Sun Oct 8 20:19:20 EDT 2000
ICANN, Humpty Dumpty; Humpty Dumpty, ICANN
"When I use a voting method," Humpty Dumpty said in a rather a scornful tone,
"it means just what I choose it to mean -- neither more nor less."
Over the past months, ICANN has sent out numerous announcements to MAL participants (not "members"). Why, then, seven days into the eleven-day election -- or, accounting for time zone differences eight days for the majority of MALers -- did they finally send out a notice with a pointer to an explanation of the methodology by which the election results will be tabulated? Presumably, they were getting wind of the fact that MALers didn't understand it. Judging by the burst of messages distributed on Dave Farber's "Interesting People" list, MALers still don't understand it either.
ICANN's message to the "At-Large Members" list, dated Saturday, 7 October, (11:21:40 -0700), points at a sort of explanation of the Alternative Voting System (AVP) or Single-Transferable Vote (STV) method in the MAL FAQ:
Members will rank the candidates in order of preference ("1" for their
first choice, "2" for their second, etc.). Members may rank as many or
as few candidates as they choose. The votes will be tallied according to
the first preferences (the "1"s). If at that point one candidate has an
absolute majority (50% + 1) of the vote, he/she is selected. If not, the
candidate with the fewest votes is eliminated.
So far so good, right?
The eliminated candidate's votes are redistributed to the next ranked
candidates (the "2"s).
Uh...yeah... (Kudos to ICANN's staff for meticulously cutting and pasting this spotty description from the Election Committee's "Recommendations on Election Procedures.")
The problem is, which "2"s? There are 3 kinds of "2"s: (1) the candidate (or, if tied, candidates) who received the second-greatest number of #1 rankings in the first tabulation; (2) the remaining candidates (statistically, probably all of them) who received #2 ratings without regard to whether those rankings were eliminated in the first tabulation; and/or (3) the candidates who received #2 rankings from voters whose #1 rankings were eliminated in the first tabulation.
The ambiguity lies in the fact that, from a voter's perspective, AVP/STV is more accurately a differential ranking system than a zero-sum "voting" system. In effect, if your #1-ranked candidate gets thrown out in the first round of tabulation, your #2 ranking becomes a #1 (and #3 a #2, etc.) ranking and the results are retabulated, and so on -- until one candidate receives an "absolute majority" of adjusted #1 rankings. I think. As we have seen, ICANN's FAQ only muddies the waters; and neither of the "useful (and entertaining) background resources" the FAQ points to do either. One of them, "The Muppets Use Instant Runoff Voting" (IRV), presents a decent narrative of how an AVP/STV/IRV election unfolds (though a better explanation can be found here). But the Muppets aren't imposing an unaccountable globalized intellectual property regime on the net, now, are they? And that minor difference may prove to have a major consequence, as we shall see.
roving_readers of course now understand AVP/STV/IRV intuitively and so will have no problem making short work of this non-hypothetical eventuality: Let's say you hold four out of seven candidates in such abysmally low esteem that you refuse to rank them at all -- that is, you rank only three. Well, first off, you'll be fed a response form that looks like an error (though there are worse fates). But let's say you persevere nonetheless with your three rankings. Now, let's say, further, that none of the candidates you ranked achieves an "absolute majority" in the first tabulation. Or in the second tabulation, where your #2 becomes your #1. Or in the third tabulation, where #3 becomes #1. Or in the fourth tabulation, where...oops -- you didn't provide a #4 rank. Off to the bitbucket with you...right? Not necessarily. Your failure to rank them may not have been the protest vote you thought -- you just didn't provide an explicit basis for redistributing your rankings. And we have ways to compensate for your pathetic schemes...
Now, whether election.com will assign variables or tinker with statistics in cases where candidates weren't ranked isn't clear. Brian Reid mailed the election.com help desk to get a better understanding, but their answer was more confusing than ever. Not to fear, though: they'll publish a "white paper" -- after the election. Unless they're doing this out of the goodness of their heart, then ICANN proceeds apace with PR first and "technical coordination" second. In any event, the roving_reporter looks forward to hearing what ICANN's Election Monitoring and Oversight Panel has to say about this all.
But lest the roving_reporter seem too persnickety and negative, take a look at Safevote's newsletter, The Bell, whose current (September) issue -- available as a free sample -- discusses the importance of open-source software in electronic elections. Election.com, deemed by ICANN to be the "leading global Internet election company," uses proprietary software.