On-chain governance vs social governance, and why do they have to be handled very differently?

• DAO tokens only control the code on the chain, where the code is the law.
• DAO tokens do not control people (coders, validators, liquidation bot operators, businesses, etc.), but can be used for off-chain sentiment signaling where law is law.
• Developers and others cannot respond to on-chain stakeholders (DAO token holders, etc.) People who make agreements carry huge legal risks.
• Developers must keep any major work they do on the protocol confidential to avoid legal obligations to become token holders and subject the entire protocol, the DAO, and all participants to onerous financial regulations.

The main purpose of the protocol DAO

• The main purpose of Protocol DAOs (and the "governance tokens" that make them up) is to give users of autonomous digital infrastructures (i.e. smart contract systems) a say in whether and how any potentially changing functionality of these systems can be changed. This is achieved through direct, binding on-chain control of these systems by the requisite majority vote of DAO token holders. In the MakerDAO community, these are called "execution votes".
• In reality, Protocol DAO is just a massively multiplayer online game for tuning the parameters of an ownerless, decentralized software system. This is also why DAO tokens are primarily distributed to users of these systems (via liquidity mining, etc.) - users need to have a strong say in the systems they rely on.
• For on-chain governance, DAO tokens and DAO voting are "God Mode" - they are the first, final and only authority, operating under the principle of "code is law". Whatever quorum and majority rule rules are written into the code, they must be strictly followed.

Secondary Purpose of the Protocol DAO

A secondary purpose of Protocol DAOs is to participate in off-chain social coordination based on rough social consensus on issues more loosely related to autonomous systems. This includes chat, "governance" forums, Twitter -- all the social media platforms where the community communicates around the system. In this case, DAO token holders may sometimes vote on various things that they believe should happen on the social layer — such as writing a new major code upgrade, or the community adopting certain social goals. MakerDAO refers to these as “signal votes,” which include unanimous expressions of social value, such as “green” environmental technology funding initiatives.

• These votes are not binding - they represent sentiment. No person, group, or business is required to "follow" the outcome of these votes, and votes are not subject to the strict requirements of a quorum or specified majority criteria. They are simply expressing the sentiment of the community on a topic.
• Furthermore, these votes do not even represent the sentiment of the entire community, but only the sentiment of governance token holders. Since the full community around a protocol is usually broader — including bot runners, developers, and even validators/miners on L1, etc. — governance tokens cannot be the sole authority on these social topics. They are just one input, albeit an important one, in the larger process of "rough social consensus".

Confusing primary and secondary purposes is legally disastrous

• Unfortunately, many people either confuse primary and secondary purposes, or expect secondary purposes to work differently. For example, they want governance token holders to have binding authority over certain people in the social layer. There are many reasons why this doesn't work and why it's a dangerous and bad idea. I'll focus on explaining the legal reasons, although there are non-legal ones too.
• Assets that give their owners binding social voting rights are legal contracts—in most cases, highly regulated “securities.” A company's stock is regulated in large part because it carries specific legal rights, including the right to appoint company directors. Elected trustees have specific legal obligations to shareholders. For example, directors of a corporation have a fiduciary duty to make corporate decisions with one goal in mind: maximizing shareholder value. If they don't follow through with this goal, they could be prosecuted.
• Since DAO tokens are not regulated, treating them like company shares or other securities would violate many financial regulations. Worse, unlike the company's directors, the so-called trustees in this case would have no insurance, no protection and unlimited liability - a very bad situation for software developers and others involved in the agreement the result of. Therefore, everyone involved in a DAO must handle and respond to sentiment votes very carefully.

Example - code upgrades and how to handle them

Let's take an example:

A signal voted support for a certain software update to the protocol, but no one has yet coded that update, it's only described conceptually, and it's not even clear that it will work. It would take a lot of time, resources, and talent to code, test, and deploy the upgrade, and the code could fail completely and never be adopted.

In the United States, the Howey test says that an asset may be a security if one group relies on the entrepreneurial efforts of another group to realize its value. Therefore, by law, if a team announces that it will upgrade within the next 6 months, it will essentially be interpreted as a commitment to token holders to complete the work required by the emotional vote. DAO tokens become something more like company shares - securities. This is true even if the team is a new one, not the one that originally built the protocol — the SEC also refers to these new teams as “active participants” who could be held liable under securities laws. Not only are these activities risky for the development team, but, as we saw in the recent CFTC v. Ooki DAO case, they are also detrimental to the DAO itself, as each participant in the DAO could be accused of running an off-chain business.

In this example, if the development team wanted to build the idea of social support for code upgrades, what should they do?

In the new normal where DAOs and DeFi face enormous legal risk, development teams must learn to work in secret until the work is completed and submitted to a binding on-chain vote. From an outsider's perspective, it's impossible to know if they're working on something unless the work is done. This achieves two important goals:

• Reduce legal exposure to the team and others in the community
• Avoid speculative pump and dumps, where the value of an experimental potential software upgrade is built into the token price so that if the final upgrade doesn't materialize or doesn't materialize, the price will crash.

Restricting formal governance to on-chain issues isn't that bad, and is common in practice

As frustrating as the above dynamics may be, this is why smart contract systems must be open source. This way, anyone can work on the protocol, and if they submit concrete actual code to put on the blockchain, they can have a binding governance vote on it. However, for social proposals, governance votes are merely expressions of sentiment and do not guarantee any particular outcome.

Many successful protocol communities take this attitude - for example, Ethereum has achieved great success despite having no formal protocol governance, via rough social consensus, "core developers" who do not respond to ETH holder votes . The same is true for Bitcoin. Admittedly, the DeFi community is somewhat inconsistent in this regard, but, for example, the "Yearn Manifesto" clearly states that "Yearn is governed by YFI, but YFI does not govern Yearn contributors" . We’ve seen protocol communities take a different stance and get punished by governments (like the CFTC lawsuit against Ooki DAO).

epilogue

We have entered a new normal in which anyone involved in liberal technology faces enormous legal risk—similar to the risks faced by developers of encryption technology during the “ crypto wars ” of the 1990s. This requires everyone involved in DeFi — from casual users, to degens, to programmers, to bot operators, to validators, to CEXs, to social media leaders — to step up and be more aware of what they are doing. What to do, how to do it, and possible legal consequences. In the long run, this is best for all of us, because it forces us to think about what this technology and associated social formation - The DAO - are really supposed to achieve, and how best to achieve those goals, rather than Fall back to old TradFi models like corporate governance.