I wrote this on my personal blog, mainly as a way to better articulate my thoughts. I’m posing it here as most of my conclusions come from my experiences with MakerDAO and they influence my actions as a community member.
A study of “censorship resistant organizations”
On a Friday evening I had an interesting Twitter encounter which forced me to think more critically about the purpose of DAOs. “Decentralized Autonomous Organizations” have been around since 2016, when Stephan Tual first mesmerized us with a grand vision and a fashionable scarf. Yet over the past five years, despite countless explanations of how DAOs can work, I cannot find a single satisfactory explanation as to why people would ever organize themselves in such a painfully frustrating way. Now, I happen to know a few things about DAOs. Since 2016 I have been actively involved with, in, and around organizations calling themselves DAOs. There have also been fascinating legislative developments, such as Wyoming officially recognizing DAOs as a type of entity. With much respect to the pioneering officials that have achieved this legislation, these entities are DINOs (DAOs in name only). While they serve a purpose in assisting a company with a siloed crypto token-economy in claiming limited liability, they are quite literally traditional LLCs with unconventional operating agreements. As will be discussed, DAOs are a completely new kind of entity that does not so easily fit into a pre-existing mold.
In addition to not having a broadly accepted purpose, there seems to be a pervasive misconception around how DAOs actually function. Digging into the Twitter interaction referenced above, it seems that there is a perception that DAOs may produce greater collective intelligence and efficiency. While there could be a distant future where this is true, I believe DAOs are extremely far away from becoming an efficient operating structure. Yet that’s not a cause for concern — DAOs are serving a far more important function than being a better corporate mousetrap.
To study the nature of the “DAO,” we should first define it. I’ll pose this: A DAO is the coordination mechanism of actors operating a permissionless public good. It may seem redundant to specify that a public good is permissionless; isn’t the entire purpose of a public good that it’s effectively available to everyone? The nuance is in how one defines “public.” The US military is a public good, if you’re a US citizen. US citizenship is perhaps one of the most permissioned status’ on the planet. Therefore the US military is only a public good in the context of the generally permissioned network of US citizens and those the US government chooses to protect. Public blockchains like ethereum are permissionless networks, truly anyone can access them. A public good running on a permissionless network can therefore be considered a permissionless public good. The internet and websites are a good example of this relationship, yet their utility has thus far been constrained to the flow of information. Blockchains bring this property to value. [It’s worth noting that this is a paradigm shift for civilization. Prior to the advent of the internet the only permissionless public goods were naturally occuring, like air and rain.]
The internet bestowed upon information a sort of “sovereignty,” no longer was it a good in which its flow could be controlled or intermediated. In turn the internet itself began to take on the properties of a sovereign state. After all, what is sovereignty if not the ability to resist censorship? The internet tried to spawn the first DAOs in the form of open source software, but without an economic incentive mechanism coordination was difficult. This isn’t to say that open source software wasn’t successful, but it certainly did not meet the expectations of its creators — the only open source software that could be implicitly monetized to the benefit of its contributors was that which in some way handled money, and value had not yet found its sovereignty.
With the creation of Bitcoin, and public blockchains in general, a global permissionless network of value could finally exist. But unlike information, there is a need for consensus in a value network, and therefore a requirement to have a coordination mechanism to form that consensus. There are many explanations of consensus mechanisms in blockchains, so I will not go further into the topic. The important concept is to understand where sovereignty lies. As a user of a public blockchain, one needs to be sure that the accounting of the ledger isn’t being manipulated and that transactions are not being censored. The sovereignty of a value network can only be present if its users can independently verify this information. If only some users can prove that their transactions are not being censored, all others users are subordinated to their power. Thus the only way that a blockchain can take on the properties of a “sovereign state,” in a way similar to the internet itself, is to be independently verifiable by the vast majority of its users. This has had serious implications on scalability and the amount of information that a blockchain, at its base layer, can be expected to process. This same concept applies to DAOs, but we’ll discuss that later.
With this context, let’s further explore the nature of a DAO. DAOs exist to coordinate actions in a permissionless public good. Permissionless public goods, outside of those that are naturally occurring, must exist in a permissionless public network. The internet and public blockchains, specifically the Ethereum public blockchain, are two permissionless networks that give their users sovereignty over value and information. They are in a sense, a new form of State.
A key role of the State is to enforce property rights and contracts. Any study of empirical data will show that a country with strong property rights and contract enforcement has probably fostered a more prosperous society than a State which does not have these qualities. Money, being a form of debt, is almost entirely dependent on the State’s ability to uphold private contracts. The long term strength of a country’s money is intrinsically linked to this principle — i.e. good money is based upon deterministic outcomes. This adds a level of complexity to our line of thought; the sovereignty of the individual is only as strong as the determinism of the network they rely on for their contracts, and therefore the sovereignty of the individual can be in opposition with that of the State, which has its own sovereignty and will. Public blockchains have overcome this problem by creating a completely neutral base-layer where users, as long as they can verify the chain, can always enforce their sovereignty over the “State.” A contract on a blockchain is maximally deterministic and therefore stronger than any guarantee made by a legacy government. It’s no wonder that blockchains are being used primarily to make better money and financial services. Yet this superpower comes at a cost — a deterministic outcome requires an objective input.This limits the utility of the blockchain to applications that do not include any subjective measurements or decisions, which is quite a severe limitation. This is where DAOs enter the picture.
