Some Reflections on Sortition


That capitalism needs to be replaced is obvious. We need a system that values human needs, rather than profits, that lets people control their own lives, rather than being dominated by tyrants or by capital. But what should our good society look like? 

For many on the left, this question is dismissed as premature at best and arrogant and elitist at worst. But how can we convince others to join us in the struggle for a new world if we can’t, even in broad strokes, tell them what the goal of our struggle is. One of the most powerful arguments in favor of the status quo is that “there is no alternative.” Unless we can demonstrate that society can be organized to realize our values, it will be impossible to convince people to make the commitment and sacrifices necessary to challenge that status quo.

Certainly, it would be elitist and downright dictatorial for a few to impose a vision of the future on others. But reflecting about our vision isn’t the same as imposing a vision. It is simply an insistence that along with thinking about questions of political strategy, the left has to give thought to clarifying its goals. While it would be presumptuous to offer a detailed blueprint of a new society, if we can’t describe our vision in enough detail to show its desirability then it would be presumptuous to expect people to fight for it. 

Thus, the fact that there has been a renewal of conversation about the socialist vision in recent years is most welcome. There has been valuable discussion of the economic, political, and other aspects of a good society. One suggestion for the political system we need that has been getting increased attention lately is the idea of democracy by lottery, or, to use the technical term, sortition. Sortition, it is said, offers a way to build a better society, avoiding some of the objectionable features of representative democracy. 

We should note that sortition has long been part of the history of democratic thought. The ancient Greeks used it. Political theorists James Harrington, Montesquieu, and Rousseau wrote about it. The Trinidadian socialist C.L.R. James praised it in his 1956 essay “Every Cook Can Govern.” 

It is possible to use elements of sortition in a political system that is largely organized around other principles. In the United States today we have the institution of the jury, which draws a randomly selected cross-section of the population to decide criminal and civil cases. There are problems with how U.S. juries work—racial discrimination in jury selection being one of the most significant — but the basic principle that it is appropriate to have a random cross-section of the people making these sorts of decisions is well established. Various theorists have proposed using sortition to choose one house of a bicameral legislature, or to provide various supplementary advisory bodies, or to democratize unions. Various experiments have been done with sortition for planning and budgeting and addressing tough issues around the world. In an essay I wrote some years ago proposing a participatory political system, I made use of sortition as a means of choosing the members of judicial bodies so as to avoid “the counter-majoritarian dilemma.” Xtinction Rebellion, the environmental direct action organization that began in the UK, has as one of its key demands that the government create and be led by the decisions of a randomly selected Citizens’ Assembly on climate and ecological justice.

There are a multitude of ways in which sortition might be used and incorporated into the workings of a good political system. But what I would like to consider here is the wisdom of using sortition as the main structure of a new society. Some contemporary writers have urged sortition as an alternative to elections. So, should sortition be the basis for our future socialism? 

Let us consider some of the values we hope to realize through a better political system.


One value is participation. Obviously, we want a political system that produces results that benefit us—policies that enhance our lives—but that’s not enough. If good results were sufficient, then we would be happy to be led by a benign and omniscient dictator or a computer algorithm that used artificial intelligence to maximize well-being. (Imagine that algorithm from Amazon that delivers book suggestions to you or the one from Google that tailors ads to your preferences.) But even if such benign dictators or algorithms were possible, they have an inherent limitation. Politics is not just a means of attaining our ends but is also a means of defining who we are and hence what our ends are. No political system is likely to produce results that benefit us unless there is some means of knowing what it is that benefits us, and this is not given, but emerges only after we interact with and deliberate with other people. Our participation helps to define and create our preferences, which is what the political system is seeking to address.

To be sure, political participation can be overdone. Today’s political activists cannot assume that everyone has the same predilection for politics—for meetings, for debates, for reading about politics—as they do. Just as people vary in their preferences and capacities for music or crafts or mathematics, so too will they vary in their attitude toward and talent for politics. So, we don’t want a political system that requires everyone to value political participation as much as current-day full-time political activists do, or that penalizes those without a flair for or an interest in politics by somehow denying their interests equal consideration. But some degree of participation—less than that of political fanatics, but far, far more than that of most citizens of capitalist democracies—is essential.

