I'm picking off some significant issues one by one at this point. This page is about avoiding contention. We have a structure containing a large number of citizens' assemblies, more or less long lived, more or less deliberative, and of various sizes. All of them are representative, in the sense that they are populated through a sortition process, and all of them need to demonstrate legitimacy. Legitimacy for the system comes from a demonstration that the structure is working; that we have the right set of assemblies linked together is a useful way. Legitimacy for each assembly comes from being seen to be doing the job right; the appropriate information is sought and used, and the decision making process is seen to be fair.

These issues are, in the proposal, the responsibility of the Oversight Council, and, naturally, will be addressed as issues arise - perhaps even before anyone else notices them. However, there are some things that seem worth mentioning at this stage, perhaps to reassure myself (if no-one else) that deliberation can be effective and practical. I am not concerned here with the validity of participants' opinions, nor the ability of participants to listen to others. Rather, I want to think about how a deliberation reaches a decision. This has two aspects - one is how to terminate the discussion in a reasonable time, and the other is to avoid feelings of unfairness or ill-will in those who might disagree with the published result.

One model that crops up in this context is the Quaker meeting model¹, where everyone gets a say and the discussion continues until consensus is reached. This is designed to give equal weight to all participants and to reduce feelings of unfairness to a minimum. There are two problems with this. One is that it only really works at small scale, and the other that people want to get home sometime tonight. The meeting might be hurried to a close by coercing dissenting voices using emotional pressure - don't rock the boat, for the sake of the community, and so on². In a similar vein, the Occupy movement developed a flat structure whereby everyone has a right to express an opinion. In this case the discussion is managed by a chair (the person with the loud hailer) who can terminate the discussion when it doesn't seem to be getting any further. At that point any dissenting voice acts as a veto on whatever was being proposed³. These two meeting styles have a similar basis but achieve opposite ends, in that the one achieves a majority decision, while the other is restricted by a minority veto. Neither mechanism is going to work in a citizens' assembly: the social cohesion of a citizens' assembly is not enough for emotional pressure (and such pressure would be unacceptable anyway), and having the possibility of an absolute veto from a single voice would be inhibiting to the point of complete immobility.

The principle that everyone should have an equal voice is, of course, fundamental. The problem is to terminate the discussion in a way that produces a useful result without leaving a bad taste. The answer more or less has to be a voting process, used with care. Well, I say a process. There's really going to be a number of processes chosen for the type of proposition being voted on and the number of participants in the assembly. The goals are in part to ensure that the result is demonstrably the result of deliberation and subsequent agreement between parties, and part to make certain decisions more difficult. The basic principle that addresses these two goals is a rather negative one: simple majority is not appropriate.

For the Agenda Council⁴ we have three types of decision: one to decide whether the issue is worth handling at all, one to decide on the general approach to be taken, and one to agree that the material to be forwarded to the relevant Interest Group is now ready to be passed on. The creation of documents is a more of a team effort. All of these are important decisions that have potential to be divisive, although the middle one - choosing an approach - is perhaps the most problematic. After all, if the issue is rejected out of hand at the first stage it can always be resubmitted, and, by the time documents are prepared and ready all of the serious decisions will have been taken, based on that middle stage decision. What I am thinking of here is that all three votes should require a super-majority for a decision. In the first case, a super-majority is needed to reject the issue. In the last case a super-majority is required to accept the documentation. In the middle case, though, we might want a three step result. A super-majority to accept a proposition, a super-majority to reject a proposition, and an 'undecided' result in between, in which case the issue goes back for further discussion. Since the Agenda Council sets the running for all issues, we cannot risk laws being proposed, or rejected, on a 50% vote. Something nearer 75% would give more confidence that the result is, in some sense, the right one.

A Policy Jury, on the other hand, is more of a hit and run set-up. Each one is convened to review a bill written by an Interest Panel and, if it passes the bill becomes law. This is achieved by the Jury receiving presentations and going through a secret ballot. My thought here is that, if there is any problem with the bill then something is wrong. I would hope for full acceptance at this point. That's unlikely, maybe, but I would still call for perhaps a 90% acceptance being required for the bill to pass.

In between, we have other panels where we can apply similar considerations to establish how easy or difficult it might be to reach their various conclusions. All in all, the voting process is a tool to either apply brakes, or allow quick reactions, and its proper design is an essential contribution to the legitimacy and overall acceptance of citizens' assemblies.