Tall, Dark, and Mysterious


Electoral Reform, or, In Which the Author Extrapolates Wildly From an Extremely Small and Biased Sample

File under: Character Writ Large, Queen of Sciences, Welcome To The Occupation. Posted by Moebius Stripper at 9:24 pm.

On the agenda at work today: coordinate a meeting that as many as possible of two dozen-odd clients would be able to attend. Inane administrative duty? No: opportunity for field research into alternative voting systems!

Dear clients, I wrote, I’d like to hold a meeting with you during the first week of March. Could you please email me a list of times when you will be available during that week (eg, “Monday morning”, “Thursday afternoon”, etc)? If you have some times that are better than others, feel free to send that information as well (eg, “I am free Monday and Tuesday mornings, but the best times are Tuesday, Wednesday, and Friday afternoons”). I’m going to do my best to accommodate as many of you as possible, but I can’t guarantee anything.

And I got responses, some of which provided me with nice, STV-compliant rankings. But, oh, the others - for instance, the dozen that went something like this:

dear moebius stripper

i am free tuesday mornings

Thank you for your quick response, I wrote back, But you’re a retired grandmother and like hell that’s the only time slot you have free Are there any other times you have available? I am trying to accommodate as many people as possible, and while I can’t guarantee that everyone will get their first choice, I will have a better chance of arranging a meeting that most people can attend if I have more options.

dear moebius stripper

yes there are other times i could come but tuesday mornings are the best, i do not want the meeting any other time.

Fully half of my clients did not want to provide me with their second choice because they were afraid that doing so would weaken their first choice vote. I’m wondering now how widespread this attitude is - does it account for a significant proportion of votes against electoral reform? If so, it’s both selfish and irrational, because we’re probably going to end up with a meeting date that hardly accommodates anyone.

Kind of like the government we’re going to end up with on Monday.


  1. See the University of California, Davis, voting “system” for their Student Government Senate. Voters had to rank the candidates from 1 - 13. The software ranked candidates based on number of #1 place votes. The lowest-scoring candidate was eliminated from the race.

    Even if that candidate had the most #2 place votes.

    Crazy!! Software was evidently developed by some of UC Davis’ own computer department students. Too bad they didn’t consult with sociologists, or statisticians, or . . . well, anyone!

    - Jeri — 1/21/2006 @ 8:40 am

  2. That system sounds like a variant of Instant-Runoff Voting, which is equivalent to the system that British Columbia voted on last summer. And while that system is not without issues, it’s still much better than the most-commonly-used first-past-the-post system, in which (essentially) all but the highest-scoring candidate is eliminated, and what’s this about second choices?

    Unfortunately, bo voting system is perfect, and if Davis consulted a voting theorist they’d probably have been told that. However, the Borda Count factors in second-place votes on the first pass, and so might have been more suitable for your senate.

    - Moebius Stripper — 1/21/2006 @ 8:58 am

  3. A few friends of mine are vocal opponents of IRV. The biggest flaw of IRV (which is not present in first-past-the-post) is that it’s not monotonic — ranking your preferred candidate higher could actually cause them to lose.

    - Nikita — 1/21/2006 @ 10:19 am

  4. If a preference ranking is really what you wanted, I think you might have gotten better compliance if you asked people specifically to rank their top n preferred meeting times instead of merely asking them the open-ended question of when they are available. The Tuesday-morning guy might have put “1. Tuesday morning, 2. Tuesday morning, 3. Tuesday morning…” and others might only have three or four times they can attend at all, but I’d bet you could have gotten almost everybody’s top three or so.

    - Wacky Hermit — 1/21/2006 @ 11:54 am

  5. This is marginally related.

    While reading up on the proposed mixed-member proportional representation scheme for Quebec, I’ve been growing disenchanted with STV as a scheme to elect representatives to a parliament. But it seems like a good approach for the problem of picking just one outcome from a list, as opposed to a set of outcomes.

    However, I was astonished to learn that allowing a second vote under MMPR can introduce new paradoxes. See page 11 of In Search of a Compensatory Mixed Electoral System for Québec, from le Secrétariat à la réforme des institutions démocratiques et à l’accès à l’information. Crazy stuff.

