In a favorite post of mine I listed some institutional reforms that might help expand the collective cognitive capacity of humanity— our ability to coordinate to solve problems and improve our lives. In this post I dive deeper into one of them: replacing electoral democracy as we know it today with randomly selected deliberative citizen committees. This idea has gained mindshare recently, but I think it is still underrated. Elections and political parties are immensely costly blights on our lives and our culture. I hope to convince you that there is a better way to achieve the legitimate and crucial goals of electoral democracy.
Why (large scale) elections are so bad
These are some classic failure modes of large-scale elections:
bad incentives for candidates to become panderers and put image over substance
political parties as engines of polarization, cognitive distortion, and corruption
governing impulses veering between establishment control and populist ferment, both of which undermine a society’s sense of self-government because both rely on a political class (whether experts or demagogues) to govern
poor levels of voter information, because voters have minimal individual incentive to become better informed and no deliberative forum where their individual voice has meaningful weight
skewed representation due to turnout problems, gerrymandering etc
I’m sure we could list many more, but hopefully this suffices to establish that election badness is a Big Problem generally, and a problem that spans generations and countries, not just something that started in the US in 2016.
Note how important the qualifier large-scale is. Elections in sub-Dunbar-number organizations or polities are far less likely to have any of these problems: think of the classic New England town meeting. But we can’t reduce our governing institutions to that scale and still have a modern civilization. We need some way of providing the benefits of democracy to polities with millions, even hundreds of millions, of citizens.
Note also that commonly proposed reforms to elections, e.g. approval or ranked-choice voting, multi-member districts, multi-party proportional representation systems, etc, do little or nothing about the above problems. That’s not to say they can’t do considerable good— in general I support these kinds of changes— but they are not substitutes for more fundamental reworking of governance.
What should we want from democracy?
To understand what alternatives to large-scale elections have to accomplish, we need to think about what democracy is good for. Of course this is a huge topic that political scientists have been arguing about for generations, and I am not a political scientist. I don’t claim to have the best possible answer, but I do think a reasonably well-informed, thoughtful citizen ought to be able to come up with a reasonable answer that sheds light on possibilities for institutional reform. So here’s my answer.
A modern liberal society, it seems to me, needs a governing mechanism that:
Provides for periodic peaceful, law-governed transitions of power that are broadly regarded as legitimate.
Subjects governing officials to popular accountability, i.e. makes them answerable to the governed populace for doing a good job as most of that populace would define “good job,” and makes it reasonably easy to dismiss those who do a bad job.
Achieves legitimacy through political equality. Political equality roughly means that among those subject to the laws of a jurisdiction, there should be no class of aristocrats with more legal power-per-person to influence government than others, and no disfavored groups systematically excluded from influence.
Also, we observe that most historical governments converge on the same general structure in practice: some combination of
executive officials with broad managerial authority over areas of governmental function, with a chief executive who can make them coordinate and settle disputes among them
legislatures to pass generally applicable laws and set budgets for agencies
an independent judiciary to try cases and interpret laws
and popular referenda for deciding on basic constitutional changes and passing laws the legislature can’t or won’t.
So if we can retain that structure; do as well or better than traditional election mechanisms at achieving the above three aims; and avoid the inherent downsides of large-scale elections, that’s a promising combination.
Sketch of an alternative
Executives chosen by jury
The first major pillar of the alternative system is that executive officials, including chief executives, heads of agencies, and other similarly powerful figures, would be appointed and reviewed periodically— perhaps as often as annually or as infrequently as every 5-6 years— by randomly chosen citizen committees. These committees would be similar in size to grand juries and aim to be similarly representative of their jurisdictions, but they would operate more like executive search committees or committees for evaluating employee performance1.
As with a private-sector search committee, the choice and review processes could include confidential interviews, research and deliberation, with the committee members voting on key steps in the process as well as on the ultimate outcome. Committees, like grand juries, likely should have subpoena power to ensure they have access to all relevant information. Candidates for executive office could propose themselves to the relevant juries or be solicited by them. The research and deliberation processes would need to be time-limited, taking weeks to a few months overall. Officials appointed by committees might be subject either to hard term limits (after X years, or upon the official reaching the age of Y, the committee must choose a new person) or soft limits (e.g. initial appointment and ordinary reappointment takes only a majority vote of the committee, but retaining an existing official after X years or after the age of Y takes a supermajority).