Standard capitalist democracies with representative systems are seriously deficient in terms of providing opportunities for participation. Voting once every two or four years is an incredibly minimal level of participation. The percentage of the population that goes to a rally, writes a letter to the editor or to Congress, or contributes money to a campaign is tiny. Representative systems tend to discourage participation. Why spend a lot of time learning about some issue—indeed, why even bother to vote—if the politician who ran on a platform promising to take one position, does the opposite when in office? Lyndon Johnson’s “No Wider War” pledge or George H.W. Bush’s “Read my lips: No new taxes” are famous examples, but more telling have been the ubiquitous pledges to “help the little guy” over the past four decades, while presiding over the largest transfer of wealth in history to the super rich. And any voters who bothered to follow politics closely and decided to punish their representatives for failing to represent his or her constituents find that they can’t get rid of the scoundrels because incumbents are nearly invulnerable. Thus, representative systems produce extremely strong pressures discouraging participation.

Referenda and initiatives were welcomed in many U.S. jurisdictions as a way to make representative democracy more participatory, but these have by no means solved the problem. Voter turnout in the United States is low in all elections, but it is even lower on referenda. And money is just as important in these contests as in other elections. And where gathering signatures to get a question on the ballot might have been an opportunity for people to talk to each other, signature-gathering is increasingly being done by private companies, paid for by big money. 

So how does sortition fare on the metric of participation? If the people who are going to be deliberating on the issue are a small group of people chosen by lot from the population, then only that small group will be participating. In a representative system while the representatives do the bulk of the deliberating, there is some small incentive for the voters for those representatives to do some thinking about the issues. But under sortition, for the bulk of the population there is not even this limited incentive to involve oneself in the issue. Thus, to the extent that we want a political system that incentivizes participation, sortition doesn’t do very well.

The sortition legislators will come into contact and deliberate with a cross-section of their fellow citizens. Through this process they will come to better understand people in different circumstances and appreciate the ways in which all their interests are intertwined, a crucial component of building feelings of empathy and solidarity. But this crucial process only takes place among the small group of people who have been randomly selected from among the population to be legislators. The benefits of deliberation, of building empathy and solidarity, accrue to the randomly selected few rather than to the many. 

On the other hand, a system built on participatory structures, with councils in which all citizens are involved, would allow these benefits to be shared by everyone.


A second value that we care about in a political system is accountability. That is, we want a political system where the people maintain their control over the system and can assure that the system continually reflects their interests. Ideally, if the people change their minds about what policy best meets their needs, then policy should change as well. Likewise, if people discover that those who are making decisions on their behalf no longer have their confidence, then those decision makers ought to be replaced. Thus, the principle of recall has long been a fundamental one for leftists. As Engels wrote in his postscript to The Civil War in France, the Paris Commune—“a new and really democratic state” — filled all posts “by election on the basis of universal suffrage of all concerned, with the right of the same electors to recall their delegate at any time.” 

In a representative system like that of the United States, there are various mechanisms that are supposed to achieve the accountability of representatives. The chief mechanism is reelection. Representatives who diverge substantially from the wishes of their constituents—whether because the voters have changed their minds or the representative has shifted position—face the possibility of not being reelected at the next election. This mechanism doesn’t work for presidents or most governors in their final term, or state legislators in their final term in the 15 states with term limits. For other officeholders, though, the prospect of losing the next election is a potential check on representatives’ views differing too much from those of their constituents. But there are several reasons why this mechanism for assuring accountability doesn’t always work. First, as already noted, it is hard to dislodge incumbents. Second, the elite domination of the media that skews elections in the first place also makes it hard to really know what one’s representatives are doing. Third, voters’ memories are short and what a representative did two, four, or even six years ago may well not be remembered on election day. And, finally, even if the representative does lose the next election, he or she still has had a long time to conduct mischief. (Several years of war-mongering or climate-destruction may well be irreversible.)

Recall is a much stronger mechanism for ensuring accountability, since there is no need to wait until the end of the term to replace the unrepresentative representative. In the United States, there is no provision for the recall of federal officials. (Impeachment is a different process, for this allows removal only in the case of corruption—high crimes and misdemeanors—not whenever the official’s views diverge from those of his or her constituents.) Twenty states have laws allowing the recall of some state and local officials, and another 19 allow recall of just some state officials. Typically a certain number of signatures triggers a special recall election. Gray Davis, for example, was recalled as governor of California in 2003, a few months into his second term.

How would this work in a legislative body chosen by sortition? Consider three cases. 

