14 votes

Could "fuzzing" voting, election, and judicial process improve decisionmaking and democratic outcomes?

Posted by dredmorbius
Tags: ask.discussion, sortition, politics, voting, elections, reform, science.political

Voting is determinative, especially where the constituency is precisely known, as with a legislature, executive council, panel of judges, gerrymandered electoral district, defined organisational membership. If you know, with high precision, who is voting, then you can determine or influence how they vote, or what the outcome will be. Which lends a certain amount of predictability (often considered as good), but also of a tyranny of the majority. This is especially true where long-standing majorities can be assured: legislatures, boards of directors, courts, ethnic or cultural majorities.

The result is a very high-stakes game in establishing majorities, influencing critical constituencies, packing courts, and gaming parliamentary and organisational procedures. But is this the best method --- both in terms of representational eqquity and of decision and goverrnance quality?

Hands down the most fascinating article I've read over the past decade is Michael Schulson's "How to choose? When your reasons are worse than useless, sometimes the most rational choice is a random stab in the dark", in Aeon. The essay, drawing heavily on Peter Stone, The Luck of the Draw: The Role of Lotteries in Decision Making (2011), which I've not read, mostly concerns decisions under uncertainty and of the risk of bad decisions. It seems to me that it also applies to periods of extreme political partisanship and division. An unlikely but possible circumstance, I'm sure....

Under many political systems, control is binary and discrete. A party with a majority in a legislature or judiciary, or control of the executive, has absolute control, barring procedural exceptions. Moreover, what results is a politics of veto power, where the bloc defining a controlling share of votes effectively controls the entire organisation. It may not be able to get its way, but it can determine which of two pluralities can reach a majority. Often in favour of its own considerations, overtly or covertly --- this is an obvious engine of corruption.

(This is why "political flexibility" often translates to more effective power than a hardline orthodoxy.)

One inspiration is a suggestion for US Supreme Court reform: greatly expand the court, hear more cases, but randomly assign a subset of judges to each case.[1] A litigant cannot know what specific magistrates will hear a case, and even a highly-packed court could produce minority-majority panels.

Where voting can be fuzzed, the majority's power is made less absolute, more uncertain, and considerations which presume that such a majority cannot be assured, one hopes, would lead to a more inclusive decisionmaking process. Some specific mechanisms;

  • All members vote, but a subset of votes are considered at random. The larger the subset, the more reliably the true majority wins.
  • A subset of members votes. As in the court example above.
  • An executive role (presidency, leader, chairmanship) is rotated over time.
  • For ranged decisions (quantitative, rather than yes/no), a value is selected randomly based on weighted support.

Concensus/majority decisionmaking tends to locked and unrepresentitive states. Fuzzing might better unlock these and increase representation.

Notes

  1. A selection of articles on Supreme Court reforms and expansion, from an earlier G+ post: https://web.archive.org/web/20190117114110/https://plus.google.com/104092656004159577193/posts/9btDjFcNhg1 Also, notably, court restructuring or resizing has been practiced: "Republicans Oppose Court Packing (Except When They Support It)".

9 comments

  1. [4]
    nacho
    Link
    I think a much more elegant solution that avoids the additional democratic deficit built in randomness of fuzzing causes, and also minimizing the existing democratic deficit by having votes thrown...

    I think a much more elegant solution that avoids the additional democratic deficit built in randomness of fuzzing causes, and also minimizing the existing democratic deficit by having votes thrown out is to implement leveling seats.

    Making votes count, even if you just get a couple percentage of the vote spread across huge geography, means that the electorate is much more fairly represented. There are a number of ways of implementation, but true for all of them are the following:

    • Gerrymandering doesn't work if you have leveling seats.
    • New parties can emerge with leveling seats.
    • Old parties can gradually fall out of favor with leveling seats.
    • The impetus for voting increases tremendously as your vote always counts with leveling seats.
    • Concerns not just for one geography, but for the whole body you make decisions about come to the forefront, so there's much less incentive to only want benefits for YOUR voting district.
    • The tendency towards two major parties is greatly reduced, so you're more likely to find a party closer to more of your personal political priorities
    • Leveling seats are a tried and tested system that demonstrably works (and has been part of the reason why countries at the top of so many statistics have reached those tops
    • Votes in all voting districts have a more similar value in parliamentary representation relative to each other.
    14 votes
    1. [3]
      vord
      Link Parent
      Philadelphia has a pretty decent system for city council: 17 members total 10 by district 7 at-large seats, but only 5 can be chosen on the ballot, insuring some minority party representation. The...