Service on these committees would be a significant time sink, so committee members would have to be, and should be, well paid in order to get a reasonable uptake rate of randomly selected people agreeing to volunteer— you can’t really, and shouldn’t want to, force people to serve. Moreover, the number of committee-appointed executives (as opposed to lower level positions appointed by those executives) would have to scale with the size of the jurisdiction to ensure that a typical citizen would not be selected for committee service more than, say, once or twice a decade. This is especially important because the system could not simply rely on motivated volunteers, since these are nearly certain to be unrepresentative. You’d want to ensure that along lines such as race, gender, age, and socioeconomic status a committee would typically “look like” the jurisdiction it would serve.
Advantages of this system would include:
Committee members could be chosen to be far more cross-sectionally representative of the jurisdictions than voters in a typical election.
They would have a far better incentive structure and resource environment to enable them to make well-informed and well-considered decisions, especially for “boring” positions where elections are especially subject to low turnout and insider capture.
Candidates would have to make a far more thoughtful and less mob-pandering type of appeal to a committee than to an electorate. Nonpartisan, independent, and otherwise nontraditional candidates would have a better chance.
Most ordinary people would, several times in their lives, get to participate in self-governance in a situation where their individual voice and judgment really mattered.
You could extend this system to cover agency heads which are now typically appointed by chief executives and confirmed by legislatures, increasing the direct popular accountability of those agencies.
Legislatures chosen by sortition
The second major pillar is that, just like the committees convened to select or review executives, members of the legislature could and should be chosen randomly. The nature and term of service, however, would be quite different for legislators, and the opt-out rate probably higher; nonetheless, choice mechanisms could be designed to correct for differential volunteer acceptance rates and ensure a cross-sectionally representative legislature.
To provide for frequent turnover in response to changing times but also allow for legislative tenures long enough to gain useful experience, you could have the legislators selected in multiple staggered cohorts, like the US Senate. Legislative sessions would have to be arranged to allow legislators to be part-time, as is already the case in some US states, but again, legislators should be paid well enough that taking the job would not be a hardship for almost any selectee.
Other than that, the business of the legislature would proceed mostly like that of an elected legislature. Except that the members would be nonpartisan, truly free to make decisions according to their own consciences, and truly representative of the people on whom those decisions were binding— all of which elected legislators aren’t and can’t be.
Judges and referenda
Judges, traditionally the most indirectly accountable and independent of governing positions, could be chosen by several means:
by a citizen committee like those that choose executives;
or by a traditional chief executive nomination with legislative confirmation;
or by some mixture, e.g. executive nomination with committee confirmation.
In any case the experience of modern democracies’ judiciaries gives reason to want judges chosen for long, fixed terms— longer than legislative or executive terms— but not for life tenure.
There remains the question of whether and when to allow issue-specific popular referenda as a backstop mechanism for using a mass vote when other mechanisms fail. Referenda might be required, for instance, to ratify constitutional changes passed by the legislature, or for certain cases of impeachment (see below); or they might be allowed when an exceptionally high number of signatures are collected, or when a committee or legislative process deadlocks. A referendum vote avoids some of the downsides of electing a person: it’s less likely to be partisan and its participants have a somewhat better incentive structure. But low-information voters and demagoguing pressure groups are still major problems, and the corruptions of an elected legislature which make referenda attractive as an alternative path for legislating should be less of a problem with a sortition-based legislature.
Impeachments and vacancies
All the above processes presume a regular cadence of selection and review. What about cases where a new selection needs to be made out of schedule, either due to a resignation or death or to gross officeholder misconduct warranting early removal?
For sudden “benign” vacancies in an executive position, a deputy appointed by the prior executive could step in, or the chief executive could appoint an acting replacement, until a special committee could be convened to choose a successor out of schedule. Similarly, a legislator who resigned or died could be replaced by a one-off lottery selection.
Impeachment for misconduct is trickier, but variations on several traditional mechanisms could be used, not mutually exclusive:
A supermajority vote of the legislature could immediately remove an official…
… or cause a special committee to be selected to decide on their removal and/or replacement, maybe with a supermajority committee vote requirement…
… or a referendum, or even just the collection of enough popular signatures, could establish such a committee.
Personally I would favor all of these being available options, so as to allow for both impeachments requiring immediate action and those requiring greater deliberation while still setting a high consensus bar for making such a decision.
Details that need more fleshing out
First, let’s do some back of the envelope math on the burden of committee service in this system. Suppose that in our polity one in 1000 adults is a governing official (including all local, state, and federal officials) of the type that would “normally” be elected. And suppose that officials are chosen and reviewed by a committee of 20 people (i.e. about the size of a grand jury) every two years. Then if I’ve done the math right, that means 1% of the adult population would be serving on committees at any given time— and if typical committee service takes about a month, you’d expect a typical adult to spend that month on a committee about once every eight to nine years. This seems reasonable: infrequent enough not to interfere with other life projects, but frequent enough that almost everyone would really feel they got their chance to participate in the business of government.
