Dear Greg…

With all due respect Greg, I had no idea you’d gotten experimental surgery to have your balls removed.

That line is from Talladega Nights and in no way do I mean anything bad about Greg in it, so don’t take it as that. I like Greg and have a great deal of respect for what he’s done for College Swimming. I also love the idea of a dual meet poll to give us something to talk about beyond national championships. This is especially important in D3, in a year where both championships (usually it’s just the men) are decided (as far as teams go) before a stroke is, well, umm, stroked?

In this article, Greg defends his polls and what they stand for. Let’s look at these:

1. “they’re a source of discussion” – While you do point out, truthfully, they’re not the BCS as they don’t actually decide anything, I believe they’d create just as much discussion if improved. The discussion would, of course, be different. It would be a discussion about small errors and matchups rather than why a team which received no votes could beat one ranked in the top 20. It would be a discussion based on the rankings themselves rather than the process in general. Teams then could also be more proud of what they represent and where they end up being ranked, rather than seeing their number and only hearing how they were overranked.

2. “we’re not out to cure cancer here” – True, but I don’t think even those who damn the polls the most would put cancer and swimming polls in the same category. Let’s not compare apples to handguns. Also, I don’t think we’re calling for a complete overhaul of the poll process, but simply an improvement. Let’s call them tweaks. Even Bobby Flay tweaks his recipes from time to time.

Here are some tweaks I’d like to see:

– Rather than giving coaches just a top times list, we should give them more. There used to be a cheat sheet with a few top swimmers and their dual meet scores. That was likely hard to continually update, but maybe in conjunction with coaches, SIDs and fans/alumni, this could be done once again.

– After the first poll, someone could easily tell voters what has happened to any team who got votes. Did they lose to Amherst? Beat U of Chicago? Fight closely with an unranked opponent?

– I don’t really know how many coaches vote, but some of them are clearly clueless. Can we encourage those coaches to abstain? It might be a tough conversation, but “hey buddy, you voted for Kenyon’s women over Emory’s women… next time, don’t vote, just focus on your team”. I doubt we can really do this, but seriously. Maybe the way to do it would be to publish everyone’s ballot. This way coaches are responsible for their votes they cast.

3) “What we’re trying to do is strengthen (and expand) our sport” – has done a better job of this than anyone. I say this honestly. I don’t know as though a decent but not outstanding poll strengthens swimming, however. If the idea is to get more names out, the poll isn’t the format. The poll features those who are the best, not those who are improving or doing a good job. If someone is getting a vote because “while they aren’t really a top 25 team, but they are making strides”, then that’s a negative for someone who did deserve the vote.

I voted in this poll when it began. I supported the idea and still do. I also understand how hard it is to pick a full list of 25 teams. I don’t envy those coaches who are already very busy during the year. That being said, with some small tweaks such as, accountability and information, this poll could go from upsetting a lot of people who have no power to change it, to being something teams really take pride in.

8 thoughts on “Dear Greg…”

  1. I agree that the polls aren’t that accurate, especially the futher down the list you go. My recommendation would be that instead of opening the polls to all coaches, or website visitors, etc, each year a small committee of 6-10 people vote. There are too many teams and results to expect a large number of voters to have comprehensive knowledge of, but if you had a small number of interested volunteers it could improve.

    If sparking discussion is the goal, even a poll put togehter by ONE voter could do the job. We get Josh’s, or Cheadle’s top 25 on the d3page, and then we’re free to chime in with our opinions. That single voter can view all the input and adjust accordingly. It could be a bit of a wikipedia model: one main editor with a bunch of volunteers contributing research and ideas.

    Also, I wanted to say how much I am enjoying the reboot of this website. Thanks to the contributors for the time and effort put into some interesting original content and links!!

  2. Someone must have some programming knowledge that would let fans/alumni/coaches keep an updated “best meet lineup” on file somewhere, and then just have an algorithm run every team vs every other one, and then from that figure out who should be ranked number 1.

    The advantage to something like that would be a fair representation of how that exact team lineup would work in a dual meet. But obviously what it doesn’t do is allow for coaches to switch swimmers around from meet to meet, which would better match certain teams to play to their strengths in the meet, and change the outcome.

    And then there could be some ability to move around the poll if high ranked teams lose to low ranked/unranked teams, or vice versa. Like an “expected” win value. I don’t know, but there must be a programmer out there who could put that together. That’d be awesome.

  3. Funny you say that mxskeir… I finally got around to seeing The Social Network last night and the swim nerd in me wondered if you could use the same algorithm they applied to matching up girls to teams swimming against one another. I do think it would create a lot of problems for teams like Kenyon, who swims very few d3 teams straight up. There also probably aren’t enough meets to make it work.

    • A) Awesome movie!
      B) It would “hurt” some teams, but what “we” could do would be make it like 90% times based, and 10% actual matchup based. So when Kenyon beats Denison, then they’ll be ahead of them.

      To use the Williams/Amherst meet strictly as a recent big-level competition, if the two teams make their “best overall lineup,” not the team specific one, then perhaps Williams beats Amherst in the “times” catagory by a little, but Amherst actually won the meet, so that would help pull Amherst’s rank up. Amherst could still be behind Williams because Williams could do better than every other team below them, whereas Amherst might not be able to, but beating Williams was a good way to pull them up.

      I don’t know if anything like this is possible, but I’m sure a mathematician out there could give a crack at it, and programmer could put something on a program, and people from each conference could go through the lineups and determine, by say USA Power-Points rankings, or how close to the B or A cut they are, the best possible lineups, and people from each conference (or multiple) could keep a list updated.

  4. I’m a d3 swimmer, and Math major – this problem is a very interesting one. Any algorithm that depends only on dual meet results (Ie: which teams have beat other teams) to rank wouldn’t work, because there are probably teams that really shouldn’t be at the top of the rankings, but have gone undefeated, simply because they don’t compete against very good teams.

    A system that could generate fake dual meets, and score them, and then use those when real meets aren’t present might present a very interesting ranking. Also, it could “predict” winners and scores for future meets (maybe users enter a matchup, and can see the results as predicted by the program) would further encourage some interesting discussion.

    If volunteers could be found to manage A lineups for each team that has a chance of being in the top 25, then such a poll wouldn’t require anything too new, just a considerable amount of programming.

  5. The programming/math wouldn’t be that hard? I figured it the other way around, haha!

    I think the only fair way to do it would be to have each teams “best” lineup. What that means is not what the team would put up against an individual team, but rather what would be best overall. And I think the only way to do that is to use a % of an A-cut from dual meet swims only. No invitationals, and this is key!

    And by splitting it up where the majority of the ranking is based on the times, and only a bit is ranked on the actual head-to-head results, based on expected results, then the teams can move.

    So to explain, for example, lets choose the Williams/Amherst meet, because I know the most about the NESCAC. (Disclaimer: just using this as an example, not picking sides…)

    Let’s say for arguments sake WIlliams has a higher ranking “time” portion than Amherst, and MIT beats both of them on “times.”

    Perhaps Williams truly is the “better” dual meet team, but lost the head-to-head because the matchup of amherst against williams played to amherst’s strengths. And Williams would perhaps beat more high quality teams than Amherst because of this. So this way, Williams would be ranked above Amherst, but not by much.

    However, after beating MIT, Amherst clearly is a great dual meet team, and because of that, they would move ahead of Williams, but maybe not MIT because MIT could possibly beat Williams.

    If you are following this logic, I commend you. Any idea how something like this could be put into place? I’m sure we could have volunteers on the forums to keep an eye on each conference, or the teams who usually are the top 50 or so, and keep that running list.


Leave a Reply