Thursday, 21 April 2011

Elections are about more than counting votes

Democratic elections are generally characterised as a competition between political parties for votes, and hence power. Supposedly 'the people' are presented with a menu of options by the politicos and make a single choice from that menu depending on what they prefer on election day. Social choice theory is the academic discipline most concerned with elections and it focuses on analysing the advantages and disadvantages of different ways of counting votes in terms of fairness. But something is missing from its analysis: elections are about more than counting up votes. Elections are also an extended strategic engagement between the voters and the politicos to decide what goes onto the menu of the ballot card.

Now of course there are lots of ways in which 'the people' may influence politics outside of their votes in elections, through the activities of civil society organisations, political party membership, demonstrations, the media, etc. But here I want to focus on election voting and show how this apparently closed and narrow choice is actually far more open and extended than often appreciated.

Elections formally take place on 1 particular day, but actually take much longer. Before elections, the politicos engage in election campaigns. Why do they do this? The standard view is that they are trying to win over potential voters to their brand, to capture market share from their opponents. But in that view the politicos would only be concerned with the persuasive presentation of their position: voters preferences are taken as fixed and up for grabs. Politicos talk, voters listen.

But the actual situation is much more interesting, and it has more active players whose best strategies depend on those of other players. In other words, we are in the domain of dynamic strategic interactions where perceptions of other players intentions is central. For to start with, the politicos are engaged in strategic competition with each other. In order to gain market share, they must find out what the people want to vote for and against, and then promise the good stuff and abjure the bad in ways that are credible and consistent with their other commitments. Elections are thus not just about style - the politicos have not yet decided on the content that they will be putting on the menu. So parties conduct private opinion polls and focus groups and analyse what different kinds of voters say they want and who they are planning to vote for in order to see what offerings might bring the greatest returns. And of course the other politicos are experimenting with their own strategies to outflank each other and win support.

But information about what other people want and how they are considering voting is too valuable to leave in the private hands of politicos. Public opinion polls published in newspapers place this information in the public domain, where everyone can make use of it. It is often remarked pejoratively that opinion polls are inaccurate because respondents lie. It is also argued, in direct contradiction, that polls influence people to vote for the winner. Both critiques misunderstand the nature of such polls, and are firmly based on the prejudice that voters are idiots.

The key question in such polls is a counterfactual one: "If the election were tomorrow, who would you vote for?" The answers provide valuable information, but not information about which party people will in fact vote for in the real election (if that were the case, we wouldn't need to bother with the expense of actual elections). What polls do is allow 'the people' to communicate directly to political parties about their feelings and intentions in a non-verbal way. It reveals something very like a 'price', or level of effective demand, for different policy bundles in the market for political support that complements the verbal public deliberations taking place in media debates and analysis.

But it also enables voters to communicate with each other, and this is key to voter power (this is perhaps the real reason why polls are banned in some countries before elections). Now not only do the politicos in the centre-right party know that a lot of people really don't like them (compared to the last election), but they know that the voters themselves know this and are likely to use that information to improve their own strategies. Voters know that in order to get what they want they have to act together. By understanding how other people are thinking of voting they can see which parties have a reasonable chance of success. Within that limited set voters can for example choose to keep their most disliked party out of power by choosing its closest competitor. The politicos are forced to change their positions to adapt to this dynamic interaction, to protect themselves from being outflanked by voters' strategic choices, for example by moving to the middle ground so they are less detested.

One thing politicians fear is the strategic co-ordination of the voters against them: the voters' implicit, non-verbal, challenge, 'You know how unpopular you are, so what are you going to do about it?' It's an important mechanism by which voters fight back against the politicos' power to set the agenda, to determine that short menu of options on the ballot paper itself. After all, votes are a pretty pathetic method of exercising democratic control over governments. On the face of it all votes can actually achieve is to passively endorse one predetermined option. And in any case, an individual's vote literally cannot even be counted - the measurement error, even of computerised systems, is always greater than 1 (because of human error in following voting instructions, administering the system, etc.). There is no rational instrumental justification for the act of voting (i.e. as a means to achieving something you want), only the intrinsic values of expressing your preferences and performing your citizenly duty.

As in actual markets, voters can only make their voice count in the aggregate (just as one person's boycott of Walmart is invisible). They must therefore rely on collective pressure to push politicos to make products they actually want. Unlike in normal markets, elections present the informational challenge of establishing fair value in the case of a single unique transaction. This can only be overcome by extending the election process to create shadow markets for voting intentions, in which voters threaten to vote instrumentally in the real election.

Social choice theory is concerned with how a snapshot of people's preferences (an election) can be translated fairly into a collective choice. A great deal of its work consists of developing and testing different aggregation rules, of votes to outcomes, to see how they work in theory and how they measure up against our intuitions about fairness. For example, they point out that under first past the post election systems it is easy to get a winner whom the majority of the voters do not like at all, because information about 'dislikes' isn't counted in the system. This seems unfair and somewhat arbitrary since it implies the rule of a minority. Furthermore, in order to prevent such outcomes people often have to 'lie' - to vote against their true preferences in order to prevent their least favoured party winning most votes. Social choice theorists therefore recommend alternative systems that capture and utilise more information, although every system has severe limitations, even in theory.

But there seems to me to be a problematic assumption at the heart of social choice theory's approach, in that it treats elections as snapshots rather than as an extended dynamic strategic interaction between the voters and politicos. For this reason it takes its goal as fairly and truly translating people's preferences into the people's choice. That means that it sees strategic voting as a problem. The aim is to build all the fairness into the system itself so that it is fit for use even by morons who know nothing about politics apart from their true preferences.

Reasonably adequate election systems are essential to democracy and far more significant than the delicate mechanism of extended strategic interaction I've described. Nevertheless the best of social choice theory may be the enemy of the good if better fairer voting systems change the wider system and prevent or impede democratic engagement between the politicos and the voters-as-players, not just as preference-holders. The Alternative Vote election system about to be put to referendum in Britain, for example, seems designed to squeeze the entire strategic interaction of an election onto the ballot form itself.

No comments:

Post a Comment

Comments will be moderated for civility and relevance