    Not that this excuses the grannies you’re dealing with, of course :)

    - Matt Corks — 1/21/2006 @ 1:09 pm

  6. They’re actually considering some system in Quebec that doesn’t hugely favour rural areas and disfavour Montreal and, to a lesser extent, other cities? No way.

    - wolfangel — 1/21/2006 @ 7:36 pm

  7. Well, I’d just throw out granny’s vote, or just let that stand as first choice and say she squandered her 2nd, 3rd, 4th, etc. choices. Since she decided not to let other preferences be known, that’s her own problem.

    Also, Wacky Hermit’s idea will help you in the future, when you have to do this again. See, you made the erroneous assumption that no one you deal with will be a selfish bastard.

    - meep — 1/21/2006 @ 10:27 pm

  8. Nikita: I’m fairly opposed to single-winner IRV as well. Does your opposition extend to STV, which has basically the same theoretical properties? In my case, I think that pragmatism makes STV the best choice for a multiple-winner method.

    Basically, the only computationally feasible multiple-winner methods are extensions of plurality or IRV. So if you want something monotonic and reasonable to implement, you don’t even get to use ranked votes. And STV is at least pretty good at being proportional, and at ensuring that a lower-ranked choice doesn’t hurt a higher-ranked one (which could certainly help Moebius Stripper try to persuade Granny).

    - Rob — 1/22/2006 @ 1:08 pm

  9. Rob and Nikita, thanks for your comments. I have issues with IRV/STV as well, and during the BC referendum I was curious as to why the pro-reform groups had chosen this method to promote. (STV also violates Independence of Irrelevant Alernatives, if I recall correctly.) I still felt (and still feel) that it was better than FPTP, but it’s not the one I’d have picked. I do like IRV’s immunity to strategic nomination, but I nevertheless would have proposed implementing Borda (which is monotonic) rather than STV, if it were up to me. I’ll probably Borda-ify the rankings I received, actually.

    How much do you want to bet that the people who selected only one time slot are going to send me emails complaining that I picked a time they weren’t available?

    - Moebius Stripper — 1/23/2006 @ 11:12 am

  10. However you settle on a meeting time, be sure to include the analysis in the announcement of the meeting time. Because you know that Granny et al. are going to call you up and scream about how they asked for *Tuesday*morning*!*, and now they are going to have to *plan*ahead*, and did you even pay attention to our preferences?!? I mean, I guess you could wait until they splutter about it before you share the analysis with them, but if it is in the announcement itself, you can reply with “yes, all that was in the email. Asshole.”

    - Tarid — 1/23/2006 @ 11:26 am

  11. Ask each client to provide in order, three meeting times. Assign each client one vote but divide that vote among the three choices a decimal value so that value adds to one (vote). For example, first choice 0.5 vote, second 0.3, third 0.2. Granny only wanted one time so she only gets 0.5 of a vote. Three votes for the same three times only counts once. Explain this system to your clients as it should be simple enough for them to understand .

    The slots with the highest fractional vote counts can be narrowed quickly and, if desired, further analysed for the most voters selecting the time.

    - William — 1/24/2006 @ 1:42 am

  12. I’ve found it useful to ask for a list of times when people have prior commitments that ABSOLUTELY CANNOT be changed. I still occassionally get “all times except Tuesday morning”, but much more rarely.

    - JeffE — 1/25/2006 @ 4:01 pm

  13. re: meeting times.

    The thing you described is EXACTLY WHY I have “retired” myself as a campus committee chair and a church-group committee chair this year. The leadership part I actually like, the being-responsible-even-if-someone-else-does-something-horribly-wrong is okay, but the scheduling meeting times was making my blood pressure go up.

    I had someone, who teaches a less-strenuous schedule than I, inform me that he “didn’t do Friday meetings.” No reason given - no, “I’m a divorced dad and Friday is my day to get the kids” no “I have regular physical therapy on Fridays.” Other people absolutely refused morning meetings. And so on.

    I suppose I could have just said “We will meet at time X” (where X = most convenient for me) but if I did, and didn’t have a “quorum,” I’d have to reschedule, which would be an even bigger headache.

    - ricki — 1/30/2006 @ 7:41 am

RSS feed for comments on this post.

Sorry, the comment form is closed at this time.