One of the more interesting consequences of the decentralized finance (DeFi) boom is how it revived the five-year old concept of the DAO, or decentralized autonomous organization. DAOs are a fairly wide umbrella term for any type of enterprise organized primarily through blockchain smart contracts. Like all things in crypto, DAOs carry a lot of promise, but their overhyped picture fails to stand up to deep scrutiny. Even worse, some DAOs are not decentralized, not autonomous, and sometimes they’re not even real organizations. But more than that, there are certain cases where centralized organizations offer a genuinely better alternative.
Perhaps the most famous example of a DAO is, well, The DAO. Despite the somewhat unoriginal name, The DAO was supposed to be a truly groundbreaking concept: basically an on-chain investment fund managed by the Ethereum community aimed at growing the ecosystem.
The DAO got hacked in 2016 while holding 15% of the entire Ether supply. The extent of the damage was so large that it resulted in a significant political mess and the hard fork that created Ethereum Classic. In brief, the community (represented, in large part, by the Ethereum Foundation) proposed to manually adjust the blockchain so that The DAO money could be returned to the original holders. The highly controversial proposal fractured the community into “code is law” maximalists and pragmatists, the former migrating to Ethereum Classic, the original chain. The pragmatists who maintained the Ethereum brand, of course, seem to have ultimately won.
With the bitter disappointment from The DAO, the entire idea temporarily fell off the crypto community’s radar (although there are a lot of true DAOs outside Ethereum that don’t use that term). The pure ideals of The DAO were eventually replicated by the more “centralized” phenomenon of ICOs, or initial coin offerings, which allowed startup founders to collect money for their business ideas from the community — usually this was done on Ethereum. Nexo was one of hundreds such projects.
Now, DAOs are back in the public’s focus, and its staunchest proponents will often say how they are the new paradigm of human organization and governance, or similar hyperbolic statements. Now, hopefully we can all agree that doing the same things on a different platform is hardly a paradigm change, so to confirm the hype we need to look into other potential forms of innovation brought by DAOs.
Decision-making and politics
One way we could imagine DAOs to be an astounding innovation is if they would somehow rid us of politics. Imagine if every decision could be objectively taken by an impartial algorithm that simply takes some data and inputs a clear response. No emotions, no biases, just an algorithm that instantly reaches a decision without the endless debates and attempts to persuade each other.
Unfortunately, this utopia clashes with a fundamental fact of life, which is that there are often no certain answers to a particular question. Imagine a computer trying to deal with a real-life trolley problem: a pandemic virus approaching your country. Let’s say, as a simplification, that there are only two choices: to do nothing, or to lock down the population. In the former, thousands of people will die, while the latter will cause immense economic damage.
The problem isn’t even so much that both are hard choices. The computer could come up with a monetary cost for everything, including loss of human life, and just pick the smaller number. The real issue is that the computer cannot have all the data it needs to make an informed decision. In this situation, virtually nothing is known about the virus: how many lives are saved (or indirectly lost) by locking down, how many lives are lost by not doing anything, and what could be the ultimate consequences of both decisions.
Uncertainty is the reason why politics must exist when making decisions. It’s usually very ugly, because most of our issues cannot be solved objectively. As long as that remains the case, DAOs themselves will always be political. And that presents a serious problem for cryptocurrency startups, the most prevalent category of organizations where DAOs are being used today.
Why you can’t always afford to be political
There are tons of decisions to be made when building a company and a product. Answering questions like, “Which design do we use?” or “Is this feature worth the extra cost?” and “What price do we charge?” is а large part of a CEO or product manager’s day. Very few of these decisions can be made with absolutely certain input data.
As a rule of thumb in corporate product design, the more people are involved in making a decision, the more time it will require. A corollary is that, most of the time, a collective decision will be worse than that of a single qualified person. There’s a reason why the term “design by committee” is a pejorative.
Decisions made by a large group often turn out to be worse simply due to the disparity in opinions and necessities. This political process often involves making extensive compromises to please everyone, but unfortunately compromising in a particular direction all too often turns out to be wrong. In a business context this often leads to losing to a more decisive competitor.
