3 ms·

> votes secret with software Any form of voting can use secret ballot or not. Liquid democracy and ballot secrecy are entirely orthogonal issues and neither ne

• notahacker 4y ago
  > Any form of voting can use secret ballot or not. Liquid democracy and ballot secrecy are entirely orthogonal issues and neither necessitate software. But they're not orthogonal issues (particularly not in the format discussed in the paper, where chains of neighbours may delegate to each other until one person in the chain casts a vote). You can't have a system in which multiple votes are delegated to X and yet the identity of X is indistinguishable from the single votes of Y and Z. > This isn't something that would happen. One would be made aware one had become a delegate. That depends heavily on mechanism design though. If I'm aware that I've become X's delegate, the objection upthread is sustained because no secrecy exists and it is practical for me to purchase X's vote (and indeed to purchase and validate that I possess the votes of tens of thousands of indifferent voters). If I am simply aware at some stage that I have become someone's delegate and am not made aware that person is X, the level of secrecy means that I may rationally choose to delegate my votes to X without realising the two delegations cancel out.
  • guerrilla 4y ago
    > But they're not orthogonal issues (particularly not in the format discussed in the paper, where chains of neighbours may delegate to each other until one person in the chain casts a vote). You can't have a system in which multiple votes are delegated to X and yet the identify of X is indistinguishable from the single votes of Y and Z. This is just one form [1], there are many. The two variables are independent. You can have direct, liquid, representative democracy with or without secret ballot; therefore, they are orthogonal issues. > If I'm aware that I've become X's delegate, the objection upthread is sustained because no secrecy exists As I commented elsewhere, you don't need to be aware that you are anyone's delegate, just that you are a delegate. You don't even need to know how many votes you have to cast. The only thing you need to do is cast the votes that you have, however many, an to whomever they belong, which you need know nothing about. 1. https://en.wikipedia.org/wiki/Liquid_democracy https://en.wikipedia.org/wiki/Liquid_democracy
    • notahacker 4y ago
      > The two variables are independent. You can have direct, liquid, representative democracy with or without secret ballot; therefore, they are orthogonal issues. The two variables are not independent because the delegation system destroys the principle feature which makes ballot secrecy trivial (every person casts the same number of votes in the same way over the same time window). Workarounds for this have significant cost (requirement to use "black box" software or requirement for delegation to take place strictly prior to the actual vote and for a person with a lot of delegated votes to fill in potentially hundreds of ballot papers, or requirement for delegates to fill in special multiple-vote ballots which may end up being personally identifiable despite best endeavours to anonymise them). > As I commented elsewhere, you don't need to be aware that you are anyone's delegate, just that you are a delegate. And if you don't know whose delegate you are, you might unintentionally cancel out a vote you would normally delegate by delegating to someone who delegated to you. As I commented in the very next sentence. Individually identifiable delegations or permitting edge cases with cancelling-out delegations resulting in no actual votes cast by a particular group, pick your poison. (I guess you could also avoid this collision by only permitting delegation to a separate class of people who are not allowed to delegate their own votes but that's a completely different system to the one described in the paper which will yield completely different results...)
  • mistermann 4y ago
    > That depends heavily on mechanism design though. > If I am simply aware at some stage that I have become someone's delegate and am not made aware that person is X, the level of secrecy means that I may rationally choose to delegate my votes to X without realising the two delegations cancel out. Isn't the obvious solution to simply make the software enforce such situations? Or am I misunderstanding?
    • notahacker 4y ago
      Not really in an acceptable way, no. If the software blocks me from delegating my vote to X because X has already delegated their vote to me, then ballot secrecy is violated. If the software blocks me from delegating my vote to anyone because somebody has already delegated their vote to me, any third party willing to "burn" a vote on delegating to me (to troll me, perhaps) can change my vote from my original intention to delegate it to $subjectexpert and force me to take a stance on the polic(ies) themselves or not vote, which seems like a pretty undesirable outcome even if it doesnt benefit the troll. Both measures partially violate ballot secrecy in another way (if the software doesn't block me from voting to delegate to a person, then it reveals they didn't delegate their vote to me) The only way around it is for delegates to self select [or be selected] in advance as a separate class of voters unable to delegate their own votes. This avoids this conflict and could work, but isn't the system described in the paper and will have different dynamics as a result.