The Lib Dems have always selected their candidates by “one member one vote” (OMOV). It has always seemed the most logical and transparently fair system, and it is certainly better than having candidates hand-picked by an inner cabal. It still does a fairly good job at selecting candidates for the House of Commons, although as membership levels drop that is becoming less true. But it has been quite inadequate for selecting candidates for larger constituencies, particularly for the European Parliament and London Assembly.

Here’s the fundamental problem: a significant proportion of our members are concentrated in our held and target constituencies. Target seats become target seats because they have a larger pool of activists from which to draw. In turn, in order to become winning seats they have to recruit more activists. The more tightly we focus on target seats, as the Lib Dems certainly have for the past two decades, the more the gap between target seat and what we sometimes euphemistically call “development seats” widens.

If you are a candidate for an internal election spreading across a whole region – even a relatively compact region like London – you have a formidable task. OMOV however makes it easier. If the vast concentration of members live in a relatively small area, then focus on them. You can be reasonably certain that your rivals will do the same. In the case of London (people from the rest of England, Scotland or Wales will have their own examples), this means focusing your activities in the South and the West. The rest of the city gets cheerfully ignored. No surprise therefore that since Lynne Featherstone became an MP, we have had no Assembly Member from north of the river.

(In the case of the London Assembly the situation is actually worse than with European Parliamentary lists. In London last year, the focus of activity was based on winning the South West constituency. So not only did the South West get to dominate the list but it got a second bite of the cherry as well).

At this point you may be thinking to yourself “well boo-hoo!” At least all party members are being treated equally. The problem however is that if our party can’t demonstrate that it represents the whole of London, then why should the rest of London vote for it? Many commentators have pointed out that Ken Livingstone’s failure in 2008 was that he was seen as being an “inner London” politician. With our vote floundering at the same time, can we really claim we don’t have a comparable image problem? And, dare I say it, what does it say about us if we fail to field a single BAME candidate with a chance of winning in a city which is almost a third non-white? There comes a point when holding onto OMOV starts to look less like fairness and more like vanity.

Far from ignoring this problem, the truth is the party has always had procedures to mitigate it hard-wired into its constitution. The conference representative system which we use gives local parties with fewer members more representation proportionately at party conferences to prevent larger local parties from dominating procedures. We use this system to elect, among other things, our federal committees (and when the membership was asked back in 2000 whether it wanted OMOV for this, a clear majority of respondents said they didn’t). However, it is hard to see how the conference rep system could be justified as a suitable system for selecting parliamentary candidates at a time when people are complaining that such procedures shut too many people out.

What’s the solution? Both Conservative and Labour politicians have been talking recently about primaries and indeed the Tories ran a primary election for London Mayor in 2007. I believe it is time the Lib Dems similarly looked at opening out our procedures for selecting our Mayoral and Assembly candidates for 2012, and possibly the European Elections in the longer term.

One thing I should be clear about: primaries are not a particularly good tool for increasing political participation. If you are serious about democratic renewal, then you have to support electoral reform. What they are good for is reviving political parties, something we could do with a bit of. Indeed, that is the crucial lesson we can learn from the US. In the US, candidates use primaries to build up their supporter base and use those supporters to drive their subsequent election campaigns. The UK has nothing comparable. Moribund areas remain moribund and we do nothing about them.

I don’t believe the Lib Dems can afford to run an open primary across the whole of London along the lines of what the Tories have recently done in Totnes. That would cost somewhere between £2.5 million and £3 million. Instead, I would like to see us run a series of caucuses.

Another system which is used in the us (most notably, in Iowa), caucuses work by allowing any registered elector to turn up and vote. They have lower participation rates but are more deliberative. We would have to publicise the events, but they would be a lot cheaper to run than an open primary. And they would work as recruiting grounds: everyone who attends should be given a membership form.

The process I envisage for London Assembly and Mayor (for European selections it would be much simpler) would operate over a series of weeks. Each of the 14 London Assembly constituencies would have their own caucus at which the mayoral candidate, assembly list and constituency candidate would be voted for. Because not every caucus could practically take place at the same time, and in order to mitigate the “snowball effect” of a candidate doing well in the early caucuses, they should take place in a strict order according to the number of members in each constituency, starting with the lowest upwards. The results of each constituency would count equally (or alternatively be weighted according to population size) which would mean that no candidate would have an incentive to ignore any given area.

Party members would still have a central role. As well as having as much right to vote as anyone else, they would be gatekeepers to the process. To stand, candidates would have to secure the nomination of a certain number of members from each constituency (10?) in order to move onto the caucus stage.

A final rule: we should adopt the practice of having our mayoral candidate appear on the assembly list automatically. The position they appear on the list should be one place lower than the total number of list candidates that were elected in the previous election. For 2012, that would mean the mayoral candidate appearing fourth on the list (in 2008 Brian Paddick would have appeared in sixth place under this rule). This means that in lieu of winning outright (which will always be a tall order for a third placed party), the mayoral candidate still has a goal to strive for and the campaigns for Assembly and London Mayor complement each other better.

Such a system will almost certainly mean more work for candidates, but no more so than we should expect of individuals who are seeking to represent the whole city. The benefit of the system is a candidate who emerges not only with a higher profile, but with a higher profile across the whole city. The system could also be used as a recruiting tool. Would there be a risk of other parties and groups flooding caucuses? A small one perhaps, but not significantly more so than we already risk by allowing anyone to join the party, and organising such a campaign on such a large scale would be a major feat indeed.

I, for one, would be much more excited about that than selecting candidates by putting numbers against a load of people who I don’t know and have spent the selection ignoring me because I live in the wrong part of town. The Lib Dems have fared appallingly in region- and London-wide PR elections – it’s time we started being a bit more imaginitive.