Democratic Weights and Persistent Elections

This is an idea I've already written about at length (opens new window). In brief:

• Each citizen has a pile of democratic weights, which they place on elections they care to participate in according to how strongly the citizen cares about that election. They can move these weights around at any time.
• Elections are persistent, as in they don't close, but rather smoothly switch outcomes whenever the citizenry votes for something else, using stabilization buckets to prevent things from switching too often and nomination buckets to prevent someone attacking the election with many nonsense candidates or otherwise the election having too many candidates to be reasonable.
• Because elections in this style are better able to be iterative and not prone to capture from small coordinated groups, it's safe to use them for documents. So constitutions, tax codes, laws, budgets, etc, can be voted on directly by any size of electorate.

Democratic weights can also be used for other things than elections, such as several of the other ideas in this grab bag.

I conjecture that the problem of dysfunctional strategic voting (opens new window) wouldn't be relevant to persistent elections, since it seems intuitive that since the election itself serves the same role as polling in a normal election, the ballots of all interested voters would converge to their most competitive and strategically fit version. This would imply the best voting mechanism to use within a persistent election would be the one where the strategic and fully honest ballots had the smallest effective difference.

Weighted Score Voting and Persistent Range Voting

Score Voting (opens new window) is simply any voting system where votes are cast as a "score" in some numerical range, which can be as simple as any discrete range such as just 0 or 1 (which is Approval Voting), all the way to a full continuous scale such as the rational numbers (opens new window). The scale can allow negative scores.

In score voting you simply add up the scores and the candidate with the highest sum is the winner.

The only downside of Score Voting that needs to be addressed is it allows "score extremism", where people give every candidate they even slightly like the maximum score and all others the minimum score, even if that isn't an honest scoring. However score voting systems always satisfy the monotonicity criterion (opens new window), since it's never rational for a voter to score a candidate they like less above one they like more.

(It actually seems intuitive monotonicity would be a required property for strategic convergence, since it would imply strategic adjustments would only be made in one direction, ensuring that no "flip-flopping" or cyclic equilibrium (opens new window) would happen.)

Because of my above conjecture about the strategic convergence of persistent elections, I don't think score extremism is worrying, since if someone maximally scores a less preferred candidate, they will only down-score them as the viability of their more preferred candidates and the non-viability of their most feared candidates is revealed.

Weighted Score Voting is where democratic weights are placed on the election as a whole, and then all the scores are multiplied by that weight. This has the effect of gauging how much concern a single voter has for the results of that election specifically.

Weights are used on elections rather than directly on candidates because elections are highly non-substitutable (opens new window), meaning that if you care about the results of one election, it's very unlikely we could find a different election where you would trade success in that election and be completely indifferent (opens new window) about the trade. Stated simply, you wouldn't be willing to trade one election for another, they're just separate and different to you. (At least this should be the case in any reasonable constitution.) However candidates in elections are frequently quite substitutable, theoretically even coming from the same political party or having very similar positions. This means that it's more likely for you to be indifferent between two different candidates, so it's more dangerous to allow your vote to be split (opens new window) between them as would be the case in using weights directly.

Range Voting (opens new window) is just score voting where a candidate's final result is the average of scores rather than the sum (sum of score values / number of score values rather than just sum of score values).

The reason Range Voting is interesting is primarily that it allows unknown candidates to break through and get attention merely with a small number of activists liking them, since they can jump to the top of an election with only a small number of actual voters.

In normal "one-off" elections this could be dangerous, since an obscure candidate could surprise everyone and create "election hangovers. We don't have to worry about this when using persistent elections because nomination buckets generally keep nonsense candidates out, and an unknown candidate shooting to the top of the election isn't "dangerous" since stabilization buckets give the rest of the electorate time to notice and respond to this with their full attention.

Democratically Weighted Sortition

From the wikipedia on Sortition (opens new window):

In governance, sortition is the selection of public officials or jurors at random, i.e. by lottery, in order to obtain a representative sample.

Sortition fairly cleanly addresses the main corrupting forces (called agency problems (opens new window) in economics) we often deal with in government:

• Some people want power for it's own sake (opens new window), for the status and ego and ability to dominate others. Sortition (and effective democracy in general) spreads power among many people, using a system of rules to make decisions instead of trusting individuals.
• Some people can be bribed or black-mailed or otherwise co-opted to advance the interests of some outside power-seekers against the interests of the public. Sortition (and democracy in general) makes it difficult to corrupt enough people that such power-seekers can get their way without being discovered.
• People always have their own selfish interests that go against the interests of the public, and some people are willing to betray their duties to advance those interests. Sortition (and democracy in general) ensure that many selfish interests are all thrown together, and counterbalance each other and prevent any individual from gaining undue advantage.

Democracy in general has these anti-corruption properties (if designed effectively), but sortition allows those same properties to apply to much smaller and more focused groups. The ability to trust smaller and more focused groups is important, since many decisions in society are some combination of boring and complex, and won't necessarily inspire mass participation from many people.

I think there's one small improvement we could make to sortition though: democratically weighting the random selection. Random really does mean random, and it would be nice if people who are generally liked and trusted by their peers, and especially aren't distrusted and disliked by their peers, were a little more likely to be randomly selected.

To democratically weight sortition, we allow voters to place democratic weights to change the likelihood of someone being selected by sortition. They can place them positively, meaning it becomes more likely that person will be randomly selected, or they can place weights negatively, meaning it becomes less likely that person will be randomly selected.

If you want to ensure that no person can have zero or negative probability of being selected, some sort of function that has an exponential limit toward zero (opens new window) and otherwise grows normally could be used to convert the weights to the actual probability, so more and more negative overall weight only pushes them toward zero but never fully to it.

Implementing weighted random selection is easy using the Roulette Wheel Algorithm (opens new window).

Persistent Delegation

If you're using persistent voting to select anything, from documents to representatives to administrators, you can also choose to give extra delegative weights to specific groups, weights they can only use for specific decisions, with the intent of partially delegating those decisions to them. Since this smaller group has more weights to make those specific decisions, they can achieve things relatively quickly without all of society needing to be mobilized to make it happen; but the rest of society can still intervene directly if those delegated are running afoul of society's broader values or preferences.

Why do this? Some decisions in society are boring or very niche, and only a small number of people can or even should spend the time needed to fully understand them. But this doesn't mean the decisions shouldn't have democratic oversight.

Just because some small group has expertise or interest in a topic, doesn't mean they will perfectly represent everyone's values. Technocracy (opens new window) may be appealing to some, since it's true many societal decisions are made in complete ignorance of important expert knowledge. But expertise is just a tool that can be used to drive toward many different incompatible goals.

Saying we should put experts in charge of society is like saying long distance rally racers (opens new window) should decide every detail of your family road trip. Sure those racers are probably the best people in the world at figuring out how to quickly plan and execute a land journey between any two places, but that doesn't mean they'll choose a destination you even want to go to, let alone make the journey enjoyable.

Expertise is just a tool, a set of skills and knowledge you can use to achieve goals. Goals on the other hand are chosen based on values and preferences, which aren't objective and aren't subject to expertise. Choosing the destination is different than making the journey. We should definitely consult experts to understand where we could go, and we should delegate experts to actually get us there, but we must ourselves choose the destination and the constraints we demand for the journey.