Liquid democracy

Liquid democracy (explained well, if slowly, by Jakob Jochmann) was first implemented for citizen voting (i.e. not only for internal use of an organization, usually a political party) in the Landkreis of Friesland in northwestern Germany in 2012.

It would be interesting to see liquid democracy implemented on a national level. Friesland shows that the way this is going to happen is likely from the bottom up: starting with local politics and progressing to regional, then (maybe, one day) to national politics. But before liquid democracy starts to be used for Big Decisions™, it must first solve its major issues.

The problems with liquid democracy are like those with electronic voting in general. Electronic voting is not anonymous and not secure. Bruce Schneier lists 6 properties which an electronic voting system must satisfy in his book Applied Cryptography, which I will term the ‘Schneier Criteria’:

  1. Only registered voters can vote
  2. Voters can only vote once
  3. Nobody can determine a particular voter’s vote
  4. Nobody can change a particular voter’s vote after it has been submitted
  5. Nobody can duplicate another voter’s choice
  6. Voters can verify that their vote was counted correctly

Paper ballots satisfy all the Schneier Criteria except for the sixth. For the purposes of liquid democracy, the fifth criterion can be partially ignored since transferring your voting power to someone else is the entire point of the system. However, nobody should be able to duplicate someone else’s vote without their permission (i.e. those who have been entrusted with others’ votes should not be able to vote on behalf of anyone except those who have chosen to delegate their votes to that person).

The only way a voting system can satisfy all these criteria is by using cryptography. Current research on so-called end-to-end verifiable voting systems is spotty. One problem is deciding what should be secret. A system which revealed the identities of those who voted on a particular proposal would be open to bribery (and fail criterion 3). Would it be sufficient that the author(s) of a particular proposal are named, but those who voted for it are anonymous?

Since cryptography is involved, liquid democracy is also going to have to move away from the web browser as its application platform. Web browsers are a bad platform for cryptographic applications. What’s needed is a standard protocol so that liquid democracy apps can be created for many platforms, and they can all work together.

A potential distraction from this is the problem of decentralization. It’s not important that a liquid democracy tool be decentralized since, if used for national politics, we will all depend on trusting the person who runs the central voting server — just as we must, in the end, trust the returning officer in our elections today.