Strategies to implement sortition?

I’ve been thinking recently about strategies to implement sortition – especially incremental strategies that start with relatively small, feasible steps, and then gradually build up to more ambitious goals. I have a special interest in strategies that could be feasible on a municipal level in or near the city where I live (San Francisco), because I’d like to be part of such a project, but I’m also very interested in any strategies people have suggested that might work in any part of the world.

Harald, you recently proposed two strategies:

  1. Add a small number of allotted members to an existing elected legislature, then increase the number of allotted members
  2. Conduct one-off sortition exercises (with the allotted bodies having some decision making power) on first a few issues, then more

Kevin O’Leary proposed starting by creating an allotted legislative body (the “People’s Assembly”) that would be advisory only, then later giving it legislative authority.

I’ve been thinking about three other strategies:

  1. Conducting a foundation-funded, local government supported experiment with a municipal level second chamber chosen by lot (in US cities, the legislative branch is usually unicameral).
  2. Creating an actual allotted second chamber in a municipality, with legislative authority; then doing the same in a few more municipalities, then replacing one branch of a state legislature with an allotted chamber in a small state or province.
  3. Creating a single purpose allotted legislative body (say, on health care or environmental legislation), then adding more.

What other strategies do you folks know of?

24 Responses

  1. Hi David,

    Devolution of power through a multi-year trial is an option you may wish to consider. We are working with a council in Sydney who will use a randomly selected group of 45 to set service levels and funding models for the council’s next 4 year operations of plan.

    This was achievable because the councillors reached a point where they saw greater community credibility in this decision making model than normal forms of engagement – and also the benefit from de-politicising the process. It was wryly observed that if you mow the parks every fortnight people complain about the rates, and if you mow them every 6 weeks he’ll get 200 letters about why the grass is long. But if the community sets the level of service and how its paid for he has a happier citizenry and fewer angry letters to deal with!

    One of our assets we bring to the table is a creedentialled group of retired politicians – allowing a local council to point to the Foundation and say “they designed this process, not us” and expect to receive a greater level of public trust.

    Happy to chat further if you want to get in touch.

    newDemocracy Foundation


  2. I am somewhat skeptical of advisory panels. I think people behave quite differently when they know their decision matters, than when they are unsure. And an advisory deliberative assembly will probably be dominated by those who most strongly believe they have actual power, optimistic idealists.

    It is better to give an assembly full power over a limited issue, than dubious power over many, I think.


  3. Iain, these are some useful ideas (multi-year trial; finding a situation in which elected officials benefit from sortition; and having a credentialed group of retired politicians). I’d love to talk with you more.

    Harald, I’m not going to argue about whether it’s better to give an assembly full power over a limited issue, versus starting with an advisory-only role on a range of issues. Instead, I’m going to suggest four dimensions (not mutually exclusive) along which is would be possible to implement sortition in an incremental way:
    • political units – from one small municipality to more, to a province or state, to a nation, etc.
    • issues – from one issue, to more, to all
    • degree of power – from advisory only, to limited powers (like forcing votes on bills that are stuck in committee, and veto power), to full legislative power
    • number of members – as you’ve suggested, add a few allotted members to an elected legislature, then increase the number of allotted members


  4. Sure David – drop me an email at and we can set up a time to talk.

    I actually agree with Harald as well though. One of our mandatory requirements of representatives is that they agree a pre-defined level of influence that will be ascribed to the assembled panel – the decision is binding and will be enacted.


  5. In the thread “Which actors for each activity?” I note that there are too many fetishes for two or more chambers per legislature. There should be a much, much larger, but unicameral, legislature: (

    Such a large legislature, spread out over multiple cities, should make China’s National People’s Congress look so miniscule.

    The definitive immediate or intermediate step towards having full sortition are random balloting and mixed-member, “party”-based proportional representation (MMP), like what the German Bundestag has currently. After all, Probability-Proportional-To-Size Sampling fits both PR and statistical sampling as the basis of random sortition. Anyway:

    1) With the “party” basis, there is potential for mandatory random selection or at least (somewhat) random balloting of officeholders by electoral “parties” and political parties themselves. Full effectiveness here means a PR system that is “party-recallable” and closed-list; outside-legislature “party bureaucracies” should have the power to yank and parachute legislators at will.

