While the idea of primaries and supporters is being framed as a way of making the Liberal Party more open, it is in fact one of the most disempowering proposals heading to the convention. Shifting from a delegated convention to weighted one member one vote was a good change, but we should keep our leadership selection process limited to people who believe in the party, our values and our policy vision strongly enough to become members.

At next week's Liberal Party convention, delegates will be voting on a series of constitutional proposals. Most of the proposals, which Jeff Jedras has helpfully annotated, are not very significant in terms of the future electoral prospects of the party. I support most of them and they are worth implementing, but I think it is foolish to look for the seeds of electoral success in internal constitutional tweaking. We shouldn't get too excited about this particular convention.

The changes related to creating a category of "supporters" and electing our leader through a series of staggered regional voting days (aka primaries) have probably garnered the most attention. I'm opposed to both ideas. Jeff Jedras and Andrew Coyne have already articulated some good reasons. Please read both articles if you haven't already done so.

Weighted one member one vote is a good system (WOMOV). It is very likely to elect a leader who resonates well with Liberals throughout the country. I see no reason to move away from it before we try it out.

Data collection is a red herring
Adding supporters to our constitution has nothing whatsoever to do with identifying supporters. We do that already during elections (and, more rarely, in between elections). Although it is true that we would collect data during a registration drive for a primary, it's not the only way to do so. I am a big believer in data and I find the "we need data!" line of argument to be less than compelling. The acknowledged leaders in collecting data on voters, the Conservatives, have managed to do so not only without a primary system for electing their leader, but with no leadership election at all since 2004. It's been all Harper, all the time. They have a money advantage and they are using their money well to create an information advantage. Frankly, the successive elections that we caused during the minority parliaments have probably been the biggest boost to their data collections efforts and ours.

The proposal is not really about data, which we can and do obtain in other ways. It's really about broadening the franchise for leadership elections and candidate elections to include supporters who do not want to become members. The onus is on proponents of the primary proposal to convince delegates why we should do so.

Leadership selection
Given the choices of Stéphane Dion and Michael Ignatieff as leader and our results in the 2008 and 2011 elections, I can understand why people are focused on our leadership selection process (disclaimer: I like both Dion and Ignatieff). However, neither leader was selected by our existing selection process of WOMOV by preferential ballot. Our next leader will be. Whereas Stéphane Dion was elected with the support of 2,500 delegates, our next leader will probably need at least 45,000, depending on our membership numbers. Stephen Harper received more than 67,000 votes on the first ballot of the CPC leadership vote in 2004. That is a huge difference from 2,500.

Furthermore, no selection process is risk-free or a sure way of selecting an excellent leader. We should focus on the pros and cons, and the practical considerations, of the different methods.

What might a primary system mean in practice?
Warning: numbers and future babble ahead.

Some numbers may help to put the issue into perspective. In the last general election, all Liberal candidates combined received 2.7 million votes or 11.5% of the voting eligible population. In better times (2004), our candidates received 4.9 million votes or 22% of the voting eligible population (VEP). That gives a pretty good sense of what the upper limits on the number of supporters might be.

Turnout in US primaries varies hugely, from 2.6% of VEP in the Wyoming Democratic caucus to 32% of VEP for the Democrats in the Vermont primary (check the data out yourself). The Vermont numbers, which are the very best for any party during the 2008 primaries, are very similar to the 32.6% of VEP that Obama received during the 2008 presidential election (52.9% of the votes times 61.7% of VEP = 32.6% of VEP).

In Vermont, during the actual presidential election, Obama received 44% of the VEP (66% times 67%). So let's take that as an absolutely best case scenario: in regions where we are relatively strong, we might expect 75% of our voting day supporters to bother voting in a primary (32/44). In Wyoming, Obama received 21% of the VEP (32.5% of 64%) during the actual election and Democrats all together received 2.6% of VEP in the Wyoming caucus. So worst case scenario, in areas where we are relatively weak, we might expect to see 12.3% (2.6/21) of our voting day supporters turnout.

