|Home - Blog - Publications - CV - Scribblings - Photo Album - Funny|
Over a decade ago I wrote up some ideas I called “Delegative Democracy,” which has also become known as “Liquid Democracy.” My draft paper from 2002 took a first stab at laying out the general idea, though I never finished or tried to publish it. Since a lot has happened in this space since then, and I get regular inquiries about it, I thought it was high time to revisit the idea and review more recent developments.
Many people including myself like the participatory ideal embodied in direct democracy, but this ideal does not readily scale beyond small communities. As I will shortly be moving to EPFL, I look forward to learning first-hand how the Swiss make direct democracy work for them — but Switzerland is a small country divided into even smaller, fairly autonomous cantons. A basic criticism of direct democracy is that expecting that every eligible voter to attend all the meetings and keep up with all the issues, rather than actually getting more people involved, may instead disenfranchise those without infinite spare time or patience.
Delegative democracy attempts to make direct democracy scale, by allowing anyone who cannot participate directly in a particular vote to delegate their vote to someone they trust to participate and vote on their behalf. This process may be viewed as a mechanized equivalent of seeking advice from a friend and voting based on that advice. Delegative democracy differs from representative democracy in the principle that each voter should have free, individual choice of their delegate — not just a choice among a restricted set of career politicians — just as we already make a free, individual choice of our friends. Further, voters may choose to participate in some meetings directly, overriding their delegate’s choices in those meetings, and voters may revoke or change their delegation at any time.
While corporate governance practice often allows the related idea of proxy voting, a proxy vote often amounts to a “pre-packaged mandate”: e.g., a directive to cast an unconditional YES or NO vote on my behalf regardless of what new information might be presented at the meeting or what discussion might ensue. The spirit of delegative democracy is closer to representative democracy, in that delegation is not about delivering my “canned decision” but about asking my trusted delegate to participate in the full deliberative process on my behalf. My delegate can and is expected to listen and engage in the debate, consider the information available, and make what his views as the best decision on that basis. If I dislike his decision, I can choose a different delegate before the next vote.
Perhaps unsurprisingly, I was only one of many people to think along these lines over the decades. In Lewis Carrol’s Principles of Parliamentary Representation of 1884, candidates could “club” their votes together, allowing a candidate who received more votes than required to win a seat to delegate the “excess” votes toward the election of other candidates. James C. Miller in 1969 foresaw electronic voting and proposed a delegate proxy voting scheme, to enable more widespread — and more propoportional — representation in political decisions.
Starting around 2002 the Internet exploded with reinventions of this and similar ideas. Besides my Delegative Democracy proposal, there was Dennis Lomax’s Beyond Politics, Jio Ito’s Emergent Democracy, Sayke’s Liquid Democracy and voting system, Mikael Nordfors’ Democracy 2.1, James Green-Armytage’s delegable proxy system, and Mark Rosst’s Structural Deep Democracy. The idea of allowing voters to delegate their participation to individually-chosen representatives is central to all of these proposals, though details differ.
Exploration of this idea in the academic literature has been more sparse, no doubt in part because it’s hard to convince reviewers of the novelty of an idea that’s been around in some form since the 1880s or 1960s, depending on how you count. However, Rodriguez et al’s Smartocracy develops a trust network and decision ranking algorithm embodying a form of delegative democracy. Paolo Boldi et al developed Viscous Democracy, a transitive proxy voting algorithm based on Google’s PageRank and in a vein similar to Structural Deep Democracy. Wybo Wiersma explored the integration of delegative democracy into social networking systems such as FaceBook.
Taking an economics perspective, James Green-Armytage has developed an axiomatic model and analysis of delegate proxy voting systems. His work formally models the expressive loss of alternative voting systems — the collective decision-making “error” induced by voters who may have limited information about issues and/or limited choices of possible representatives. This analysis mathematically confirms the intuition motivating all the incarnations of delegative democracy: namely, that voluntary and free choice of delegation yields decisions with no more error (and most likely less) than conventional alternatives such as direct or representative democracy.
Yefim Leifman has explored ways to implement transitive delegation schemes using cryptography. This is a topic I am particularly interested in as well, since a major motivation for my Dissent project was to build practical anonymous communication systems that would be good not only for communication but also for secure deliberation mechanisms such as voting. I hope more system-builders and security/privacy researchers will be bold enough to venture into this space.