    2) Deductions are made from “party” totals so as to maintain overall proportionality. This is a crucial feature of MMP.

    3) Thresholds for “party” entry should be reasonable. 5% is too high, and 1% is too low.

    4) One or more chunks of this large legislature could be (s)elected on the basis of single-winner districts / geographic constituencies. As much as possible, the Condorcet methods (like the Schulze method) are to be applied. Random balloting, which would in fact be conducive towards Condorcet methods, is most relevant to this chunk or these chunks. I don’t know whether plurality/first-past-the-post or instant runoff should be kept, because they tend not to produce Condorcet winners.

    5) One or more chunks of this large legislature could be (s)elected on the basis of one or more forms of single transferrable vote (STV), with the basic being Droop Quota and more complex being Schulze STV and CPO-STV.

    6) One chunk of this large legislature could be demarchic regardless of whether “parties” employ random selection of their candidates.

    7) One or more chunks of this large legislature could be (s)elected on the basis of various forms of open-list PR (D’Hondt, Sainte-Lague, etc.). The “party-recallable” feature should, as much as possible, apply here.

    8) The rest of this large legislature should be (s)elected solely on the basis of party-recallable, closed-list PR. If ordinary voters wish to have influence on who gets (s)elected here, then they should apply for organizational membership.


  6. Looks awfully complicated, and a “much much larger body than China’s National People’s Congress” sounds to me like a recipe for rational ignorance. Here’s an opposing view for any other simple-minded parliament fetishists like myself:


  7. Did you read the Q&As on that website? That was just the tenth question. Here’s the third one:

    I hope that one deals with the subject of “rational ignorance.”

    Anyway, the core of my post are #1, 2, 3, and 8, and also the paragraph above 31.


  8. Jacob, any chance of a short summary of how one vote in 30,000 doesn’t invoke the rational ignorance theorem? (sounds a bit like the miracle of the feeding of the 5,000) — I confess I don’t have the time to go off-site for background reading.


  9. I agree with Harald.

    Peter suggested a while ago, I think, starting with promoting sortition as a decision making tool in areas where elected officials have a clear conflict of interest, such as determining the salaries of elected officials. This may be a good starting point, but it does run the risk of relegating sortition to a niche, rather than being an effective beachhead.


  10. One vote in 30,000? I’m confused. The site isn’t advocating a 30,000-member Congress, but a Congress with one rep per 30,000 voters. Please summarize the rational ignorance theorem as it pertains to critiquing my proposal.


  11. In the case of one rep per 30,000 voters (or anything in that order of magnitude) voters have no incentive to do the necessary tedious and time-consuming work, studying the issues and candidates, as their individual vote has negligible causal power — the result would be the same whichever way they voted. Thus voter ignorance is deemed rational (from a public choice perspective). This is the prime argument for sortition as it reduces the numbers to the level where every vote counts — usually reckoned as being in the low hundreds. But you already know this, so I’m not sure why you’re asking me to restate it.


  12. Yoram,

    You wrote, “Peter suggested a while ago, I think, starting with promoting sortition as a decision making tool in areas where elected officials have a clear conflict of interest, such as determining the salaries of elected officials. This may be a good starting point, but it does run the risk of relegating sortition to a niche, rather than being an effective beachhead.”

    Good point. On the one hand, Peter’s suggestion sounds like a very promising way to start, as does Iain’s experience in Sydney. On the other hand, there’s the strategic question of how to keep these experiments from relegating sortition to a niche.

    (By the way, just for the record, I too agree with Harald’s point that giving an allotted body real decision making power is a big step forward from making it advisory only.)

    Iain, are there plans to expand the use of sortition in Sydney if the project you describe is successful?

    Does anyone here know of any experiments with a general purpose allotted chamber? I’ve seen posts about parties proposing this in Spain and Canada and Wales, but has anybody gotten to the implementation stage?


  13. To implement a stategy on a large scale and foreign to the general public’s tradition is in my view a major undertaking requiring a lot of energy and money. I believe the best way of introducing a new system to the public is on a tiny scale that can educate the masses as it grows. I live in a small village of 110 families in Belize and am educating them one at a time. There is corruption here but if I can convince enough of the families to ramdomly select three of the 7 council members, it would be a start.


