Resisting the pull of cynicism since 1969.

Monday, October 15, 2007

If IRV is the answer, you've lost sight of the question

During the Ontario electoral reform referendum, a number of people--nearly all of them Liberals--kept popping up to say that they didn't like the system the citizens' assembly proposed, but that they really liked a different reform called Instant Runoff Voting (or IRV) instead. And since the rejection of Ontario's proposed MMP reform, these voices have only gotten louder.

What is IRV? Although originally created for single-winner elections like those that elect mayors or presidents, this is the system Australia currently uses to elect its House of Representatives. It is also known as Alternative Vote, or AV. In this system, voters receive a ballot on which they have to rank-order their preferences. If no candidate receives a majority of first choices, the candidate with the fewest number of votes is eliminated, and ballots cast for that candidate are redistributed to the continuing candidates according to the voters' indicated preferences. The result is a system which, when it is used to produce legislative bodies, tends to reduce the full spectrum of political choices to two large parties, like you tend to find in Australia (or the U.S.).

But if you think IRV is the answer, you've lost sight of the question.

Let's back up for a moment and remember what the problems are in Canada that electoral reform is supposed to fix. Our current voting system of first-past-the-post is designed for places that have a clear two-party reality--and it reinforces that reality by preventing alternative choices from gaining a toehold. But over the last few decades, Canada has managed to defy those odds and develop various alternatives to the Big Two. This means that what we have right now is a multiparty reality held hostage by a voting system designed for a two-party reality.

Now, if this were a simple matter of unfairness toward small and midsize parties like the Greens and the NDP, the issue might not gain much traction. But the thing is, this disconnect between our political reality and our voting system leads to weird artifacts such as "majority" governments that get as little as 37% of the vote (and therefore only represent 37% of the voters), and which sometimes even achieve this dubious status without winning the most votes. This doesn't happen in a two-party reality, but because we no longer have a two-party reality and haven't for a good long time, the voting system has stopped functioning normally. This affects not just small and midsize parties, but everyone. And if we want to avoid those weird artifacts, we need a new system that is built for our new multiparty reality and truly accounts for the full spectrum of choice available in that reality.

Any system based on proportional representation (i.e., systems that make sure every vote ends up counting toward electing someone) would address these problems. IRV, on the other hand, because it is yet another system that's designed for an essentially two-party reality (although perhaps one with a small handful of tiny parties), would address the problem of...I'm not sure what, honestly. The Conservatives being in power sometimes, maybe? It certainly would solve that little "problem" either federally or in Ontario--because the Liberals are the second choice of a lot of Conservatives as well as a lot of New Democrats in both places, they would be the clear beneficiaries of an IRV system. And as a system that severely punishes any parties but big, established ones, it might even solve the "problem" of the existence of small and midsize parties, too.

But here's the fly in the ointment: while that scenario might be Jason Cherniak's wet dream, it's not exactly one that would accurately represent the choices of most Canadians. There's a word for that: undemocratic. Not to mention the fact that replacing an outdated voting system that's making problems for us with a newfangled voting system that makes those problems even worse is a pretty ridiculous idea.

If you really like the idea of a ranked ballot, though, you're in luck--there's a way of achieving that while still keeping the element of proportional representation that we need to fix the disconnect between our multiparty reality and our two-party voting system. The Single Transferable Vote system, or STV, came within three points of meeting the 60% threshold necessary for adoption in B.C. in 2005. And because it came so close, B.C. is getting a second kick at the STV can in conjunction with the next provincial election in May 2009, with assurances that there will be better financing for education this time around. (If this idea appeals to you, you can get involved with B.C.'s push to change the system over here.)

10 comments:

BROKEN LADDER said...

IRV was born when someone decided to use STV (single transferable vote) in a single-winner election. So you have your history a little backwards.

In any case, Reweighted Range Voting and Asset Voting are the state of the art in proportional representation, and were invented in the modern era by a Princeton math Ph.D., whereas the other systems were mostly created decades ago, prior to most of the greatest advances in election method theory. Check out
http://rangevoting.org/PropRep.html

CLAY SHENTRUP
SAN FRANCISCO, CA
clay@electopia.org
415.240.1973

Idealistic Pragmatist said...

broken ladder,

IRV was born when someone decided to use STV (single transferable vote) in a single-winner election.

Perhaps, but as I understand it, the single-winner election version was then adapted yet again in Australia to elect a legislative body. That's always seemed a bit silly to me--why not go straight to the existing system of STV rather than trying to shove a single-winner system into the task of electing the House of Representatives? I'd love to know how that came about at the time (any suggestions on reading material on that, peanut gallery?).