  1. Imagine that 55 percent of the population supported position X on a particular issue, but in the sortition legislature—because it was chosen at random—only 48 percent supported position X. The legislature is thus not an accurate reflection of the political views of the population.
  2. Imagine that a majority of the population takes position Y on an issue as does a majority of the legislature. Say that because of new developments, the population or the legislature—but not both—changes its view so that the two are no longer in alignment. Again, the legislature is not an accurate reflection of the political views of the population.
  3. Imagine a legislator whose voting behavior seems to be motivated by selfish considerations rather than the common good (perhaps voting on the siting of community amenities near the legislator’s own home). After a number of such votes, some members of the population lose confidence in the disinterested judgment of the legislator.

In case #1, it is unclear how recall could help bring the legislature into alignment with the popular will. Who would be recalled? In case #2, a new legislature could be selected—though that means that the learning curve of the new legislators would have to start over. And who would be authorized to call for the new selection? And in case #3, who would make the judgment that the legislator was compromised? It would seem odd to say that the population will vote on it (after a certain number of signatures are obtained), given that the population didn’t vote for the legislator in the first place. 

In general, the problem seems to be that while we understand what it means to say that legislators should be accountable to the constituents who elected them, what does it mean to say that legislators should be accountable to the group from which they were selected at random? A group of voters becomes a community of sorts when they engage in the common activity of participating in an election campaign. But does a group of people become a community by the mere fact that legislators were chosen randomly from among them?

Note that most sortition proposals suggest that there not simply be one big random sample chosen from among the population, but a stratified sample where separate random samples are drawn from relevant demographic groups, and maybe also geographic and attitudinal groups. This assures that subgroups are adequately represented, but it does raise the question of whether a black, female, cisgender, southwestern, pro-capital punishment legislator, for example, “represents” the society or only the black, female, cis, southwestern, pro-capital-punishment members of the population.


Another problem for sortition is the question of legitimacy. Any political system has to not only make political decisions, but have those decisions be broadly seen as legitimate by the population, including by those who supported the losing side. If the losers take up arms whenever they lose a vote, the political system has failed in a fundamental respect. 

In standard representative governments, the votes for the representatives and the votes by the representatives are conducted using majority rule. Many have been critical of the majority rule principle—why is 51 percent more morally acceptable than 49 percent? 

Some have urged that we use consensus decision making instead of majority rule. Indeed, working toward consensus is a good idea wherever possible. However, insisting on consensus in every case gives any individual the power to block the overwhelming majority. Such an approach is ill-advised. It is sometimes said that even a large group should be forced to respect and acknowledge the sentiments of a single dissenter who feels strongly on an issue. Respect and acknowledgment are fine; but the question is whether the strong feelings of the lone dissenter should invariably be able to block the equally strong feelings of everyone else? People should talk, they should debate, they should respect the moral seriousness of each other, they should try to find some areas of common agreement. But if, at the end of the day, they cannot reach consensus, a vote, decided by majority rule, is the only just option. To allow the lone dissenter to block action is to deny the overwhelming majority ultimate authority to decide their own fates. Even to allow a much larger minority to block the majority would mean that we are assigning more moral weight to that minority than to the majority. But the fundamental principle of equality means that everyone is entitled to the same moral weight. There is nothing magical about 50 percent plus one, but it does deserve more moral weight than 50 percent minus one.

In a standard representative government, decisions are said to be viewed as legitimate—even by those who were outvoted—because of the principle of majority rule. Socialists, of course, understand that capitalism undermines democratic legitimacy in multiple ways: There is the impact of money on elections, the power of the corporate media, and the threat of the capital strike. But in a society that had eliminated inequality in wealth and economic power, majority rule truly provides legitimacy. In a socialist society with participatory structures rather than representative ones, it would still be the case that majority rule—this time in the councils—would provide legitimacy.

But what about in a sortition system? Will there be legitimacy for the decisions made by a randomly chosen legislature? 

In particular, random processes always have sampling errors. On a close vote in the legislature, it is possible that the margin of the vote is less than the sampling error. This means that we can’t be certain—or even 95 percent sure—that a majority of the population (with appropriate deliberation) would reach the same decision as the majority of the legislature. This will make the decisions of a sortition legislature potentially much less legitimate than that of a body that more accurately matched the views of the population.


That modern representative legislatures are far from representative is well known. Women, minorities, the working class, the poor—all these folks are seriously under-represented in the legislatures of the United States and other capitalist democracies. 