Paying 1% of the population continuously to do this service at a middle-class wage is a significant expense, but:
seems worth it if it produces a large improvement in quality of government leading to a sustainably higher economic growth rate, and
might function as a sort of stochastic UBI/job guarantee, partially offsetting the cost of other income support/redistribution programs.
I do worry that maybe I’ve done the math wrong or my assumptions are off by an order of magnitude, though.
You might also wonder if this really works at federal scale, e.g. whether having 20 randomly selected people from all across the US choose the President is really a good idea. A variation that might address this would be to have more local-level committees choose representatives to larger committees— something like the Electoral College as it was originally envisioned, hopefully with better representativeness. Alternatively or in addition, some especially high-stakes committee decisions might be cross-checked or ratified by an independently chosen second committee, with or without an “appeals court” sort of structure, as is sometimes done for corporate performance evaluations today.
Should committee votes be open or secret ballot? Legislative votes are open, and these committees are somewhat legislature-like; but secret ballots are better for allowing people to vote their consciences. Both have advantages and disadvantages as far as avoiding corrupt influence on committee members. It may be that votes should be open within the committee but confidential relative to the public, as is the case for trial juries.
Other points that probably need more thought:
Degree of allowable delegation of rulemaking authority by the legislature to executive agencies. This is already a controversy in our system but probably has different implications with a sortition legislature and committee-chosen executives.
Interactions between a chief executive and agency heads who might normally be chosen at the chief’s discretion but are now chosen by committee instead. How important is it for a chief executive to get to pick “their team”?
Opt-out norms and eligibility qualifications for committee service. Do new parents get an automatic deferral, or people in certain demanding jobs, etc? What mental disorders or limitations, criminal records, conflicts of interest, etc would be disqualifying?
What could go wrong?
As with all large-scale governance systems, this one attempts to balance independent expert judgment with trust in ordinary citizens, and in many respects moves radically toward the latter. Experiments would be needed to judge whether this changed balance is viable.
A lot depends on the composition of committees and legislatures actually being random and representative. There would be tremendous incentives to manipulate this, and an opportunity for corruption in that the legislature, with judicial review, would set the rules for its own selection. But this is a problem for our present system too— election process integrity is hard and crucial, and it’s not clear that lottery process integrity is any harder.
There are probably other lessons to be learned from the successes and failures of the trial jury system, which is by far the closest analog to this proposal in our present polity. Some of them are positive: citizen committees would not be subject to the distortions of adversarial voir dire or the artificial rules of evidence necessary for trials. Others might be negative: some of the strategies lawyers use to manipulate juries could transfer to interest group manipulation of citizen committees. Certainly some committees would be just plain dumb, as are some juries— but so are some electorates, and again, there’s no obvious reason the madness of crowds would be any worse in this system, and several reasons to think it would likely be less bad.
Further reading
Some books and articles that have influenced this proposal include:
The case for abolishing elections, a solid introduction to sortition based governance
The wisdom of small crowds, a primer on modest uses of small citizen juries
On cultures that build, excellent on the practical import of self-governance as a cultural value and habit
10% Less Democracy, on how governing officials often make better decisions when they don’t have to stand for election
The Great Covid Panic, contains controversial claims about Covid, many of which I do not endorse, but the key part here is their case for citizen juries at the end
I served on many such committees at Google, and for all their artificiality and expense, they were among the most consistently fair-minded, deliberative decision-making bodies I’ve ever seen.
This is one of the best outlines I've seen for how Citizen Assemblies could be integrated with the system the US has today.
The big question it leaves me asking is "how do we get from here to there?". Moving to Citizen Assemblies seems like a big step, so what is the most plausible sequence of small steps that gets us there?
Thanks, Nicholas!
Two more possible application for citizens' committees chosen by lot from the general citizenry or from a pool of volunteer applicants (with apologies if either is covered somewhere within your post, which I just now skimmed):
1. Redistricting: drawing legislative district boundaries. (There's a list of US states that are already doing this, labeled with the "Non-politician commission" category in the "State-by-state procedures" tables at https://ballotpedia.org/State-by-state_redistricting_procedures)
2. Making decisions or proposals regarding hard problems and other, similar "deliberative tasks." (See Joe Klein's classic piece on this, in Time magazine back in 2010, https://content.time.com/time/magazine/article/0,9171,2015790,00.html)