For this reason, early-stage startups are almost always dictatorships. There needs to be at most one decision-maker, because stakes are so high that one delay or one slight misstep may be the difference between success and failure. Then, over time, startups that succeed generally become more “decentralized” and committee-driven.
Of course, I don’t intend to advocate for one-party rule in general, and the flipside of relying on one decision-maker is that he or she just may not be qualified to make calls in every situation that arises, hence a committee is actually a balancing force for good. But there are contexts where one approach fits better than the other, and DeFi and crypto are still closer to the “single decision-maker” stage.
Leaders always emerge
All the DeFi DAOs that began forming last year would apparently disprove my argument that early startups must be dictatorships in order to be successful. Except, if you look closely, they are actually embodiments of my point.
Beneath the Kumbaya chants of decentralization, the vast majority of the protocols are steered by a well-defined and powerful team of “core contributors” that are usually unelected. Sometimes it’s quite obvious, for example when an existing team makes their project a DAO — while retaining all control of development and “graciously” putting up irrelevant matters for debate. Certain DAO-native projects, for example Yearn and Sushi, are also quite open about avoiding too much politics in general protocol development.
Perhaps the best example of a truly decentralized and automated DeFi DAO is Compound, though its model is certainly closer to an oligarchy than democracy. Compound has a true class of “protocol politicians” who regularly submit governance proposals, which are engineered to be the only way to upgrade the protocol. It is worth noting that the Compound Protocol has only had a precious few changes ever since it released COMP, as governance proposals mostly deal with adding new tokens or other minor items.
Ultimately, having central custodians is unavoidable when flexibility is key. Examples like the Ethereum DAO hack response, or Yearn’s governance structure, show that even the most decentralized of protocols rely on effective “custodians of the community’s will.” Unfortunately, this exposes DAOs to the dilemma of “Quis custodiet ipsos custodes,” or how to guarantee that the custodians of power don’t abuse the inherent control that their position provides. Time and time again we see that subverting the custody of power (for example, controlling the election process) is the most effective way of subverting power itself.
The case for honest centralization
While decentralization of power is a complex issue, I don’t believe that simply putting everything on a blockchain is the answer.
The problem of fair custody of power is one too difficult to solve without fallible human intervention. Even if smart contracts definitely won’t create a perfect automated decision-maker, perhaps they may save us in the future and create a system wherein custodians cannot subvert the established process. Regardless, today that’s not the case — most DAOs use Snapshot, a centralized platform where users vote by signing with their wallet’s keys. It’s a good way to do sybil-resistant polling, but there is no ability to enforce that the team or the multi-sig holders execute the governance proposals.
Furthermore, centralization has one clear benefit: accountability. For example, should users lose funds, companies like mine would be legally liable to return them to these clients by any means necessary.
Contrast this to incidents like MakerDAO’s Black Thursday, where customers lost millions in unfair liquidations because someone was able to bid almost 0 DAI for each ETH. Call it whatever you want: faulty keeper software, ineffective keeper community, lack of liquidity, unforeseen gas fees… The ultimate issue was poor protocol design, and thus Maker’s fault. However, the community hid behind the veneer of decentralization to completely refuse compensation.
There were many arguments thrown to justify the refusal. One was that the protocol was permissionless and that anyone could use it, which somehow meant that Maker wasn’t responsible if something went bad (many physical products are also permissionless, but their manufacturers are still responsible if they fail). Another argument is that it was a failure of the market, meaning that the users wouldn’t have gotten as much as they thought anyway (but why did they get zero, then?). Finally, some community members felt that it was the Maker Foundation that had to pay, while the Foundation did not think so.
As the community discussed the finer philosophical points of whether the users “deserved” to be compensated, few cared about the message this was sending to current and future users, and how it would affect trust in crypto and DeFi. The liquidation ecosystem was later improved significantly, which was a silent admission that the Black Thursday liquidations were indeed not proper behavior. The aftermath of that crash was a clear example of a decentralized political process falling to egoism and bias.
If we want crypto to reach serious mainstream adoption, taking responsibility for mistakes is absolutely crucial to establish trust. For now, only centralized finance (CeFi) guarantees this level of accountability.