With participatory structures, the complexion of decision-making bodies would be far more “representative” because everyone would be involved. But note: This would not be the case if we simply used workers’ councils. A society that depended on such councils alone would leave retirees, homemakers, and those whose health precluded participation in the workforce disenfranchised. Thus, there have to be neighborhood councils or consumer councils as well as workers’ councils to ensure that all are represented.

With sortition, where legislators were chosen randomly from the population, the make-up of the legislative bodies would mirror that of the population. But there is a problem here. Would everyone chosen in the random procedure be required to serve? With juries, participation is compulsory, but for many the potential loss of income is considerable and they do what they can to avoid it. Presumably, adequate compensation would not be an issue for sortition legislators in a socialist society, but many people will still not be eager to take on the role, a responsibility that lasts far longer than jury service. As noted previously, not everyone shares the enthusiasms of political activists, not everyone wants to leave the careers they’ve chosen, not everyone wants to be in the spotlight of societal decision making. (We can remove the spotlight by making sortition legislators anonymous, but then accountability is reduced even further.) How well would a legislature function if it were filled with people who didn’t want to be there?

The alternative to mandatory service is to have the selection process only choose from among those who volunteer. But then we have the problem of selection bias: Our legislative body will over-represent those with self-confidence, speaking ability, and education.  

One can try to avoid this problem by using stratified sampling as a way of guaranteeing that those selected match the population demographically. But on which variables would one stratify? Race and gender and sexuality? Occupation? Political views? One thing we probably couldn’t eliminate from our self-selecting volunteers would be the over-representation of those who see themselves as leaders.

Issues Versus Agendas

Many advocates of sortition draw on the studies of James Fishkin, who has shown that a randomly selected nonexpert group of citizens, given access to expert testimony and an opportunity to deliberate, can come up with decisions that are thoughtful and more in tune with progressive values than one would expect from public opinion polls that capture people’s off-the-cuff views. Just as jurors, who receive explanations of the relevant points of law by the judge and perhaps hear from expert witnesses, can deliberate and then come to a verdict in a case, so too nonexperts can make sound political decisions on other issues as well. But there is little evidence showing that a randomly selected group of people is well-suited to agenda setting. As two scholars note, “there appears to be no precedent for a sortition body that has such an expansive mandate ranging from agenda setting to scrutiny. Specifically, we cannot find any examples of the effective use of sortition bodies to undertake agenda-setting functions.”

We can imagine that citizens randomly selected to deliberate on an issue—like climate change—could be provided with expert advice, including from contending sources, without giving the experts control over the decision making. But it is much harder to imagine how expert advice could be provided on agenda setting. The necessary information for setting an agenda is aggregated by political parties, something missing from a sortition legislative body.

Executives and Administrators

Diverse groups are often more effective decision makers than smart people. That’s the logic of juries and it is the logic of democracy in general. But what works for legislatures does not necessarily work for executive and administrative functions. 

Ancient Athens filled many positions by lot, but it didn’t do so for the commanders of the army and the fleet; they were chosen based on their military skills and knowledge. But other positions within the military (for example, archers and cavalry) were not assigned randomly either. In any case, it’s hard to believe that one could use random assignment to place the millions of people who today work for governments: Those who work for the public health service or the army corps of engineers or the IRS do not have interchangeable skill sets or training.

It is true that the division of labor in advanced capitalist societies is highly problematic. It concentrates the decision making in the hands of the few, provides opportunities that enhance advanced skills for a limited number of people, and consigns most people to jobs that make them mindless. As Adam Smith noted two and a half centuries ago, 

The man whose whole life is spent in performing a few simple operations, of which the effects too are, perhaps, always the same, or very nearly the same, has no occasion to exert his understanding, or to exercise his invention in finding out expedients for removing difficulties which never occur. He naturally loses, therefore, the habit of such exertion, and generally becomes as stupid and ignorant as it is possible for a human creature to become.

But the answer to this problem is not to assume that jobs do not require skills and training, but to make sure that everyone has a mix of empowering jobs and unempowering jobs.

Political Organizations Today

Does it make sense for left political organizations today to use sortition as a way of choosing their leaders and/or their decision-making bodies?

There is no doubt that many left organizations (like many other organizations) have an encrusted leadership that, even if elected, becomes separate from the membership, developing different interests, and over time comes to see challenges to its continuing control as subversive. There are various mechanisms available to address this problem: term limits, prohibitions on using organizational funds in elections, rotation, making more positions elective, implementing quotas for leadership positions, having multiple-person teams fill leadership slots, among others. Would sortition add something to these measures?

