Sortition by elimination

Worries are sometimes expressed about the impossibility of generating a sample of people at random in a way that cannot be manipulated by powerful actors. Sources of physical data are either too predictable to be of use or require machinery that is too arcane or sensitive to be effectively publicly verifiable. Social sources of data – such as the stock market or blockchain transactions – may be influenced powerful forces in society. Many randomizations that rely on explicit and symmetrical inputs from the public as a source of randomness have to utilize aggregation procedures that may allow those with advance knowledge of others’ inputs to manipulate the outcomes. With the prevailing mood of generalized distrust in institutions, a randomization mechanism would have to be completely open and verifiable to have a reasonable chance of inspiring confidence.

My proposal for such a mechanism is a simple elimination procedure which works as follows. At the outset, one candidate is eliminated. This candidate then gets to eliminate another, who then gets to eliminate another, and so on. The selection thus proceeds by sequential elimination of candidates until only one, or however many appointees are desired, remain.

This procedure is easily verifiable by any observer since it is self-contained and does not involve secrets, fancy machinery or fancy calculations. All the decisions involved are made in the open and cannot be foreseen in advance.

In addition to being manipulation resistant, this procedure has the advantage that it involves all interested citizens in the selection procedure and allows them to influence the outcome. By creating a new form of mass political participation, this procedure addresses the oft-heard objection to sortition that it deprives people from having influence over the appointment of decision makers.

In fact, while, like any form of mass participation, the impact made by any single decision-maker is minute, this form of participation is more meaningful than electoral participation because the choice made by each person – who to eliminate – is entirely unrestricted. This is in contrast which the electoral choice which is restricted a-priori by a primaries process in which the field of candidates is drastically narrowed-down. In the proposed procedure, citizens are completely free to make their elimination choices as they see fit, even if it may be seen as a sign of good citizenship to make this choice at random.

A minor technical point: The first candidate to be eliminated, the starting point of the elimination chain, can be chosen arbitrarily – this is not a position of decisive power, but rather the opposite, a position of disadvantage. If no other procedure is found suitable, an election can be used to select this person.

18 Responses

  1. So you’ve finally given up on your notion of sortition as a form of statistical representation in favour of a procedure where “candidates”, drawn from a small group of “interested citizens”, eliminate each other (perhaps you’ve been watching the new Korean drama on Netflix). This assumes that they know (of) each other, so would only serve to further embed Manin’s principle of distinction. And it is completely opposed to the Stone/Dowlen lottery principle, whereby sortition is a way of eliminating all reasons (good, bad and otherwise) in the selection process. Of all the objections I’ve seen to the political potential of sortition, I’ve not yet come across serious concerns regarding the possibility of an impartial lottery.

    Liked by 1 person

  2. I’m with Keith on this – this procedure would only be feasible where the candidate pool is already small. The problem of fair lotteries is already solved by institutions such as the British National Lottery, which disburses large amounts of money at random.

    Like

  3. That said, as a procedure for ensuring nobody controversial ends up on a panel of people drawn from a larger group, this has some merit. It is very far from random selection, however – it will select for the most well-liked and least well-known members of the pool.

    Like

  4. Hi Yoram, If any outgoing office-holder was the first to chose who else to eliminate, then many elections could be avoided. Only in the case where a new office was being created would there be a need to chose a first eliminator, by election, sortition or etc. To explain, existing office holders leaving the office for which the sortition was being held would be the first eliminator (as well as the first eliminated); the second eliminator would be the person chosen by the outgoing office-holder – they would chose the third eliminated, who would chose the fourth, and so forth, as you’ve described. This method would set up a one-term limit for office-holders. In sortition, this would not be likely to change outcomes, though.

    What do you all think? Brian-

    Liked by 1 person

  5. Yes indeed, the Principle of Distinction (normally anathema to Yoram) will rule. And I’ve never heard of anyone claiming that the National Lottery is fixed.

    Like

  6. Brian:> In sortition, this would not be likely to change outcomes.

    So why is the idea being floated on this blog?

    Like

  7. Yoram’s proposal is a form of election — the only difference being that it has a negative valence and the electorate at each round comprises one (self-selected) person. Whatever merits it may or may not have, it has no relevance to sortition whatsoever.

    Like

  8. Oliver,

    > this procedure would only be feasible where the candidate pool is already small

    Not at all. In fact, the contrary is true. When the pool is large, most people will not know most other people on the pool, and therefore eliminating candidates at random would be the natural thing to.

    > The problem of fair lotteries is already solved by institutions such as the British National Lottery, which disburses large amounts of money at random.

    I think you do not appreciate what’s at stake here. The largest sums ever given in lotteries are several orders of magnitude smaller than the budget of a government. In addition, control of a country implies powers that no amount of money can bestow. The CIA, say, would hardly invest its resources in winning the lottery, but may very well be interested in rigging the allotment of Congress.

    > It is very far from random selection, however – it will select for the most well-liked and least well-known members of the pool.

    I disagree. I think most people would use standard random procedures and thus the outcome would be quite similar to pure sortition, with maybe a bit of dokimasia or ostracism thrown in.

    As for the specific supposedly-advantaged groups you mention:

    1. To the extent “most well-liked” implies highly visible people, those people are, like any highly visible people, actually at a disadvantage, because it is enough that one person finds them dangerous to have them deliberately eliminated.

    2. If by “least well known” we are talking about reclusive eccentrics who have an unusually small number of acquaintances, then it is by definition a marginal group which will be eliminated by those normative people who use randomization for elimination. It is also quite possible that the few people who do know such recluses would consider them dangerous and deliberately eliminate them.

    Like

  9. Brian,

    > If any outgoing office-holder was the first to chose who else to eliminate,

    Yes – that does sound like a good idea.

