Weighted proxy voting

From InstaGov
Jump to navigation Jump to search


Weighted proxy voting (WPV) is what might be termed an allegiance system. Unlike political parties, however, allegiances in WPV are under much closer control of the user/member: there is little to no lag between action and effect, and the exact degree of allegiance is also under the user's control.

WPV was designed for use in direct democracy -- when so used, it forms the core element of liquid democracy -- but it can also be used as an aid in forming opinions about which no formal election is taking place.

Governance Application

As has been noted elsewhere, the major known forms of democracy each have systemic problems.

Representational democracy is prone to developing unaccountable power concentrations. Direct democracy, on the other hand, has problems with information overload.

Unaccountable power concentration (UPC) is when representative decisionmakers (typically politicians) are able to make poor decisions without suffering any substantial penalty -- because they are able to maintain their power base either despite voter opposition or because they are able to deceive voters into supporting their actions.

Information overload is both the need to understand a vast quantity of information in order to make rational decisions and a large number of decisions that must be made regularly. There simply isn't time for each voter to make an informed (and carefully-considered) choice on every issue, while the opinions of known experts -- those who already have that information, and have had time to think about it at length -- carry no more weight than that of anyone else.

Weighted proxy voting (WPV) was designed as a solution for both of these problems, while also taking a stab at the problems of rational allocation of expertise and influence-buying. It is a modification of proxy voting in which multiple proxies may be used in order to further distribute power.

low-level design: how it works logically

Aside: should I do a video, with diagrams and examples?

Here are the basic rules:

  • A. A "proxy"1 is a weighted average of the votes of selected other users.
  • B. For each proxy a user creates, they select (a) which other users they want to use as vote-sources, (b) how much weight to attach to each user's vote, and (c) what they want to do with the output (see item D below).
  • C. Each voter can build as many or as few proxies as they want.
  • D. Each proxy's output can be either:
    • (a) advisory -- "here's what the proxy thinks, now you can vote however you want"
    • (b) automatic by topic -- if the issue under vote is within certain categories the user has designated for that proxy, then the the user's vote will automatically be set to the proxy's output, unless/until overridden2
      • Obviously this could be extended to "slave" all of one's votes to a proxy, so "automatic global" is another option implicit within that.
    • (c) automatic by issue -- "the output of this particular proxy will be your vote on this one issue only"
  • E. Voters can edit their proxies at any time, as often as they want. Users can select whether they want the edited proxy's vote to overwrite existing votes, or only to be applied to future issues.

In other words, proxy voting is not so much a set of rules that voters have to follow, but a tool that voters can use however they want. They have complete freedom to use it as much or as little as they want, too.

usage design: what this accomplishes

WPV basically replicates and replaces the function of voting for representatives, but in a much more flexible and fluid way -- and one that is, crucially, much more directly under the control of the voter than is our current political system.

Someone whose opinions on a certain topic are known for being accurate, well-informed, and sensible should, in time, become widely adopted as an "advisor" -- a "proxy source" -- and their decisions could ultimately have as much influence as that of an elected legislator. Unlike the latter, however, if an "advisor" shows any sign that their vote is being influenced by special interests, voters are free to drop her/him from their proxy (or reduce her/his influence within it) with no warning.

Unlike direct democratic voting, however, WPV supports the existence of domain experts at every scale -- from the family member who reads a lot all the way up to internationally-renowned experts on highly specialized topics. Every citizen has an equal opportunity to be influential, and each can be crowd-rewarded with the degree of influence they have earned.

(Yes, there will be tools to help track the results of each vote, and who voted which way on it, so we don't have to rely entirely on our pathetic meatbrain memory-banks to keep track of the truly gargantuan amount of information involved.)

decisionmaking ecosystem

Neural voting is not a solution unto itself, although it would certainly improve vote fidelity even if used without any other tools. It was designed as part of InstaGov with the intention that it would be part of a larger decisionmaking ecosystem which helps improve the signal-to-noise ratio of the information by which people make decisions while also minimizing the impact of attempts to game or otherwise corrupt the system.

It is specifically designed to be a part of the liquid agenda system, by which problems are announced and solutions suggested.

It was intended to work side-by-side with the debate mapper system, which may in turn make use of it to help solve disputes about the validity of arguments.

To Be Discussed

  • potential problems with this system (including some decisions to be made about just how much data it should reveal to unprivileged users)
  • roadmap for getting from experimental and small-scale usage to nationwide and worldwide without breaking any laws


  • 1. For real-world use, instead of "proxy" we might use terms like "board of advisors", "counsel", "trustee", "genie", or something else similarly user-friendly. There has also been some discussion about whether "proxy" is synonymous with "advisor"; I'm choosing to use it in a way that is distinct from that, but this is another reason why it might be good to find a different term.
  • 2. I picture online voting working rather differently than IRL voting, especially in the time-domain. Voters could set their vote -- and change it -- at any time between when the issue appears up until the counting deadline. Some issues won't have counting deadlines, but will be more like ongoing realtime polls. Maybe I should have written about this first...