Lottery Democracy

Adding a little randomness to democracy is sometimes to the good. Why, when, and how far?

The planning meeting for the neighborhood semi-monthly party was a model of democracy. There was even a whiff of “degree affected democracy,” as those who were going to be out of town on the party night, would be otherwise engaged, or were uninterested, simply absented themselves from the meeting and its votes. (For more on degree affected democracy see Post of 9/13/14,

The Moral Underpinnings of Democracy and its Degree-Affected Variant ) This month, as usual, “Heirloom Rock,” won out as the musical choice for the event. Also, as usual, the partisans of what came in the neighborhood to be called “Millennium Rock” and “Adult Hip-Hop” grumbled a little.

Then Jen, perhaps she had been reading some Athenian history, suggested a new voting procedure for use in the next meeting. Everyone marks ballots as usual. Then they are all be put in Ms. J's large monkey pod salad bowl. Finally one ballot is pulled out of the bowl by little Zisi and whatever was checked on that ballot would be the music for the party. Jen explained that in this way the fans of not- quite-so-old rock and of rap would sometimes get their music, In the long run each of their nights would be in proportion to their numbers.

This suggestion was adopted, and it worked out even better than Jen prophesied. With the new voting procedure officially in place, members of the different camps started negotiating for mixed musical programs. Rather than risking an entire night of uninterrupted hip hop, however adult, the Heritage Rock majority was willing to compromise on giving some time to one or both of the other genres at each party. Preferring not to wait perhaps a year for their music, the younger set were happy to get some of it played at least every other party. No one was completely happy with the arrangement, but no one felt that their preferences were entirely disregarded or that they were second class neighbors.

Lottery democracy, or “sortition” does indeed go back at least as far the 500 selected by lot in ancient Athens for its legislative and jury functions. It never guaranteed right outcomes, as one widely reported criminal case for youth corruption and impiety illustrates, but it had some egalitarian virtues and benefits for diversity and against corruption. As philosopher Allan Gibbard noted, sortition also eliminates strategic voting. In a plurality election with three choices, of which you anticipate yours to be third in popularity, you may vote for your second preference to try to keep your least favorite from winning. With sortition you can vote for your favorite without increasing the chances that the greater evil will win. And it is possible that you will win – the probability if which you increase by voting. There is always an incentive to vote, no matter how small your minority.

Sortition, modified by challenges for cause and a limited number of peremptories, constitutes most trial juries in the common law tradition. Grand juries are made up by lottery with even less filtering. Robert Paul Wolff and Akhil Amar have proposed more sweeping use of sortition in electing legislators or other public functionaries. Wolff was looking for a mechanism for performing some state-like functions with as little as possible of what he saw as the moral illegitimacy of the state. Amar's chief concern was gerrymandering, wasted votes, and, in general, the unhappy lot of voting minorities.

My neighborhood party case responds primarily to the Amar family of concerns in the case of issue voting. Jen's proposal appeals to our intuitions of fairness. The results have a good claim to be appropriate, even handed, and to afford equal dignity to all the neighbors..

Its merits, however, depend upon some assumptions. Consider, first, ephemeral minorities. Suppose that the musical tastes in our neighborhood were flighty. A member of the Heirloom Rock majority is not significantly more likely to prefer that music six months from now than is a member of the Adult Hip-Hop minority. One's musical allies one month might well be one's opponents two months later. There are always minorities, but no one is in the long run more likely to be in a minority than in the majority. Under these circumstances the chief virtue of sortition disappears, and a serious negative becomes obvious.

That a minority position wins out over a majority always means that more people have their choices frustrated. The core value of majoritarian democracy is that represents the best projection of free choice and individual autonomy into the domain of social decision making. The fewest possible individuals don't get their way. Letting an ephemeral minority determine an outcome simply amounts to a random subtraction from the virtue of majority rule. More choices, perhaps many more choices, are frustrated.

Majority choice frustration happens with sortition when there are durable minorities too, of course. However, because the frustrated majority is a durable one, it is composed of voters who, even with sortition in place, get their own way most of the time. In that respect theirs seems a less serious hardship than that of the durable minority, which under majority rule is always frozen out.