Today the primary use case for a DAO is to build, maintain and update open source software (smart contracts) on a public blockchain. DAOs are most useful when these smart contracts require subjective inputs to function appropriately. By running on a public blockchain, smart contracts are implicitly permissionless public goods. Therefore, if deployed without any kind of administrative powers, they are fully censorship resistant. Since the value of a public blockchain is in its censorship resistance, any smart contract with administrative functions must have a mechanism to preserve this property. Thus we can expand our definition of a DAO to being a coordination mechanism to maximize censorship resistance. While a legacy organization organizes itself in a way which maximizes the efficiency in achieving a specific goal (e.g. profit, impact, control), a DAO organizes itself to maximize resistance to censorship in achieving a specific goal. A DAO is a censorship-resistant organization. Since a DAO must ensure its public good remains censorship resistant and continues to provide deterministic outcomes, it must logically resist censorship of itself. If the DAO itself is compromised, all of its goods and services become equally compromised. To achieve censorship resistance, a DAO must properly align the incentives of its users and bestow sovereignty into a fundamental unit of power. It must then distribute these fundamental units of power in a way that maximally diversifies away the potential for censorship while not breaking the incentive alignment. In most nation-states, the fundamental unit of power is the individual who is able to express this power with a singular vote. The fundamental unit of power in a DAO is the token. Tokens are not the ‘expression of power’, in the sense that they have the right to vote, but are the unit of power itself, the ‘individual’ in the analogy. They are explicitly not comparable to a share in a company. Traditional shares do not exist in a permissionless deterministic network, and are therefore not an extension of the sovereignty of that network but a unit of power subordinated to the State they exist in. Since DAO tokens exist on permissionaless public networks, which bestow sovereignty and determinism into the code running on them, the will of the token holders in a DAO is law.
As indicated above, by placing administrative rights into a token which exists on a public blockchain, pass on the censorship resistant qualities of that network to the token. From here we must determine the best way to use these tokens without diminishing this censorship resistance and thus negating the entire purpose of the DAO. This forms the fundamental theory of the DAO — to preserve censorship resistance, all sovereignty must remain with the token. Note that this concept is not being applied to token holders but to the collective token supply itself. A token holder can easily be censored, but with the correct mechanism design it may be nearly impossible to censor the collective. Yet this causes the same point of contention that we see at the public blockchain base layer between user sovereignty and scalability, in that DAO scalability becomes a direct tradeoff to token holder sovereignty. In an ideal world token holders would be able to process an unlimited amount of information and intelligently vote on each parameter change in a system. They would also be able to do so reliably and without intermittent periods of censorship. This is clearly not a reflection of reality, especially in an adversarial environment. While public blockchains cannot scale beyond the average computer processing power of its users without sacrificing censorship resistance, a DAO cannot scale beyond the average brain processing power of a human being. Most scaling solutions for public blockchains address this issue by having layer-2 solutions compress their data and commit it to the main public chain, with the important caveat that the entirety of the original state can be reconstructed and challenged at will. A similar concept can work for DAOs; any decision which is delegated outside of the control of token holders may be condensed and presented to token holders as a ‘conclusion’, but the entirety of the inputs must be made public and deployed to a decentralized network for preservation and redundancy. There will be circumstances where this is not possible due to the interaction between the legacy State and the DAO (i.e. legal constraints, and a DAO should NOT attempt to extend its sovereignty beyond the public permissionless network on which it runs), but a DAO and its token holders should understand that each bit of information which is siloed outside of token governance effectively leaks sovereignty from its own token. This same concept applies with even greater significance to control. Any control ceded from the token to an outside party should be considered a direct loss of token holder sovereignty and weighed accordingly, as the actions of this outside party may be censored. So in this context, we can say that transparency is the antidote to complexity. However, given the generally subjective nature of DAO decisions, transparency is only as effective as the amount of time provided for a community to process information and debate. Therefore I draw the conclusion that a DAO can only be effective at scale if all information is public and its governance is exceptionally slow.
With the key to censorship resistance being in token sovereignty, token holders must have a direct and not tangential incentive to govern properly. This means that they must be explicitly rewarded for good governance, and explicitly punished for bad governance — with the distinction between the two being an objective and hard-coded measurement. Since censorship is a total failure of governance, it is additionally beneficial if the protocol has a built in distribution mechanism in order to dilute the existing token holder pool should they be unable to govern due to censorship. For example, if a system begins to fail because parameter changes are being censored, the protocol must have a way of automatically redistributing its units of power to new un-compromised participants. These mechanisms may be unpopular with existing token holders, but systems are likely more fragile without it.
Another consideration for DAOs is how resilient they can be to a successful censorship attack. If an attacker can prevent the protocol from receiving inputs, for how long can the DAO survive? This introduces another principle, and one that is relatively well accepted in today’s DAO-scape, that DAOs should attempt to minimize governance. The number of interactions with token holders can be considered to exponentially increase the likelihood of censorship, simply by making it more effective for an attacker to execute a “siege” on the DAO. For this reason, I have been an advocate of “set it and forget it” models whenever possible. More resilient DAOs will also have a thorough contingency plan to survive a censorship attack.
In conclusion, the concept of a DAO is still in its early stages, but it would be a mistake to discount these organizations just because they are young. It’s hard to say what the future will look like, but I’d bet it’s full of DAOs. Society craves determinism, DAOs deliver it. This shift should not be particularly surprising either; we (i.e. humanity) have already seen the feudal/agrarian model organization give way to the corporate form. This happened when the means of production shifted from land to factories and other industrial mechanisms. Now the means of production are shifting to smart contracts, and in the not-too-distant future DAOs will be as common as corporations are today. We will ultimately be the generation of the permissionless public good, the generation of the censorship resistant organization.