Increasing participation and engagement in political discourse has long been a basic goal of many Internet democracy efforts. Demoex, an experiment to build political parties based on online direct democracy, succeeded in Sweeden from 2002 until this year, as recounted in The Little Horse from Athens by Per Norbäck. While Demoex pursued something closer to traditional direct democracy, it relied heavily on online participation and made provision for voters to obtain voting advice from others.
Perhaps the most exciting development is that starting around 2006, free software organizations in Germany began building two different software platforms for delegation-based online discussion and deliberation: LiquidFeedback and Adhocracy. LiquidFeedback was adopted and used for the past several years by the German Pirate Party, as recounted in an article by Björn Swierczek (original German version). Further information on these tools and their adoption is now available in a book and an electronic journal.
These success stories with delegative democracy in Germany have started to get noticed and studied internationally. Articles in Spiegel and techPresident cover the German Pirate Party’s use of Liquid Democracy, and The Boston Globe recently covered the party’s attempt at a foothold in the US. There are academic studies of the Pirate Party’s use of Liquid Democracy by Friedrich Lindenberg in Germany (auto-translation), Anna Litvinenko in Russia, and Andrea Cangialosi in Italy. Back in Germany, Anja Adler at the NRW School of Governance recently interviewed me as part of her thesis studying liquid democracy; I look forward to seeing the result.
I still believe strongly in the idea of delegative democracy, regardless of what we decide to call it. Participatory approaches to democracy are important because they address a fundamental security flaw in representative democracy, which in practice allows small groups of elite political insiders to control and limit voting to choices “between Tweedledum and Tweedledee.” Delegation is important because it addresses a basic scaling flaw of direct democracy, which becomes a critical security flaw once those members of a community with other demands on their time and attention are drowned out and driven away by a vocal minority with the time and determination to do nothing but push their agenda.
I see delegation as just one of many improvements needed to address scaling and security weaknesses in traditional democratic processes, however. I am concerned that current tools for delegative democracy appear to be designed with limited attention to other security concerns. The current tools depend on a single centralized server that all users must simply trust, and offer no cryptographic protections either to ensure the integrity of the deliberation process, or to offer privacy, anonymity, or coercion-resistance to voting users. All of these issues and many others, if inadequately addressed, could easily become different but equally critical security flaws in the overall deliberative process as it scales up and meets real-world adversaries.
I found this comment by godix in a a kuro5hin discussion to be both amusing and apt. [sic]
I can easily see the following quotes if this idea ever took hold:
“Congradulations on your new high paying job at monolithic multinational corperation. Just a few forms to fill out and you’ll be earning three figures. First off is the tax form, next is your health insurance, then we’ll be needing to you proxy your vote to the CEO….”
“Hi, I’m Monica from Friends. Sign your vote over to me. I don’t know shit about politics really, but I’m famous. Thank you for giving me political power along with fame and wealth you mindless drones.”
“No child/wife of mine is going to vote while they live in my house. Sign it over or get the fuck out.”
“Hello, we know many of you are there are used to voting straight Republican/Democrat. To continue this fine tradition would you please sign your vote over to the DNC/GOP?”
“Hi, this is the President speaking. This nation decided I was qualified enough to run the free world but COngress is causing me problems. If you support me please sign over your vote. In return I’ll make sure I’m elected again and Congress isn’t. Thank you for your support.”
“Hi, you may know me, I’m Bill Gates. I’m getting sick of government investigations. The next one million people to sign their vote over to me gets a free copy of Windows 2045. This time we fixed the bugs. No, seriously, we did.”
We ultimately need not only secure, scalable deliberative processes such as delegative democracy, but secure and scalable implementations of those processes with strong, decentralized integrity and privacy protections. As I alluded to above, building the technical infrastructure needed integrity-protected, privacy- and anonymity-preserving communication — in a decentralized model where no single person, organization, or server needs to be blindly trusted — has been one of the key goals of my group’s Dissent project for the past several years.
I believe we now have all the key cryptographic tools and protocols we need to create secure implementations of delegative democracy. To oversimplify things just a bit, what remains is SMOP, a “small matter of programming.” I would be happy to work with any talented coder/activists interested in collaborating to make this happen.