Democracy, nowadays, well, since North American Independence, refers exclusively to the use of voting. Even the UN defines the right to democracy as a right to vote¹. The founding fathers were basing their new political system on a style of representative democracy where representatives are selected because people can trust them to handle affairs in the best possible way. The selection process was by vote, and that mechanism has become the global standard. Just occasionally various people become concerned that voting isn't working, usually when it isn't working for them. We've had a flurry of such complaints in the UK recently and the proposed solution has been to improve voters' ability to make the 'right' choice by educating them in economics and social science. A similar proposal in the USA boils down to disenfranchising people who are not able to pass a test². As it happens, economics has been a disastrous source of political guidance since Reagan and Thatcher, and social science doesn't help anyone understand what's going on outside their front door in a ex steel town in Wales. It would be even worse to disenfranchise people who might well be the very people that need to be heard³.

Let's do some first principles work here. What sort of choice is susceptible to voting? When is a simple majority appropriate and when is it not? Does voting work for multiple choice situations?

Taking the last question first, the answer is generally not. Picking one outcome from three possibles means that, in a close run decision, 2/3 of the voters do not get what they voted for. Allowing for multiple outcomes, using some form of transferable vote, doesn't help much. Different proportional voting systems have been put forward over the years and none have addressed all of the problems. The problems are not so obvious where some kind of discussion and trade-off can be arranged, but this really only applies to small, well connected, groups. And, of course, the key point in that case is the availability of discussion, making the vote itself more of a formal recognition of an agreed aoutcome.

If we look at simple binary choice we can say that the options must be mutually exclusive, and that the outcome of the vote must be clear. This is not always the case. The obvious example here being the Brexit referendum, where the leave option did nothing to clarify the possibilities for the final relationship with the EU. On the other hand, if the choice goes into too much detail then one side might loose votes over a some detail, ignoring the general principle.

In a similar vein, the options should not be divisive. If one side or the other is going to be seriously upset by loosing then it might be best to discuss the matter before hand to determine whether a potential vote for one side or the other is a proxy for some other issue.

As a slight digression, issues that are potentially divisive are not normally handled by a simple majority. Organisations that wish to change a constitution, for example, may have a built in super-majority requirement to ensure that the number of dissatisfied voters is minimised.

And, perhaps obviously, the options should be relevant to the target voting audience and open to discussion. Selecting an MP is perhaps the worst example: voters consider either party to be as bad as the other, and discussion is largely replaced by predictable speeches and sound bites.

Most of these cases are easily handled within a small organisation where everyone is familiar with the goals of the organisation and discussion is encouraged. A vote might be taken at the end of a discussion simply to record a result, or to terminate discussion at the end of an allotted time. Discussion becomes more difficult as the size of the group increases and when discussion is not possible then voting should not be attempted.

Ultimately, voting works when it is preceded by useful discussion and deliberation. The discussion clears the air and allows for a proper agreement on the options that might be best handled by a vote. In the best circumstances a vote becomes a record of an agreement already arrived at.

Clearly then, both voting for a Member of Parliament and voting in a referendum are, or should be, non-starters. In a properly representative democracy there should be no need of referendums, so we can ignore that problem. Constructing a properly representative democracy is another matter.