According to Robert Goodin (2008), one problem with a deliberative forum (allotted or otherwise) is that:
conversations, seen as serial processes with dynamic updating, can easily be path dependent. The outcome of the conversation depends upon the sequence of conversational moves, particularly those early in the conversation that set it off down one path rather than some other. (p.114)
A path-dependent process displays the following features:
Unpredictability. Because early events have a large effect and are partly random, many outcomes may be possible. We cannot predict ahead of time which of these possible end‐states will be reached.
Inflexibility. The farther into the process we are, the harder it becomes to shift from one path to another.
Nonergodicity. Accidental events early in a sequence do not cancel out. They cannot be treated (which is to say, ignored) as ‘noise’, because they feed back into future choices. Small events are remembered.
Potential path inefficiency. In the long run, the outcome that becomes locked in may generate lower pay‐offs than a forgone alternative would have. (p.112)
Goodin illustrates the problem for deliberative democracy with a stastical model called the ‘Polya urn’. In this model an urn contains two balls, one red and one black. Draw out one ball. Return that ball to the urn, together with another ball of the same colour. Continue the process until the urn fills up. Then tip out the balls and count how many there are of each colour, the chances being that early selections will lead to a predominance of either red or black balls. How might this play out in a deliberative assembly with full speech powers?
Imagine a dozen‐person group trying to answer a binary true‐false question. Imagine that each member of the group has some private information bearing on the question, as well as the public information that is shared among everyone in the group. Suppose we are good deliberative democrats, so (a) we accord each other person’s views the same weight as our own and (b) we update our own views in light of what others have said dynamically as they say it.
Now suppose we go around the table clockwise, each of us announcing his or her decision in turn. The first person says ‘false’. Suppose the private information available to the second person also disposed her to say ‘false’—which she does all the more emphatically for having heard the first speaker’s ‘false’. Now comes the turn of the third man. His private information suggests that ‘true’ is the correct answer. But both of the first two speakers have said ‘false’, and he accords their views as much weight as his own. Dynamically updating his views in light of theirs, therefore, he concludes (albeit more reluctantly than the second speaker) that ‘false’ must be the correct answer, and reports accordingly. The weight of opinion come the fourth speaker is three‐to‐zero ‘false’; so after dynamic updating she says ‘false’ even if her private information suggested ‘true’. The weight of opinion come the fifth speaker is four‐to‐zero ‘false’; so after dynamically updating he says ‘false’ whatever his private information. And so on around the table. In the end, the decision (p. 114 ) is unanimous. ‘False’ has emerged as the common consensus among the group.
But suppose we had gone around the table counterclockwise instead. And suppose the private information available to the first person and the second person going around in that direction suggested that the correct answer is ‘true’. By exactly the same serial process of dynamic updating, the group would then have come to a unanimous decision in favour of ‘true’. In short, had we gone around the table in the opposite direction, the common consensus would have been just the opposite.
The general point is that conversations, seen as serial processes with dynamic updating, can easily be path dependent. The outcome of the conversation depends upon the sequence of conversational moves, particularly those early in the conversation that set it off down one path rather than some other. (p.113)
This is not a big problem for ‘procedural’ democrats, although it is a serious issue for those of us who argue for the epistemic benefits of allotted deliberation. A typical deliberative forum has a purely advisory role, the final decision being taken by elected members who may well attempt to bracket out such factors when they study the micro-deliberations. But for those of us who seek to establish allotted legislative assemblies, path-dependency is another reason to insist that speech acts should be limited to asking questions of representatives. Goodin concludes that the variations in the three Texas energy DPs (Fishkin, 1995, p. 220) is a result of the conversational path-dependency effects of small-group deliberations.
Goodin acknowledges that a Polya urn that contains an initial (say) ninety-nine red and one black ball will inevitably reproduce into an urn that is almost entirely red, so those who argue that politics is a competition between class interests (the elite and the masses) will draw comfort from this argument.
Goodin, Robert E. (2008), Innovating Democracy: Democratic Theory and Practice After the Deliberative Turn (Oxford University Press).
Fishkin, James S. (1995), The Voice of the People (Yale University Press).