The Challenges of going from E-Voting to Liquid Democracy, Part 1: General Challenges of Delegations

More direct involvement of people through use of technology may improve our often sluggish and distorted democracies. E-voting – using modern technology for the ‘traditional’ elections – is difficult in itself, yet technical solutions for Liquid Democracy – with more direct and active participation – is even more challenging. In this article I focus on the challenges raised by Liquid Democracy, and one of its core features, delegations, compared to simpler form of E-voting.

E-voting (a subset of E-democracy) attempts to use modern technology – computers and networks – for elections. It has a history of decades, but it is still far from being solved. In the last years, cryptography and blockchain technology brought a fresh perspective on one of the fundamental issues: balancing privacy and verifiability.

Liquid Democracy may combine the best of the representative and direct democracy models. Representative democracy is used in Western democracies, and ensures participation, but only in a remote, indirect way. Direct democracy assumes more active participation, and more direct impact, but at a price of more effort needed by participants. Liquid democracy allows delegation of one’s voting right, in an arbitrary network of delegations. Delegations should be done to more knowledgeable, more caring, more active trusted acquaintance, can be forwarded, can be specific to topics, can span multiple votings, and can be withdrawn at any time. Delegation is a core aspect (liquid democracy is also called delegative democracy), and its incorporation in an e-voting scheme is not easy.

A system with arbitrary level of delegations has some obvious difficulties: verification of votes is more complex, as an end vote depends on a chain of delegations (which may be quite long!). A delegation can last for a long period (spanning multiple ballots); this increases the time dependency dimension of the system.

Another subtle issue is that in a chain of delegation, individual links may form in different order. For example, Bob may delegate his vote to Carol first, then Alice to Bob, then Carol to Dave. The final chain is Alice-Bob-Carol-Dave, but when Bob makes his delegation, there is no way to tell who will delegate to him.

A liquid democracy system should be anonymous, publicly verifiable, and convenient to use – but these goals often clash with each other. The result of a ballot depends on the votes and all the delegations, and verifiability would dictate that the graph of delegations should be available. But even with anonymized identities in the graph, the set of relationships is a source information, revealing secret identities (for sure or with some probability). For example, if one would compile a graph with the world’s airport and the routes between them, keeping the identities of the nodes secret, it is possible to infer the identity of important nodes with high probability. Another example is the pseudo-anonymous system of the Bitcoin network: while identities of addresses are secret, relationships between them (transaction) can reveal a lot.

The network of delegations thus cannot be available, as a malicious resourceful entity (government) could find out most of the votes. Another twist is the fact that for convenience reasons it is good to make delegate addresses public, to avoid the need for a two-way communication when making a delegation. Such a public delegate-address mapping is an additional challenge for privacy.

There are possible solutions for solving verifiability without full delegation graph, e.g. by using different identities for incoming and forwarded delegations, but these have their own challenges: they require some form of activity done at each delegation (forwarding, creating new identities, storing identities), with trust and practical concerns.

A delegative system is also more complex from visibility perspective: who can see what. For example, may a voter see the vote/delegation of her delegate? How about the delegate of the delegate? Accountability would dictate so, but more visibility is against strong privacy. Also, one may say that if someone cares much about the faith of her vote, she should vote directly, a delegation means trust. But does it have to be blind trust? Some minimal accountability is needed for sure, e.g. to be able to see if one’s vote was at all ‘used’ by the delegates (e.g. if a delegate fails to vote or delegate, the vote may be lost).

After reading so far, dear reader, hopefully you became a stronger fan of liquid democracy (though this is not the primary goal of this article). But are you still keen on building a liquid democracy system? You should be not scared away, and hang on. In the next part I plan to extend these challenges in the context of a specific e-voting system.