Gerrymandering and Democratic Failure: Designing DAO Security to Learn from the Past
The greatest sin of tech, according to some, is its constant desire to reinvent all things, whether or not the reinventions will actually be better. The Decentralized Autonomous Organization is one of these very ambitious reinventions, it has the potential to replace the many forms of government we interact with every day. But it can only get there by avoiding the failings of the structures we already have. We have an obligation to learn from the past to build our new cryptographically governed world. This article is going to be a story about a failure of representative democracy in the United States, how that could be used to attack a DAO, and what it will actually take to secure cryptographic governance.
The structure of representative democracy is simple: you group voters and groups of voters elect people to write the laws via pure democracy. In the United States one of the legislative bodies, the House of Representatives, is organized in this way. The goal of the House is to represent the people of the USA democratically but to be assured a bare majority in the house the Democrats would have to win 11% more of the electorate than Republicans [https://www.brennancenter.org/publication/extreme-gerrymandering-2018-midterm].
How did the USA reach a point which is an apparent corruption of democratic principles? The answer is gerrymandering, the process of politically motivated individuals picking the groups of people who elect each representative. Since groups of electors are selected based on geography, this often results in districts that look quite strange visually. Perhaps because of that odd shape, mathematicians who study geometry have constructed models which can easily quantify gerrymandering based on the compactness of the district. Using these compactness models of gerrymandering it would be possible to eliminate the process in the United States, but that’s not likely to happen in the US.
Gerrymandering has existed since 1812 when Governor Elbridge Gerry drew a district to entrench his political power that looked like a salamander. However, computers have supercharged the practice by allowing millions of potential districts to be tried and the ones optimal for any party to be picked. Now we are trying to rebuild governance using cryptography and blockchain, so imagine that there is a representative democracy built in an Ethereum smart contract which allows before each election the selection of districts by submission of a supposedly random set of groups of on-chain identities. Can we gerrymander this DAO? Absolutely!
We live in a time with more data than ever before, and the on-chain identity will likely have a data pool associated with it. We can fire up a machine learning algorithm like SMV or random decision forests to take each high dimensional datapoint and classify them into categories [the districts from ye old gerrymander] using a payoff function which is that the classification is scored based on how likely it is to give one group control. Depending on how good of a machine learning engineer you are, this gerrymander is actually going to be much more efficient than any seen before. Moreover, this gerrymandering is even easier than before because we aren’t even bound to make voters look closer together geographically!
Even though we have located the organization on-chain, we still have to deal with the same problems that plague old fashioned governments. But to fit the new implementation of the government, I offer a new lens to view these problems through: structural and electoral flaws in old governments are cybersecurity flaws in DAOs. A cybersecurity flaw occurs when the code of the project doesn’t match its intention, the rules as written allow the corruption of the ideal. That’s why securing a DAO is different than securing other projects, the cybersecurity encompasses structural security of the institution’s ideals.
Smart contract auditors and security professionals, like myself, have an obligation to report to the users of the DAO about structural problems that exist which may prevent them from gaining the desired form of governance. Since cryptographic DAOs answer to the users, they can demand changes to preserve the security of their ideals and freedom within governance.
There is a problem with this though. If we revisit the question of why the United States doesn’t fix the problem of gerrymandering, we can see it. Simply: the politicians who benefit have the power to make the rules and set their enforcement through the courts. If the institution securing the DAO is not both completely apolitical in the DAO and completely free of conflicts of interest, then it cannot fill the role of securing the ideals of the DAO.
Imagine if the United States had an independent body that answered and reported directly to the voters whose whole job was to independently and apolitically protect democratic ideals, eliminate the potential for structural corruption, and secure the operation of the government. It wouldn’t fix all of the problems of the nation-state, but it could fix many of them. That is the potential role of a security firm that works with a DAO. We can work directly for the DAO’s voters preventing the pitfalls of the governments we see in the world today and securing the future of the DAO.
This content has been produced in conjunction with our proposal to become the Aragon Network Security Partner under AGP-18 which, pending AA review and approval, will be up for ANT vote on January 24th. If you are interested in continuing the discussion about these topics, please join us on January 19th for our public AMA call (see twitter & aragon.chat for details). Either way, please make your voice heard and vote on January 24th!
