Voting, Livejournal style - The Snow In The Summer or So-So

26 April 2008
The Cruddas Effect

This post is superseded, at least in part, by developments we came to know a month later.

Even after going through this one a few times, we still can't see the error in our thinking. Someone, surely, will be able to spot the blindingly obvious goof that we've made. The topic of discussion is the voting system for Livejournal's consultative board. Leave out criticism of the organisation, this is purely a post about the electoral mechanics. The process is this:

  1. Voters will be able to select their first, second, and third choice.
  2. If a candidate has 51% of the first-choice vote, that candidate will be declared the winner
  3. If no candidate has 51% of the first-choice vote, we will begin using an Instant Runoff process. The candidate with the lowest tally for first-choice vote will be eliminated, and the second choice candidate on those ballots will be substituted as first choice. This process will continue (drop lowest candidate, reassign to next vote, tally again) until any candidate has 51% of the vote.
  4. If all ballots are exhausted and no candidate has 51% of the vote, we will return to the first round and award the election to the candidate with the most 1st choice votes.

(Amended from original to remove irrelevant section, and clarify the difference between voters and candidates.)

Now, what in this arrangement causes us problems? As any half-awake psephologist will spot, it's a truncated AV system, considering only the voter's first three preferences. Readers can compare and contrast against the London mayor, which considers only the first two preferences, and bulk-eliminates all but the top two candidates.

It is not clear what happens to the non-effective non-transferable votes: points 3 and 4 refer to 51% of the vote without declaring whether this is 51% of the active vote (those papers with a preference currently being counted) or 51% of the original vote. Does the Quota reduce in line with non-effective non-transferable votes? Logic suggests that the Quota should fall, it does in London, it does in Ireland and Australia (where fully-specified votes are the norm), but it's not clear. Indeed, the final clause suggests that the Quota remains unchanged, as 51% of the valid vote, with non-effective non-transferable votes still counted.

Another unusual arrangement in this election is that the Quota is set at 51% of the valid vote. It would usually be set to 50% of the valid vote, rounded off to a nearby integer to ensure that only one candidate could reach it. How might this alter the outcome? Suppose that the election is contested by three candidates:

A - 5040
B - 2500
C - 2460

Quota - 5100

Clearly, B and C combined cannot catch A, but the Quota for victory is 51% of 10,000, so A has not yet won. Indeed, if all of C's votes transfer to B, apart from 100 non-effective non-transferrable papers, we have this final result:

A - 5040
B - 4860

Quota - 5049 (or 5100)

No winner has been found.

Under any rational system we've ever seen, the result would be to elect A short of a Quota. But not here: the process rolls right back to the initial count, and the candidate securing the most first preferences wins.

Is that a minor matter, of academic interest only, and so obscure as not to happen in real life? Er, no. Let us take you back to the Labour party's deputy leadership election last June, when the result was tighter than a gnat's crotchet. A six-candidate election where non-effective non-transferrable votes are discarded boiled down to the final two, when Harriet Harman beat Alan Johnson by 50.43% to 49.56%.

Under the rules proposed by Livejournal, the deputy leader would not be Mrs. Harman, who secured the approval of the majority of voters who expressed a preference, but not 51% of active voters. It would not be Mr. Johnson, the runner-up. No, the deputy leader would be Jon Cruddas, who came on top in the very first round, but was subsequently eliminated in third place.

And that, viewers, is what we don't understand at all. If we're right - and we're convinced that this isn't what they've intended at all - the combination of Russians and St Francisvilleains, like Mr. Brown, may be lumbered with a result they really don't want. Or a result they do want, but that goes against the democratically expressed wishes of the voters.

We're convinced that this an oversight. Concurrent with the publication of this post, we've asked Livejournal to comment, and will publish such substantive response as we receive.

Addendum, 3 May

Though we've received no communication from Livejournal, and the company has failed to clarify its post, we reckon we've cracked the idea. Rather than declaring the winner from the last two candidates, the loser's votes are then sifted for sufficient transfers to give the winner 51%. We reckon that this is an attempt to ensure that the winner has the approval of a supermajority of voters (if 51% counts as a supermajority). Quite why the fallback is to return to the first preference count is beyond us.

This post is superseded, at least in part, by developments we came to know a month later.

| Permanent link