, The Center for Election Science

    • If I understand you correctly, Clay, you’re saying that a true approval vote would allow every person casting a ballot to decide how many teams to populate it with, from one to one hundred twenty-four (or whatever the current number of D-1 teams is).

      Under that scenario, what’s to stop a group of voters from conspiring to cast ballots with a single vote of the same team on them all?

  48. Clay Shentrup

    what’s to stop a group of voters from conspiring to cast ballots with a single vote of the same team on them all?

    The fact that it’s strategically non-optimal.
    http://www.electology.org/threshold

    In a nutshell, the same reason lots of Green Party supporters vote Democrat (and would vote for both of them, if using Approval Voting).

    • I understand that in the context of a general election, with thousands or even millions of voters. But in the context of a Coaches Poll, with only sixty voters, it sure seems like it would be easy to game the results with such a strategy.

      • Clay Shentrup

        On the contrary, with a small group of voters, the difference between a sincere vote (mean-based thresholding) and a strategic vote generally becomes smaller. (Incidentally, this is also why the optimal Score Voting strategy in small groups is often not to vote in exaggerated Approval Voting style: http://scorevoting.net/RVstrat1.html)

        Think about it. Say you bullet vote for “Nader”, only to find that “Bush” squeaks by to beat “Gore”. Now you’ll be kicking yourself for not also approving of Gore. That “spoiler” problem doesn’t go away just because you have a small number of voters. Indeed, it can be worse.

        The irony is that by limiting the number of votes, you encourage the kind of strategic voting that says “vote for Gore even if you prefer Nader, so you don’t get Bush”. I.e. your favorite team is one you believe isn’t well liked by other voters anyway, and so you might as well use your limited votes on teams where you’re more likely to make a difference. The very strength of approval voting is that it gives you unlimited votes.

        At the very least, a rule change that says you must vote for at least N teams is an improvement. But even then, N (apparently 10, in your case) is some arbitrary number you chose with presumably no scientific basis for it.

        And speaking of science, one important criterion for a voting method to satisfy is “symmetry”. In other words, for any vote that can be cast, there’s an “equal but opposite” vote that can be cast to cancel that vote. The idea is that two voters with the exact opposite preferences should have equal weight, and thus cancel out. So if one voter votes for precisely 10 teams, then a voter with the exact opposite preferences would logically vote for every other team aside from those 10. But your system doesn’t allow that. This introduces noise or “loss” into the system.

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s