In the last post I grumbled my way towards admitting that I will be a grudging ‘Yes’ voter in the AV referendum tomorrow, on the basis that, while there’s a lot wrong with that system – notably its complete lack of proportionality – it’s still fractionally better than First Past the Post, since it should at least be easier for people to block the candidates they can’t stand. I ended that post by suggesting that I did have ideas for a different system, one that I think would be a significant improvement over either of the options available in the referendum, but that it would make sense to talk about it in another post. So, without further ado, allow me to present my preferred alternative to AV.
Defining the problems
The big problem with the First Past the Post constituency system is that it’s completely unproportional – a party can have a considerable following, but if that support is evenly spread across the country as a whole they may not win in any individual constituency. It’s this problem that has led to the historical under-representation of the Lib Dems at Westminster (though it remains to be seen if anyone much is still prepared to vote for them in the future), and the problems faced by emerging parties like the Greens and (shudder) UKIP.
The simplest way of creating a proportional system would be to abolish constituencies and just hold a single poll across the whole country which decides the proportion of parliamentary seats each party is to receive. There are two major problems with this solution. Firstly, it completely does away with individual accountability. If voters decide an individual MP is doing a bad job they can’t directly vote them out. They can, of course, vote against the party in the national poll, but if the candidate they’re dissatisfied with appears high up the party list they are unlikely to be unseated. Secondly, a national system would have the effect of making it very hard for independents to win election, since they usually stand on issues that have a local resonance, but a national election would encourage voters only to think about national issues.
Solving the problems (1): Accountability v Proportionality
The first part of the solution to this problem is easy: multi-member constituencies. The persistence of constituencies allows for some measure of individual and regional accountability, while the election of several candidates allows some scope for proportionality. Multi-member constituencies are not an ideal solution – the fact they are larger means accountability is somewhat diluted, while the subdivision of the country means that very small parties (those which might deserve one or two seats nationally) will fail to receive their just deserts – but they are a reasonable solution. I’m not aware of any other approach that balances the competing interests of accountability and proportionality more effectively. (If you are, please tell me about it in the comments; I’m always interested to learn more.)
The next issue that needs to be considered is how large these multi-member constituencies should be – or, to put it another way, how many MPs each of the new constituencies should elect. If the number is set too small – at two MPs, say – proportionality will suffer, since constituencies will not be big enough for small parties with evenly-spread support to gather enough votes. On the other hand, if the number is set too large – ten MPs, say – accountability will be too far diluted. I think a reasonable compromise would be to make each constituency four times the size of current constituencies, and for it to elect, therefore, four MPs. In some urbanised parts of the country it might be possible to make constituencies larger while keeping MPs reasonably close to the people that elect them, but in rural areas the distances would become too great. At the same time, a cohort of four MPs should mean that any party achieving 25% of the vote ought to be rewarded with an MP. As set out above, this isn’t ideal because of the likelihood it will exclude smaller parties, but under representative democracy something, regrettably, has to give.
Solving the problems (2): The actual voting system
This issue is slightly more complex. The simplest approach would be to allow each voter one vote; these votes would then be totalled, and the seats distributed between the parties on a proportional basis. This approach has the very great advantage of being simple, but the downside is that voters may feel pressured into voting tactically – would an elector feel confident voting for a small party, or would they worry that they ought to cast their single vote for whichever party seemed best-placed to block the party they want to see lose? As always, the problem with tactical voting on a personal level is that it prevents individuals from voting their conscience, while the wider problem is that it tends to entrench a maximum of two parties within each constituency.
One solution to this problem is to introduce some form of preference voting, with the method most often suggested in the UK being the Single Transferable Vote. (I’m not going to define STV – I’m trying to keep this simple – but the Electoral Reform Society have a good, if rose-tinted, guide here.) The big problem with most systems of preferential voting, including STV, is that a second (or lower) choice vote counts, if it comes into play, for as much as a first choice; as I suggested in my AV post, this can give rise to a situation where everyone is unhappy with the outcome of an election (though this is less of a problem in multi-member constituencies). For this reason, I’m an advocate of a solution called the Borda count (or a slightly modified version thereof, since the classic Borda count was designed for single-member constituencies).
Under this method, voters would rank the candidates in order of preference, as in other preferential systems, but the difference with Borda system is that each ranking counts towards the outcome of the election. Under the Borda system, parties receive a number of electoral points, depending on how highly they’re ranked. Imagine a scenario in which there are five parties standing for election, and that I’ve ranked them as follows:
- Conservative 
- Green 
- Labour 
- Liberal Democrat 
- UKIP 
Because there are five parties standing in this constituency, the maximum number of electoral points any party can receive is 5. On my rankings above, then, the parties are awarded electoral points as follows:
- Green – 5 points
- Labour – 4 points
- Liberal Democrat – 3 points
- Conservative – 0 points
- UKIP – 0 points
(If I had ranked the Conservatives 4th and UKIP 5th, they’d have received 2 and 1 points respectively, but I’ve left them unranked for tactical reasons; I’ll come back to this.)
When the votes are counted, the total number of electoral points for each party is added up and the available seats are distributed in accordance with the proportion of electoral points received by each party. In the example above, I’ve left the Conservatives and UKIP unranked because I don’t want them to win under any circumstances; by withholding the small number of points they would have got from being ranked 4th and 5th I’ve reduced the size of their overall points total, and so the likelihood of one (or more) of their candidates being elected.
Borda has a number of advantages over other systems. The major advantage is that the outcome of the election directly reflects the preferences of voters. In this scenario, I want the Greens to win, Labour would be a tolerable second best, and I could live with the Lib Dems, and that desire will be directly reflected in the points total for each party. This is a striking contrast with most preferential systems, under which I might have been counted as backing the Lib Dems as enthusiastically as if they had been my first choice. Another advantage of Borda is that it allows for a mixture of positive and negative voting. If I have a single vote, I can either cast it positively (for the party I want to win) or negatively (in such a way as to block the party I want to see lose), but under Borda I can achieve both goals simultaneously, being able to vote for the Greens and against UKIP and the Conservatives at the same time. A further advantage of Borda is that parties are rewarded both for inspiring strong support amongst their core supporters and for appealing to voters more widely. This is a pleasing blend of what tends to occur under First Past the Post (where building core support is important for creating and maintaining safe seats) and most other forms of preferential voting (which can encourage parties to create bland policies which don’t inspire many people, but also don’t repel many).
So, there you have it, my proposal for reform of the electoral system: four-member constituencies, with candidates elected from party lists by means of a slightly modified Borda count. I’m not an expert on any of this (just a tragically obsessed amateur), but I think this is as close to an ideal system as you can get under representative democracy; I’m ready, of course, to be corrected if I’ve missed something.
As for whether there might be an alternative to representative democracy that would be even better, well, I have ideas there, too. But you’ll be relieved to hear I’m not going to bore you with them right now.