Thanks for the link, by the way--I'll check it out. I'm geeky enough that I love reading stuff like that. I should mention, though, that I'm a little bit small-c conservative when it comes to actual reform--I prefer systems that we've been able to observe in action in other jurisdictions over the ones that have never been tested anywhere (no matter how elegant they sound in theory).

Ian King said...

IRV, on the other hand, because it is yet another system that's designed for a two-party reality, would address the problem of...I'm not sure what, honestly."

Paul Forseth, perchance?

I am not so sure that IRV was designed mainly for 2-party climates (why would those environments need a runoff?), but it's not necessarily more of the same. Instant runoff makes a difference in those three-party and three-party-plus races where a candidate whose views are not shared by even a plurality of voters ends up winning thanks to vote-splitting. AIUI, it's not there to create proportionality or "count every vote". It's not much better in those respects than FPTP. It does ensure that the representative is acceptable to the majority in the riding.

Amusing side note: the 1952 BC election used instant runoff, but it was billed as "Single Transferable Vote"!

(For those who'd rather not Google, Forseth was a forgettable Reform/Alliance/Conservative MP who repeatedly won his New West seat in squeakers thanks to the left-of-right vote not coalescing around either the Liberals or NDP. Forseth won over 35% of the vote only once, in 2000, the year the Alliance almost maxed itself out in BC.)

Idealistic Pragmatist said...

Paul Forseth, perchance?

Yes, yes, or to bring it closer to (my) home, Rahim Jaffer. I'd still rather actually have fair elections, myself.

AIUI, it's not there to create proportionality or "count every vote". It's not much better in those respects than FPTP. It does ensure that the representative is acceptable to the majority in the riding.

It's actually WORSE in those respects than FPTP, believe it or not, or at least worse than the way our multiparty reality has managed to contort itself in order to adapt to FPTP. And as for "acceptable," that's a huge assumption you're making. What if a whole slew of voters only find one candidate acceptable, or only the two candidates from the two smallest parties? (There are a lot of people like that these days.) Not only would those people never elect anybody in their own ridings--they'd end up losing any chance that people they like might get elected in some other riding, too.

My real point, though, is that IRV wouldn't accurately represent Canadians' preferences. Since the entire point of an election is to figure out what people want, it wouldn't end up being very democratic in a Canadian context.

Anonymous said...

Runoff voting is slightly superior to FPTP in that it arranges for every elected representative to nominally command at least a bare majority of votes. It is inferior in that not all voters are simultaneously given the opportunity to revote in light of the "new information" (the removal of a candidate).

There is nothing undemocratic about not every vote counting, since most issues in democratic systems will never count every vote: referenda and decisions in the legislative bodies, for example. And a change from regional direct election of representatives to one of election by party affiliation would necessitate reconsideration of constitutional law to sufficiently constrain the resulting flavour of "people's" government from abuses.

Idealistic Pragmatist said...

anon,

There is nothing undemocratic about not every vote counting

We can argue about that point, and I suspect we'd come down on different sides of the issue. The real issue in a Canadian context, though, is that most votes don't count under the current system. Reformers are looking for a fix for that, and IRV would not cut the mustard.

And a change from regional direct election of representatives to one of election by party affiliation would necessitate reconsideration of constitutional law

Not true, first of all--but more importantly, beside the point. Name me one reform proposal currently on the table that includes "election by party affiliation"? I suspect you won't find any. And there's no sense confusing the argument with something that's not currently on the table, is there? Not to mention the fact that this post is about IRV and why it wouldn't be better than FPTP, so even if it were on the table, this comment would be rather off-topic on this particular post.

Anonymous said...

>Not true, first of all--but more importantly, beside the point.

Beside the point, yes, with respect to this particular topic. But applicable in general to any discussion of changing electoral systems. Reviewing the legitimate use of legislative power is common sense when reviewing the methods of determining how that power is assigned.

Anyways, IRV is on balance worse than FPTP because of the information imbalance. The solution to that is the system used at party leadership conventions.

Anonymous said...

I am not sure if this is relevant. But when you mentioned that IRV is used in Australia and is now a suggestion by the Liberals, I can't but help seeing the same pattern I so often see. We only trust things that exist in other English-speaking places. We never go further than that, because we don't understand it (language-wise) and would have to do too much research and translation. Plus, anything in a country that speaks no English (or French) is probably 3rd world and waaay behind us.

Idealistic Pragmatist said...

final anon (can't you guys use pseudonyms, here?),

You have a point about the way Canadians react to things done in jurisdictions we're not immediately familiar with--this is something that has been a grave disappointment to me about this country. But because MMP is used in New Zealand and STV is used in Ireland, electoral reformers have the non-English-speaking problem covered, I think. :-)

Anonymous said...

How on earth can yu argue that IRV is for a two party system? That's about the dumbest thing I;ve ever heard. In a two candidate it would be mathematically impossible to be of any relevance.