      Philadelphia has a pretty decent system for city council:

      • 17 members total
        • 10 by district
        • 7 at-large seats, but only 5 can be chosen on the ballot, insuring some minority party representation.

      The Working Families party was able to oust a Republican by strategically siphoning a few votes from the Democrats, who of course screamed bloody murder about 'risking a loss'.

      3 votes
      1. [2]
        dredmorbius
        Link Parent
        How are the two non-balloted at-large seats chosen?

        How are the two non-balloted at-large seats chosen?

        1. vord
          Link Parent
          I believe it's the next highest vote earners, so in practice they were Republican seats. So the Republicans were quite unhappy with losing one.

          I believe it's the next highest vote earners, so in practice they were Republican seats.

          So the Republicans were quite unhappy with losing one.

          1 vote
  2. [5]
    arghdos
    Link
    Is the idea that given an uncertain outcome, the majority would be more willing to come up with a compromise that satisfies a plurality in order to stave off total defeat? My first instinct was...

    Where voting can be fuzzed, the majority's power is made less absolute, more uncertain, and considerations which presume that such a majority cannot be assured, one hopes, would lead to a more inclusive decisionmaking process

    Is the idea that given an uncertain outcome, the majority would be more willing to come up with a compromise that satisfies a plurality in order to stave off total defeat? My first instinct was that the parties would simply establish their preferred position and expect their members to toe the line, by and large as happens today. but I think I could see circumstances where, if the stakes of losing are perceived to be too high, or one possible outcome is seen to be too extreme, the uncertain outcome could drive compromise.

    Another procedural point (somewhat tangential to the sortition proposal) would be to make the voting process anonymous, both for the member’s who’s votes are selected and how they voted, to hopefully dilute the power of the leadership in the parties further, and give members the ability to vote how they feel without hyper partisan backlash.

    2 votes
    1. dredmorbius
      (edited )
      Link Parent
      There are a few points at play, and as I dive into Schulson's sources and revisit the court-reform proposals several are addressed there: Fuzzed voting means that any given majority is not itself...

      There are a few points at play, and as I dive into Schulson's sources and revisit the court-reform proposals several are addressed there:

      • Fuzzed voting means that any given majority is not itself absolute. Odds for "leakages" to the minority exist. This is a feature.
      • Because any one vote's significance is reduced, the benefits of motivated vote-gaming, whether within a given body or within a public electorate at large, is reduced. Put inversely: the cost of vote manipulation is increased by the fraction of votes sampled. If only 1 in 2 votes are counted, costs of vote-manipulation double. If 1:10, they increase tenfold.
      • Because control is more likely to switch among factions, the appeal to any one faction of instituting structural or rules changes which disadvantage or neutralise the minority is reduced: they will very likely be the minority themselves on occasion. This argument is strongly similar to John Rawls's "Veil of Ignorance".
      • The value of "stacking" institutions (notably the judiciary) would have reduced appeal. Presently, an appointment, especially to the Supreme Court, affects its structure for 5--10 years, and can result in individual tenures of 20--30 years. The wins make selections highly significant. Reducing the stakes, through enlargement, random selection of panels to individual cases, and other mechanisms, would preserve the capabilities and functions of the court while reducing is partisan utility. Again, manipulation costs are strongly increased while effects are reduced.
      • Institutionally, the power of extreme outlier factions is reduced relative to the median. If, say, 1:10 votes are sampled, the odds that a 10% fringe vote is selected at all is only 1:10. The result might be an impact on individual votes, but not a persistent and reliable influence on a given body or electorate.
      • Institutionally, dynamics amplifying partisanship themselves should, I hope, be reduced. Currently, such dynamics tend to be self-feeding. Under a sortition system, median voices should be more prominent, the returns to extremism less positive, and stability more attractive.