Part of the logic of sortition is that the laws of probability assure that a random sample of an adequate size drawn from a large-enough population will tend to mirror the characteristics of the population. Alas, for most left organizations the population size is too small to guarantee a very representative sample. After all, we don’t want to just change the leadership, we want to do so in a way that accurately reflects the views of the membership.

Another problem is that, except for organizations to which all members have decided to devote their whole lives, political groups tend to have people willing and able to commit different levels of time and energy. Whatever other problems plague standard methods of leadership selection, one problem they do not have is that they can be sure that those who choose to be leaders and run for those positions have the time for and interest in serving. A leadership team chosen at random from avid members and marginal members would not be very effective.

Assessing Sortition

I have reviewed a number of reasons to question whether sortition would be a good structure for the political system with which we hope to replace capitalism. I have suggested that in terms of participation, accountability, legitimacy, representativeness, and agenda setting there are serious questions about sortition. I was also skeptical of its use as the means of choosing people for executive and administrative tasks and of its use in left political organizations today. But this is not an argument against incorporating some aspects of sortition into a larger participatory vision. 



  1. CLR James “Every Cook can Govern: A Study of Democracy in Ancient Greece.
  2. See, for example, Equal Justice Initiative, Illegal Racial Discrimination in Jury Selection: A Continuing Legacy (Montgomery: August 2010).
  3. See, for example, John Gastil and Erik Olin Wright, eds., Legislature by Lot: Transformative Designs for Deliberative Governance (Verso, 2019).
  4. Simon Pek, “Rekindling Union Democracy Through the Use of Sortition,” Journal of Business Ethics (vol 155, 2019), 1033–1051.
  5. ParPolity: Political Vision for a Good Society,” ZNet, Nov. 22, 2005. See also Noah Briggs, “It’s Time For a People’s Court,” New Politics.
  6. Extinction Rebellion, “Our Demands.”
  7. For example, Oliver Dowlen, The Political Potential of Sortition (Exeter: Imprint Academic, 2008); Brett Hennig, The End of Politicians (London: Unbound, 2017); David van Reybrouck, Against Elections (London: Bodley Head, 2016).
  8. 1891 Introduction by Frederick Engels, On the 20th Anniversary of the Paris Commune.”
  9. Term Limits in the United States.
  10. Laws Governing Recall.
  11. Say you had a legislature with about 1,200 legislators randomly selected from the population. As is well known from the mathematics of polling, the legislators’ views on a particular issue will diverge from the views of the general population by more than 3 percentage points only about one time out of twenty. But if we have fifty issues, then the likelihood that the legislators’ views diverge by more than 3 points from the population’s views on at least one question is greater than nine out of ten [1-(.9550) = .92] And with smaller legislatures, the divergence would be greater.
  12. I omit here the possibility of more blatant corruption. To be sure, capitalism’s professional political class is deeply corrupt. But the potential for corruption would also be strong in a system where a randomly selected citizen can make a decision that affects lots of people. As Tom Malleson notes, corruption is a danger because “sortition members would not have the standard restraints of accountability to constituents or to a political party, which can discipline individual members who threaten its collective reputation.” (“Should Democracy Work Through Elections or Sortition?” Politics & Society (vol. 46, no. 3, 2018), 401–417, 405.)
  13. Note that if a supermajority is required instead—say a two-thirds vote—then the question remains: Why is 67% so much more morally acceptable than 66%?
  14. For more discussion of majority rule and consensus, see my “ParPolity…,” cited in note 5 above.
  15. The Voice of the People (New Haven: Yale University Press, 1995).
  16. David Owen and Graham Smith, “Sortition, Rotation, and Mandate: Conditions for Political Equality and Deliberative Reasoning,” Politics & Society (vol. 46, no. 3, 2018), 419–434, 425.
  17. The Wealth of Nations (Modern Library, 1965 [1776]), 734. 
  18. See the Parecon model developed by Michael Albert and Robin Hahnel. Michael Albert, Parecon: Life After Capitalism (Verso, 2003), chap. 6; and Robin Hahnel, Of the People, By the People: the case for a participatory economy (Soapbox Press, 2012), chap. 11.
About Author
STEPHEN R. SHALOM is on the editorial board of New Politics.

If you’ve read this far, you were pretty interested, right? Isn’t that worth a few bucks -maybe more?  Please donate and  subscribe to help provide our informative, timely analysis unswerving in its commitment to struggles for peace, freedom, equality, and justice — what New Politics has called “socialism” for a half-century.