The Problem of Procedural Perversity
It is said that in the course of studying for his citizenship test, the mathematician, logician and philosopher Kurt Gödel discovered an “inner contradiction,” or loophole, in the U.S. Constitution, whose discovery threatened to derail his path to citizenship. Though the official formalization of this contradiction was never put down on paper, it is suggested that its substance concerns the fact that Article V of the Constitution, which describes the process for amending the Constitution, can be applied to itself.
Notably, Article V contains an explicit restriction on what can be amended in the Constitution:
“…provided that no amendment which may be made prior to the year one thousand eight hundred and eight shall in any manner affect the first and fourth clauses in the ninth section of the first article; and that no state, without its consent, shall be deprived of its equal suffrage in the Senate…”
As Article V does not explicitly bar application to itself, however, the procedural barriers to amendment of the Constitution can themselves be eviscerated, leading to a greatly increased possibility that the Constitution can be amended. Gödel, having escaped from Nazi Germany, was understandably fearful of this possibility and worried that the weakening of safeguards to amendment via Article V could easily turn a democratic model for self-governance via elections into an autocracy.
Putting the merits of Gödel’s fears to the side, Gödel’s loophole is an example of what we might call the problem of procedural perversity, where the procedures for self-governance do not adequately guard against the possibility of bad actors. The problem of procedural perversity is, however, not limited to the U.S. Constitution, and can pop up in any system of self-governance, especially where the system of self-governance is not drafted by writers as illustrious as our Founding Fathers.
Procedural Perversity and DAOs
Consider Decentralized Autonomous Organizations (DAOs), which are a form of self-governance based on a smart contract in which the power of governance is distributed amongst participants, as opposed to being held by a single separate centralized entity. One prominent example of a DAO is the Decentraland DAO, which oversees the governance of the Decentraland metaverse. Governance determinations of the DAO are made by a series of three votes, where a proposal must meet a minimum level of Voter Power, and also achieve a simple majority of Voter Power in each vote, in order to be advanced and eventually adopted as a governing law. Decentraland participants each have a Voting Power based on their ownership of Mana (Decentraland’s cryptocurrency), Land and Names. When a Governance proposal is put up for a vote, Decentraland participants have the opportunity to vote on the proposal within a certain period of time. If a sufficient amount of Voting Power is utilized, and the Voting Power in favor of the proposal reaches a simple majority, then the proposal is advanced to the next vote in the series – or if a proposal reaches the final vote in the series, it is adopted as the law of Decentraland. If a proposal either fails to reach a sufficient amount of Voting Power or the Voting Power in favor of the proposal does not reach a simple majority, for any vote in the series, the proposal is rejected.
Notably, Decentraland’s use of Voting Power is itself intended to deal with a problem of procedural perversity associated with account-based voting systems, in which each blockchain address gets one vote. As noted in Decentraland’s DAO guide, if voting was distributed giving each blockchain address one vote, then bad actors might try and subvert the DAO voting system by simply creating thousands of addresses. By moving to a voting system based on Voting Power, this is prevented, since the creation of thousands of addresses will not by itself increase an individual’s leverage. Of course, basing determinations on Voting Power still creates a system in which some individuals will have more leverage than others. Individuals could simply buy up several million MANA, thereby giving themselves a greater amount of Voting Power than an individual who owned only one MANA. However, a system based on Voting Power arguably distributes leverage in a way that seems plausibly fair in some way: The system gives individuals a say in how the world is governed in direct proportion to the extent to which they have invested in the world.
Of course, this is not the only area in which there might be an issue with procedural perversity in a DAO. Consider, for instance, that under Decentraland’s Terms of Service, disputes over intellectual property are determined by the Decentraland DAO via a vote. However, nothing in the Decentraland DAO or the Terms of Service appears to prevent an interested party from purchasing an inordinate amount of Mana and using it to procure outcomes favorable to itself when a dispute over intellectual property in which it has an interest is presented to the Decentraland DAO for determination.
There are also potentially problems similar to the one that Gödel feared. The earlier example involving Gödel’s loophole was an example where the procedure for amendment could be applied to itself, thus allowing for the possible evisceration of the barriers to amendment of the Constitution, leading to a greatly increased possibility of an autocratic government. Similarly, the Decentraland DAO does not appear to contain any bar restricting application of its voting mechanism to the requirements for passage of a Governance Proposal. So, a Governance Proposal could, hypothetically, revise the minimum threshold for passage of a Governance Proposal, for instance, which might arguably go against the intentions of the initial DAO creators.
Of course, while each of the above illustrations indicates a way in which a DAO’s procedures may be subverted via a loophole, it is a related, but different, issue as to whether said loophole represents a substantial problem to be fixed. For the purposes of this blogpost, we will not delve too deeply into any loophole utility analysis. However, a good analysis would seemingly involve, inter alia, an examination of 1) how likely the loophole is to be utilized, 2) how disastrous the consequences are of the loophole being utilized, and 3) the benefits of maintaining a system with the loophole. Moreover, while one might be tempted to analyze a DAO by direct comparison to real-world democracies such as espoused by the U.S. Constitution, it is worth noting that such analysis has its limits, as DAOs are not like typical real-world constitutions in important ways.
Looking at the Decentraland DAO, for instance:
- As noted before, unlike with a real-world democracy, individuals can amass Voting Power based on one’s degree of ownership in the system. This is in contrast with most real-world elections, in which voting is restricted to identified citizens of the nation, and individuals are restricted to having one and only one vote, regardless of ownership.
- Unlike with many real-world democracies, DAOs such as a the Decentraland DAO involve direct voting, where individuals can vote on individual issues in order to determine governance outcomes. This is in contrast with a representative voting system, where individuals primarily vote on representatives to make determinations about governance outcomes.
- Unlike with many real-world democracies, votes in the Decentraland DAO also have the potential to be either anonymous or non-anonymous. For any entity whose blockchain address has been identified, the exercise of Voting Power is never anonymous. This is because, at least in Decentraland, voting is partially mediated by the Snapshot tool, which publishes how every blockchain address votes for a particular Proposal and the amount of Voting Power expended. At the same time, for any entity whose blockchain address has not been publicly identified, the exercise of Voting Power is anonymous. As a result, it is possible for unknown entities to exercise an inordinate influence on governance outcomes.
How these differences may affect an analysis of a DAO’s procedures is unclear, but the fact that the Decentraland DAO may allow for an anonymous entity to buy up inordinate Voting Power, and vote directly on issues in which it has an untoward interest, may give one reason for some concern.
Representative DAO Voting?
Notably, Decentraland appears to be addressing some of these issues through the encouragement of something akin to representative voting. Under the Delegation program, citizens of Decentraland can delegate their Voting Power to one of a number of designated Delegates, where Delegates are chosen based on an application that includes specific information about the Delegate. The intent of the program is to encourage “a more healthy [Voter Power] distribution” by allowing more passive participants in Decentraland to transfer their Voting Power to Delegates who are active in the Governance process and who have an interest in the success of Decentraland. As Delegates are chosen by a Decentraland participant based on some evaluation of the Delegate, the voting of the Delegates arguably represents in some way the views of the Decentraland community as a whole. Moreover, in the case that Delegates with a bona fide and impartial interest in the governance of Decentraland are chosen, this would seem to lessen the effects of possible procedural perversity, especially where the Delegates control a very large amount of Voter Power.