[EM] Re: majority rule criteria--alternative nomenclature

[James Green-Armytage]

>
>
>    Mike Ossipoff answered you pretty well in his "counting time"
>response. 

	Have you read my reply to his reply yet? That was posted January 14th. In
that message, I describe this counting method in more detail. Basically,
it works when you have a small-enough group to do quick pairwise
comparisons by a show of hands. If there are more than 30 or so people
voting, it will take more than a few minutes.

>In addition, a big problem I have with your answer is
>that I suspect it's totally theoretical. Have you ever actually tried
>to use a ranked ballot method in a meeting? 

	Yes, I've done demonstrations of the method I described in the 1/14 post.

>You talk about it
>taking only "a few minutes," but I think you are being wildly
>optimistic unless you are thinking about a very simple 3 or 4
>option decision by a very small group of people. 

	Yes, that is what I was talking about. The method of counting by show of
hands I describe can be used in a larger group (e.g. the US Congress), but
it will take longer. (For the US Congress, I would suggest that they rig
up their desks with an electronic interface, so they can tabulate any kind
of vote instantaneously.)

>But in real
>life, things are often much more complicated.

	Granted. I thought you were talking about small groups.
>
>   I'll cite just two examples of actual decisions I was
>involved with. One was at the 1996 founding convention
>of an organization tentatively named "The Alliance." There
>were over 300 people at the convention, and one decision
>they needed to make was to choose a permanent name
>from a dozen or so choices that had been winnowed down
>from a much larger list by a name committee. I had learned
>about Condorcet voting from Steve Eppley, but any kind
>of ranked ballot method would have been out of the question.
>The best method by far would have been Approval, but at
>the time neither I nor anyone else at the convention knew
>about it. I believe a two stage process was used, with the
>final decision made by plurality from among a smaller
>group of names that had been chosen by plurality in the
>first stage. Nevertheless, the actual vote and the counting
>of it took several hours. A one step approval vote from
>among the entire original list of names would have gone
>more quickly and probably would have been a more
>satisfactory method. I was happy with the name that
>was chosen (Alliance for Democracy), but like many
>other people at the convention, I had misgivings about
>the voting process used.

	Okay, fair enough. This seems like a pretty good situation to use
approval. One, it's too large a group to count quickly by show of hands.
Two, the stakes are relatively low, majority rule isn't crucial,
widespread tactical voting is unlikely, and consensus is the goal.
	Let me just clarify something, though. When you say "a one step approval
vote", do you mean the following procedure?: 
1. One person calls out a potential name. 
2. People who approve of that name raise their hands. 
3. Someone counts the number of votes for that name, and marks it down
somewhere.
4. This repeats for all possible names.
5. The name with the most approval votes is adopted.
	I can see how that would be quicker than the procedure I describe. Fewer
counts. Still, they are comparable. Plus, it might be possible to do a
combination. After doing the approval counts for all possible names, you
could do a pairwise contest (a runoff) between the top two approval vote
getters. (Or you could do my pairwise-count-by-hand procedure between the
top three or four approval vote getters.)

>   Another example was a board meeting of a national
>organization with a dozen or so people present. A
>decision had to be made about where to hold the next
>meeting. Approval voting was used, and it went well, with
>everyone very satisfied with the process. An unusual
>aspect of it was that instead of making a list of possible
>meeting locations and then voting on them, locations
>were suggested one at a time and approval votes were
>taken after each name was suggested. Again, ranked
>ballots would have made the process much more time
>consuming and probably less satisfactory. Ranking
>many different possible meeting locations also would
>have been difficult for most people, whereas it was
>fairly easy to decide whether to approve or disapprove
>each proposed location. So I think ranked ballot voting
>would have been overkill in that case and would have left
>people less satisfied with the process if not also less
>satisfied with the outcome.

	Okay, interesting. The number of people is small enough to do quick
pairwise counts by hand, but the number of options is unknown. Also, once
again, majority rule doesn't seem particularly important in this scenario,
severe tactical voting is unlikely, and consensus is the goal. So,
approval might make sense in that situation.
	However, if you were more concerned with majority rule, a couple
alternatives spring to my mind. One is to do the nominations beforehand,
and then to do the pairwise-by-show-of-hands  method I described on 1/14.
Another is to do the pairwise-by-show-of-hands method between the top
three or four approval vote getters.

Sincerely,
James Green-Armytage
fc.antioch.edu/~james_green-armytage/voting.htm