If we were to adopt a primary system with voting staggered by region or provinces, we might expect participation along the following lines, using 2011 as a baseline and 12.3%, 42.5% and 75% as weak, average and strong turnout discount factors applied to our most recent electoral results to estimate turnout in primaries (you can see my worksheet here). Thanks to the Pundits' Guide for election results by region.

Projection (using 2011 as baseline)

Atlantic Canada (strong): 250,000
Central Canada (average): 825,000
Western Canada (weak): 61,000
Northern Canada (average): 4,500

Total: 1.2 million

So, I think that it's fair to say that if we switched to a primary system, and if we succeeded in registering a very high proportion of our supporters, we might expect 1.2 million or so voters to participate in our primaries for a leadership election. This is 10 times as many as we might expect to participate in a WOMOV system (the Conservatives had just under 100,000 in 2004).

To review the various systems and the approximate number of votes likely needed to win:

Delegated system: 2,500 votes
WOMOV: 50,000 votes
Primaries: 600,000 votes

Reaching and mobilizing this many voters would of course come with financial and human resource implications -- similar in scope to a major provincial election, judging from the numbers. Increased spending requirements would likely limit the field of initial candidates. The winnowing process inherent in staggered voting would further narrow the field as we moved across the country.

Finally, and most importantly, adopting primaries would reduce the influence of members to roughly 1/10th of the influence they enjoy now under the WOMOV system. Considering the situation we are in right now, I don't think that disempowering members in this way, and on this scale, is a good idea. I don't know why most people would continue being or choose to become members if we made this change.

Bold, but not smart
In speaking with proponents of this proposal, it is clear that they are very excited about the possibility of opening up the party, reaching out to more Canadians and making a bold change that they believe will make a big difference. Even though I disagree with the proposal, I'm glad they are so actively engaged with it and enthusiastic. Unfortunately, the rationale for the proposal is sorely lacking. We are being asked to abandon a perfectly good system without crucial information about the costs of this change, how exactly the voting days would be staggered, how higher spending limits would influence the field of candidates, etc. Compounding matters, we are facing a fiscal crunch with the phase-out of the per vote subsidy and we don't have a lot of time to adopt and implement this change in advance of the 2013 leadership vote -- look at how much time and effort goes into the US primaries. The idea is well-intentioned, but it is rushed and misguided. I believe we can reach out to Canadians and better inform our campaigning without enacting the proposal.

Membership has to matter. It has to mean something. If you will be attending the convention, I hope you will join me in voting against this proposal.

Note: you may want to read up on the US primary system, as it is the one we are superficially familiar with. It's actually fairly complicated, with state primaries or caucuses determining the number of delegates that the respective state Democrats or Republicans send to their national conventions, where the nominee is officially elected by the delegates. Fans of US history may recall when Gerald Ford defeated Ronald Reagan at such a national convention. Usually the outcome of the national convention is known in advance, as most of the delegates are bound by the results of the primaries and caucuses in their home states.

Update 6 Jan 2011: Based on the results seen in the Socialist Party primaries in France (2007 presidential election and 2011 open primary), I would expect participation in a Liberal Party primary to be between 470,000 to 850,000, with the lower number based on 2011 election results and the higher number based on our 2004 election results. Broken down by region, that would mean as few as 24,000 in Western Canada (2011 baseline) and as many as 585,000 in Central Canada (2004 baseline).

Roughly 2.9 million people voted in the Socialist Party primaries, which is 17.3% of the 16.7 million who voted Socialist Party in the presidential elections and 6.5% of the voting eligible population (2.9/44.4 million).

Update 6 Jan 2011: The Alberta Liberals used a primary system for their leadership selection. They signed up 27,000 or so supporters, 8,640 of whom actually voted in the leadership election (31%). Details at Blunt Objects.

• Previous Entry: Unofficial calendar of events for Liberal Party convention
• Next Entry: Globalization and free trade