The more durable the minority the stronger the argument for sortition, the more ephemeral, the weaker. Where any current voter is as likely to be in the majority as the minority by the time the next election rolls around, then this core value favoring sortition disappears altogether.

A second criterion for attractive sortition is that the minority's choice would not be too awful in the majority's eyes. In a early draft I had oompah bands and heavy metal as my alternatives to Heirloom Rock. You may well, know, however, someone who would be driven out of the party straight away by oompah and perhaps someone else who would react similarly to heavy metal. (Imagine an oompah band heavy metal fusion!) There is one neighbor who just loves Oktoberfest and there are two metalheads. Probabilities under sortition should not have these genres succeeding very often, but should they ever succeed? In my sketched case, their success might well be self defeating, as the party simply wouldn't populate. The more nearly intolerable a minority position is, the less attractive is sortition.

There has to be a caveat here. Certain kinds of awfulness ought not to count against sortition. Seating an African American on a Mississippi school board might well have been intolerable to the majority of a majority white district. In this case, however, sortition or some other mechanism, e.g. proportional voting, would be morally required. This obviously normative consideration would require a good deal of spelling out. What follows will do a little of that work.

An issue that affects a durable minority more than it affects the majority will be more sortition appropriate – assuming, at least, the differential effect is not taken into account in some other way, such as by giving multiple votes to the more affected minority. If the majority is more affected, then sortition is less appropriate.

What about the importance of the issue? It was on purpose that my party music case involved a decision of no very great moment. It is not the death penalty or the risk taking of big banks or unlimited election funding. It is a greater unfairness if a durable minority always loses on important issues, yet randomly taking an important issue away from the majority is also a greater assault on the values of individual choice and autonomy. I tend to think that as importance increases these values underlying majority rule gain weight more rapidly than those underlying sortition. This is on the assumption, again, that the degree affected is roughly the same between majority and minority. Where the minority is more affected greater importance increases the value of sortition; when the majority is more affected by an issue, a high level of importance forecloses sortition.

At this point I do not have any real argument for my contention that majority rule is preferable to sortition in proportion as issues become more important, degree affected being equal. It might well be that importance is too crude a category. Perhaps what is crucial is the way in which an issue is important.

There is a generalization of issue sortition democracy to super-majority situations. If a 2/3 vote is now required to pass a proposition, the alternative sortition procedure would be to draw three votes out of the bowl. The motion passes if and only if at least two of the ballots are marked in favor.

When both the issue and the status quo are so important that a super-majority procedure is attractive in the first place, however, a sortition alternative will almost always be unattractive. Why, if we want to guard against change by requiring a super-majority, would we accept the possibility that the issue be decided by the very good luck of a small minority? (Perhaps we could tell a story about a generally change averse society that still likes the spice of occasional unpredictable novelty, but I doubt it is going to be a story with much of a moral for political philosophy.)

As well as working out the theory and the details of the proper trade off between the values underlying majority rule and those underlying sortition, any serious consideration of issue sortition is going to have to deal with the practical problem of reconsideration. My party music example implicitly excluded the possibility that the losers could get the issue reopened. The ballots were cast; one was randomly selected; it controlled; and that was that for the coming party. In many assemblies, however, a motion for reconsideration, is in order and will prevail if it has enough support. The majority, with probabilities on its side, can be assured of winning out if it can keep trying.

So an issue sortition process requires some period during which reconsideration is off the table. This need not be the same period for all issues. It might be left to the authors of the proposition to be voted upon. Sometimes the period is all but dictated by the content of the proposition. Reconsidering the building of a bridge ¾ completed is rarely a good idea. In other cases there will be room for some games-think. Longer reconsideration ban periods give your proposition more time to take effect if you win. However, you must wait longer before trying again if you lose, and shorter periods might attract more sitters from the fence.

Postscript: With the success of our semi-monthly parties, a neighboring neighborhood adopted the idea. One of its residents designed a “smart jukebox.” Party goers would put in their choice of genre, among five, for the next set of tunes and the box's computer would select at random among all the choices submitted. (Well, pseudo-random for sticklers, but a good pseudo-random.) This neighborhood made the boast that their idea better combined sortition's fairness to minorities with a greater sensitivity to the degree participants were affected, as only those then present and then interested “cast ballots.”