  14. Yes good point. But does your village council exercise real power? If so, then it would be an interesting experiment to introduce allotted councillors. And what would you anticipate would be the outcome regarding corruption — surely allotted councillors would be easy to corrupt after they were appointed (arguably more so than elected councillors, as they would not have the need to seek re-election as a way of regulating their behaviour in office)?


  15. newoutofthebox,

    This is interesting. What is the current situation of the council? Is the corruption widely recognized? How difficult is it for an average person to get elected? I would suspect that with such a small group, creating electoral alternatives to politicians who are widely perceived as corrupt should not be too difficult.

    Please keep us up to date with your efforts.


  16. I’m wondering about what would be a good size for an allotted second chamber in a city. I imagine that you’d want a big enough number of members to be a good sample of the whole adult population, but small enough to facilitate good deliberation and to contain the cost of implementation. And I’m also wondering about what would be a good term of office, and about how much time a member of the AC would have to devote per week.

    Have these subjects been discussed here already? And if so, could someone direct me to the right posts?


  17. I think a good strategy to get people familiar with sortition is to look for opportunities to adopt it outside of government. So for example in the governance of sporting bodies and clubs. Or in the selection of corporate committees.


  18. David,

    Regarding size, the considerations that you mentioned are the primary ones, I think. I think ~200 seems reasonable in the sense that 20 is too small and 2,000 too big. This is also the same order of magnitude that is customary in parliaments today.

    Regarding length of term, the term has to be long enough to allow and motivate forming an agenda and a thorough understanding of the needed policy changes and short enough so as not to allow the formation of elitist ideas and interests within the allotted chamber. Again, I don’t see any reason to change the customary 4 years (or so).

    We discussed these issues before in comments, I believe, but I don’t recall any specific posts.


  19. In looking for non-governmental opportunities for sortition, it is crucial that the members in the lottery have a genuine concern about the health of the organization, and will be significantly affected by decisions of the organization. For example, many organizations have a core of dedicated members and a huge number of peripheral members who wouldn’t agree to serve. Sortition may not work well in that situation. Organizations where members have a sufficiently deep interest in the future of the organization might be employees of a business, or members of a union, or a geographic entity (where “exit” is not an easy out).


  20. Yoram,

    I would think that allocated chambers have one size criterion to meet that elected chambers don’t. Since one of the purposes of an allocated chamber is descriptive (“mirror”) representation, it seems to be especially important to have enough members to be a statistically valid sample. Any idea what that range would be? Has that been discussed here?


  21. The point Terry is making is good.

    In general, different situations imply different considerations on the people involved. Sortition happens to be particularly well suited to the situation of the state. Other situations (localities, special interest organizations, workplaces, etc.) may require quite different mechanisms, or at least require a significantly different design of the sortition mechanism.


  22. David,

    Right – that is the constraint from below. We had some discussions of the representativity of a chamber with 500 people – in the comments thread here.

    As the sample size drops, larger variations from the expected become more likely. If there is a minority group in the population whose proportion is, say 30%, there is a 1.7% chance that it will have a majority in a chamber of 20 people.

    This may not sound like much, but this is just the chance that a given specific 30% minority (say, college graduates) would gain a majority in the chamber. There are a whole lot of groups in the population and we would like to have each of them represented approximately according to its relative size. The chances that some identifiable group gains disproportional influence could be much higher than the chances that a specific group does so (the exact probability would depend on the number of identifiable groups and their structure in the population).


  23. Size should be more closely tied to the importance and longevity of the decisions than the size of the community’s population. A city of 50,000 would need the same size sortition bodies as a city of 5 million, for example.

    Some decisions may be less important than others, and could be safely handled by small groups, because the harm of an unrepresentative decision is relatively small. Sometimes it matters less what the actual decision is than that the decision not be made by a corrupt self-serving elite power group, that is using the decision to amass more power.

    For example, a city might trust a very small sortition panel to decide the allocation of street re-paving funds for one year, but not whether to build a new sports stadium that will have a 40-year impact on taxes and development.


  24. Yes – that seems very sensible.


Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s

This site uses Akismet to reduce spam. Learn how your comment data is processed.

%d bloggers like this: