• Random sampling and mini-public deliberation

    • Basic principles: the statistical representative property of a “good” sample

    • Athenian democracy

      • Primary method of appointing public officials.

      • From Wikipedia: “In Athens, to be eligible to be chosen by lot, citizens self-selected themselves into the available pool, then lotteries in the kleroteria machines. The magistracies assigned by lot generally had terms of service of 1 year. A citizen could not hold any particular magistracy more than once in his lifetime, but could hold other magistracies. All male citizens over 30 years of age, who were not disenfranchised by atimia, were eligible. Those selected through lot underwent examination called dokimasia in order to avoid incompetent officials. Rarely were selected citizens discarded.[11] Magistrates, once in place, were subjected to constant monitoring by the Assembly. Magistrates appointed by lot had to render account of their time in office upon their leave, called euthynai. However, any citizen could request the suspension of a magistrate with due reason.[12]”

    • Venetian democracy

      • From Wikipedia: “New regulations for the elections of the doge introduced in 1268 remained in force until the end of the republic in 1797. Their intention was to minimize the influence of individual great families, and this was effected by a complex electoral machinery. Thirty members of the Great Council, chosen by lot, were reduced by lot to nine; the nine chose forty and the forty were reduced by lot to twelve, who chose twenty-five. The twenty-five were reduced by lot to nine, and the nine elected forty-five. These forty-five were once more reduced by lot to eleven, and the eleven finally chose the forty-one who elected the doge.[19] Election required at least twenty-five votes out of forty-one, nine votes out of eleven or twelve, or seven votes out of nine electors.[20] A detailed description of this process, and the ceremonial procession that followed, is preserved in Martin Da Canale's work Les Estoires de Venise (English translation by Laura K. Morreale, Padua 2009, pp. 103–116).“

    • Trial by jury

      • Government-supported “draft” tries to reduce self-selection bias

        • Employers required to let employees attend without penalty, etc.

      • Still has self-selection biases, biases based on chosen jurisdiction, … 

      • Deliberate biases due to juror elimination processes: good or bad?

    • Public opinion polling

      • No force of law, more self-selection biases: friendliness, time, question wording, physical polling environment, weather, …

      • Many ways to try to adjust for biases, but how reliable/complete?

      • Much of polled population may not have a strong or informed opinion

    • Randomness in popular voting

      • Risk-limiting audits

      • Random-sample voting

    • Deliberative polling

      • Diversity of opinion, perspective

      • But in considered, informed (by experts) form

      • Still need a moderator/organizer for agenda-setting, choosing experts…

      • Many self-selection biases, maybe worth due to higher time demand

    • Potential benefits flow in two directions

      • Voter -> process: better information from more diverse viewpoints, less likely to miss/neglect important issues or minority concerns, etc.

      • Process -> voter: expose voters to more diverse viewpoints, educate on mutual understanding and consensus-building, reduce polarization

    • Scalability: practical for many groups to deliberate on many issues concurrently

  • Technical challenges to unbiased, trustworthy random sampling

    • Vietnam war draft bias incident

    • Numerous hacked lottery incidents, electronic or not

    • FIFA “warm versus cold balls” rigging controversy

    • Computer-generated randomness challenges:

      • Producing any “good” randomness, given processor determinism

        • Operating system sources: keypresses, packet arrival, etc.

        • Hardware random number generators

        • Quantum sources of randomness

        • Lenstra’s parking lot cam, Cloudflare lava lamp cam, … 

      • Producing trustworthy distributed randomness

        • So we don’t have to trust any single contributor

        • Strawman approaches: just combine, or commit-and-reveal

        • Shamir secret sharing; RandHound/RandHerd


Post-lecture blackboard snapshot 2019:



Last modified: Thursday, 29 October 2020, 16:10