      There are some possible negatives:

      • Large-scale change or rapid adaptation might be reduced. If a minority group is considered one example of a "fringe" its ability to influence policy through coalition politics might be reduced. Likewise adaptation to a rapidly-evolving period of change. Neither seem especially well-supported by the status quo.
      • Minority "third parties" --- typically Greens or Libertarians in the US --- wouldn't see any particular benefit. Some thing like @nacho's "leveling seats" proposal would be required for them to see representation.
      • Antidemocratic natures of institutions such as the Electoral College would still largely remain, though might be somewwhat blunted.

      Unlike vote-levelling, though, vote fuzzing achieves reduced partisanship through a procedural rather than structural change. I see fuzzing as a more generalised solution.

      I've been vague about just how "fuzzed" votes should be. I'm not sure, and possibilities range from drawing a single ballot from a cast set (this resembles the original Greek method -- counting all votes was far more difficult), to drawing a sample of cast votes. With a true random sample the accuracy of an estimator is almost entirely independent of sample size. For a fuzzing operation, smaller samples would be preferable in the sense of increasing randomness. Generally, "large sample" statistics begin at a sample size of 30, or n=30. With a history of votes, it's possible to model possible alternative outcomes through Monte Carlo modeling: repeatedly re-running the election by selecting different subsets of the reported cast vote. The number of times subsets of given votes differ from the total count shows the level of random fuzzing.

      5 votes
    2. [3]
      cfabbro
      (edited )
      Link Parent
      How would you ever hold politicians accountable if all their votes were anonymized? Politicians could simply claim to their constituents and the press that they voted one way, when they actually...

      make the voting process anonymous, both for the member’s who’s votes are selected and how they voted

      How would you ever hold politicians accountable if all their votes were anonymized? Politicians could simply claim to their constituents and the press that they voted one way, when they actually did the opposite. And a a result, that sort of system would be absolutely ripe for corruption and bribery, IMO.

      5 votes
      1. arghdos
        Link Parent
        Hmm, good point. I’ll admit it’s not my brightest idea ever, but I also don’t know if it’s that much worse than our current system of non-transparency on the influence end (e.g., the ‘legal’...

        Politicians could simply claim to their constituents and the press that they voted one way, when they actually did the opposite

        Hmm, good point. I’ll admit it’s not my brightest idea ever, but I also don’t know if it’s that much worse than our current system of non-transparency on the influence end (e.g., the ‘legal’ corruption of PACs and dark money here in the states), and hyper transparency and public excoriation at perceived disobedience that makes it effectively impossible for most politicians to go against the party line. I’m not sure how to get out of that without providing some measure of anonymity (probably less than full), unless the fuzzing itself can change the structural and procedural incentives enough to break the accelerating trend towards hyperpartisan top-down leadership, as @dredmorbius suggests.

        2 votes
      2. dredmorbius
        Link Parent
        Fuzzed votes need not be anonymised. The anonymous vote has value for the electorate, to avoid vote buying or selling. Though fuzzing itself is at least a partial protection against this. For...

        Fuzzed votes need not be anonymised.

        The anonymous vote has value for the electorate, to avoid vote buying or selling. Though fuzzing itself is at least a partial protection against this.

        For representatives or judges, selectively dropping votes is one method already practiced, in Olympic figure skating, for example:

        12 judges evaluated each skater, but only nine of those votes, selected at random, actually counted towards the final tally (the ancient Athenians judged drama competitions in a similar way). Figure skating is a notoriously corrupt sport, with judges sometimes forming blocs that support each other’s favoured skaters. In theory, a randomised process makes it harder to form such alliances. A tit-for-tat arrangement, after all, doesn’t work as well if it’s unclear whether your partners will be able to reciprocate.

        From TFA.

        For elected reepresentatives, all votes could be recorded but only a subset selected. Again, at random.

        1 vote