    Like

  10. Happy frothing, Sutherland.

    Like

  11. Yoram,
    While I appreciate the “outside the box” thinking, this procedure creates at least as many problems as it solves. First to help clarify… by the word “candidate” in your initial description… I assume you mean that ALL adult residents are automatically initially candidates (until eliminated)… not some pre-selected volunteer pool, correct? The massive reporting and record keeping of eliminations creates a huge opening for manipulation and a near impossibility of knowing if an elimination was real or fabricated by the record-keeper. Once a person is told they have been eliminated, I suspect MOST people would say “so what,” and NOT participate by bothering to eliminate another person (seems like effort with absolutely no consequence for the person, nor feeling of civic pride). Or at the very least, not bother to take an elimination action until pestered several time by the authorities over a few weeks. So, the chain of eliminations would constantly come to a dead-end and need to be restarted by somebody (or software) randomly picking somebody to eliminate, which defeats your goal. If you literally had a single chain of a single eliminated person eliminating one more, this chain might take years to come to an end in a large society.

    You objection to using some publicly confirmable number is not really a manipulation problem…There are countless easy completely unpredictable and non-manipulatable ways to do this. As an example, just take the second decimal place and fifth decimal place digits on a precise (to 5 decimal places) public ceremony stop clock that measures the moment a dropped ping pong ball hits the sensor as your seed numbers to plug into a simple formula to take names from a n ordered list of names that is public (like start with page 6, name 8 and take every 1,000th name after that.)

    Liked by 1 person

  12. Terry:> I assume you mean that ALL adult residents are automatically initially candidates

    Given Yoram’s consistent opposition to quasi-mandatory sortition, his “all interested citizens” would suggest the jury pool would be composed of a small group of activists. If likely to be cancelled (quasi-)randomly, then why would Ornery Joe bother to participate? Those who did want to be included would not be a representative sample of the target population, but Yoram has long given up on representation (or even democracy, as the word is understood by everyone else). People, on the whole, understand that random lotteries include people in an impartial manner, but being eliminated by the choice of somebody else is unlikely to go down well. As for historical parallels, the Athenians made no connection between sortition and ostracism — in fact ostracism was superseded by the extension of jury trials. As for dokimasia, this was a rational, public assessment, not an act of random bigotry.

    Like

  13. Terry,

    > by the word “candidate” in your initial description… I assume you mean that ALL adult residents are automatically initially candidates (until eliminated)… not some pre-selected volunteer pool, correct?

    Naturally.

    > The massive reporting and record keeping of eliminations creates a huge opening for manipulation and a near impossibility of knowing if an elimination was real or fabricated by the record-keeper.

    I don’t see the problem. If record is kept of the elimination chain (i.e., who was eliminated by whom), then verification becomes straightforward. Any person can find out who they were eliminated by and any interested investigator can easily draw a sample of eliminated people and make sure they were eliminated properly.

    > Once a person is told they have been eliminated, I suspect MOST people would say “so what,” and NOT participate by bothering to eliminate another person (seems like effort with absolutely no consequence for the person, nor feeling of civic pride)

    First, I don’t see why participating in the elimination chain would not be as much a source of civic pride and be considered as consequential as voting (if not more so).

    But that said, it makes sense that only some fraction of the citizens would bother to take action (as presently the turnout rates are not close to 100%), so indeed some sort of a fallback mechanism would need to be employed. Let’s say that experience shows that 50% of citizens participate. Then each participating citizen could be asked to name, in addition the citizen to be eliminated, a fallback elimination. In case the chain snaps because an eliminated citizen does not respond, the chain continues through the most recently nominated fallback elimination.

    > public ceremony stop clock that measures the moment a dropped ping pong ball hits the sensor

    But what is the way to guarantee that the clock was not rigged to produce a desired outcome?

    Like

  14. Yoram,

    I feel you underestimate the time it would take to run your system on even a municipal level. Let’s suppose you have a city of a million pool members. Suppose it takes one minute for each pool member to pick the next link in the chain, day or night, whatever else they’re doing. The process would take 695 days to be completed. And that scales linearly, so for a city of 10 million, it would take nearly 20 years. In a country of a hundred million, every member of the pool would be dead before the chain was completed. This process has very hard limits at a very small number of participants.

    Like

  15. > ALL adult residents are automatically initially candidates (until eliminated)… not some pre-selected volunteer pool.

    OK, but we should remember that that average acceptance rate for sortition pools is 3%, and people who accept the invitation have a reasonable chance of their lot being drawn (as they were included in the original statistical sample [several thousand]). But if the vast majority of a pool of several million will be eliminated, then why should anyone bother to participate? And Yoram is right to compare it to election, as the process doesn’t involve sortition (hence my “frothing” at it being posted on this forum). Besides which, as Oliver point out (albeit more politely), it’s spectacularly silly.

    Like

  16. Oliver,

    Yes – I agree that slowness is a serious problem with the process in its simple form as described above. I suspect, however, that if the principle is accepted then some faster variation – i.e. one where very many eliminations can take place in parallel – can be developed.

    Like

  17. Yoram,

    One way to do it might be to have each eliminated pool member eliminate two others, turning elimination from a linear to an exponential process. The issue there is that at the end of the process there would be a very large number of competing claims to eliminate a relatively small number of people, producing the problem of how to determine who is left un-eliminated without introducing a weak point for interfering agents to attack.

    Liked by 1 person

  18. The labyrinthine political arrangements of the Most Serene Republic of Venice were straightforward by comparison. Whilst this thread might provide amusement to a few nerds with nothing better to do on a Saturday afternoon, the probability of anyone in the real world taking it seriously is zero.

    Like

Leave a Reply

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

WordPress.com Logo

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

Google photo

You are commenting using your Google 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: