In this analysis, Tom Flanagan - the former advisor to Stephen Harper (and also his former professor) - explains how the preponderance of moderate voters in the political middle ensures that the parties on the left and the right will converge ‘in the middle’ in a two party system. While some of his conclusions are not far off the mark, Flanagan’s analysis itself couldn’t be more wrong, mostly because he has made two serious errors: he completely misunderstands the nature of the moderate voter, and he has too much faith in the one-dimensional model of politics.
Lets look at the second error first. The one-dimensional model states that ‘political space’ can be described with one parameter, usually by a big government index. According to this model, the NDP is on the left, the Bolsheviks are on the far left, small-c conservatives are on the right, and Ann Rand objectivist are on the far right. But if this model describes the key features of politics adequately, how come Hitler’s National Socialists are universally considered to be on the far right? And how come socially conservative, big government populists like Pat Buchanan and libertarians - who believe in small government and personal freedom - are both classed as right-wing parties. If the one-dimensional model is right, shouldn’t Pat Buchanan be a lefty moderate while a gay libertarian is a right-wing extremist!
In reality, one parameter is nowhere near enough to describe important features of the body politic.
In science, there is a saying: all models are wrong but some are useful. What makes the one-dimensional model useful is that the parameter it uses – economic freedom – is usually dominant. But this is not always the case. Take Quebec, where politics is split on federalist-separatist lines. There the conventional left-right split occurs within the parties themselves. In the real world, all political parties are expedient coalitions of diverse interests and how they come together depends on specific circumstances.
Another place where the one-dimensional model doesn’t work is a blue collar riding like Oshawa where voters are socially conservative but economically socialist. Because the main parties are split along conventional left-right lines, there is no party that really services them. They alternate between Conservatives and the NDP, depending on whether social issues (like gun control or family values) or economic issues (like free trade and middle class entitlements) predominate. A conventionally moderate candidate who is socially liberal but economically conservative, like say John Tory or Scott Brison, are poison to them. The one-dimensional model does not explain the Oshawa voter.
Now lets tackle Flanagan’s first error, a misunderstanding of what political moderation is. As I explained in this post, the moderate voter in the middle is not somebody whose political philosophy lies in the statistical midpoint of politics space. Rather, he is simply an apathetic voter who doesn’t care about politics at all. What political scientists like Flanagan see as a preference for moderation is really just a sensible desire to not rock the boat when things are OK. These people believe that there are a lot more ways to break things than there are utopian schemes that will actually make things better.
The flip side of this sentiment is a preference for decisive action when they see the wheels coming off the machine. This is why Margaret Thatcher, Ronald Reagan, Mike Harris and Rob Ford are popular, even though they were decisively to the right of the people who elected them. In serious times, the moderate voter wants – nay, demands - a leader who offers him a credible answer to the problem of the day and then boldly carries it out. A ‘moderate’ split-the-difference solution just repels him. Because Thatcher, Reagan and Harris offered convincing solutions that worked, they succeeded in moving the statistical middle towards them.
So when times are good, splitting the difference Median-Voter-Theorem-style works. But when the seas get choppy, you better have an answer that isn’t a weasely half-measure.