When we designed Infinex governance in 2023, we started from first principles and tried to avoid the mistakes of early DeFi governance. Unfortunately, we started from the false premise that Infinex was a protocol. It turns out it's not a protocol. Infinex is an Application. We have learned a lot in the last two years, so it is worth once again assessing the governance problem from first principles. In this post, I will propose what an optimal Infinex governance structure could look like.

Governance is about establishing the meta-rules of the game. We must assume in a changing world, rules must also change. This assumption necessitates a legitimate mechanism for modifying the rules. Establishing these meta-rules is critical. Without clarity over the rules, no one will play. In crypto, the game is often capital formation and allocation. Crypto platforms are typically money games. This means using a platform typically requires an assessment of risk and/or allocation of capital. This is not too dissimilar to early banking (and arguably modern banking), where the bank you chose required a careful assessment for solvency risk. FDIC insurance and other global insurance schemes reduced the necessity of this over time, but have not completely removed it, especially for large depositors. See the recent (and upcoming) regional banking crisis for a modern precedent.

So, whether you are a user or an investor, the rules of the game are critically important. I will add another point that I think was lost on the early DeFi governance enjoyoooors: the simplicity of the rules is a critical advantage. Unfortunately, early users of DeFi found inscrutable and complex rules to be exciting and challenging for reasons of weaponised autism and PvP money games. This led to perverse incentives and an escalation of governance complexity.

If simplicity is one of our goals, then an elegant solution is a set of rules where those delegated authority to build the project have broad discretion with clearly defined guardrails. This is a major benefit of a non-custodial platform, because if every participant can exit freely at any point, we do not need to have undue complexity and layers of intervention and mediation.

We then need to delegate authority to someone(s) to operate within this framework. Currently, that person is me, although with significantly more complex governance processes. If we opt for simplicity, we can delegate authority, within a predefined framework, to run the project with broad discretion, circumscribed by one explicit rule: no action taken, either directly or indirectly, can threaten the platform's non-custodial posture.

But wait, is this a good idea? Do we want a single person or group to have broad power over a project? My response, with the benefit of hindsight, is overwhelmingly yes. We have clear evidence that DAO governance and other forms of governance by committee, in general, are absolutely diabolical. Many of the best founders in the space handed control to committees of apparatchiks for both ideological and regulatory reasons. Saying this grand experiment has gone poorly is being generous.

Now, I am on record saying the best ideas from early Synthetix came from the community, and I stand by this. Without debates and discussions between Degenspartan, Arthur0x, Andrew Kang (he is evil, not stupid), Nocturnalsheet, DeltaTiger, Alex Svanevik and many, many others, the project would have died in 2019. But we must understand diffusion of responsibility. In the early days of Havven/Synthetix, as the director of the foundation, I was almost solely responsible for deciding what to prioritise, but the input into this process was the community.

Governance was designed to signal that the input from these community members would be listened to, which created a virtuous cycle where many more people with good ideas stepped up to provide feedback and critique. Feedback, ideas and critique are necessary but woefully insufficient for a project to succeed. You also need people with a mandate and the ability to decisively execute on these ideas. This was the superpower of early Synthetix. We had a small team of ten people, we were brutally efficient at prioritising, and we constantly engaged the community in the process of both deciding what to do and how to prioritise.

This is very different from seeking community feedback. That is helpful, but it is a completely different process and far inferior to genuinely engaging a deeply aligned and valuable community. For Synthetix, this was all done in Discord. I lived there day in and day out, and read and responded to almost every message. Today in Infinex, we have a filtered channel in Telegram for Patron holders, which reduces the noise significantly, although arguably it reduces participation. I'm aware that one of the elements that drove participation in the early SNX Discord was the fact that everyone knew they had direct access to me and the rest of the team. And the degen trading channel.

Many of the issues with DAOs over the last five years have arisen from incentive misalignment; the principal-agent problem is pernicious in crypto governance frameworks. When we (the proto-DeFi community) began designing DAO governance in 2017/18, it was not about regulatory arbitrage; it was about removing central authority and trusted parties. It was only later that governance was co-opted and corrupted as a regulatory arbitrage optimisation vector as a necessary defence mechanism from overzealous and unscrupulous regulatory regimes.

An aside on governance philosophy: Ethereum needs credible neutrality because it evolves; Uniswap doesn’t because it is immutable code. Aave needs more credible neutrality than Uniswap because it is upgradeable, so it is critical that systems are in place to ensure users do not lose funds. This requires some kind of onchain token-based governance. Infinex is a platform; the onchain contracts are upgradeable, but they are not monolithic. This means every user must opt into upgrades. This is a huge advantage over a defi protocol that has a single set of contracts controlled by governance. This enables an ecosystem of many non-custodial “superapps”, all optimising for user experience and functionality. Because they will each be able to move far faster than if they were governed by DeFi-style governance.

So, we want wide input but narrow execution. At least within Infinex, we have a good framework to enshrine decision-making with a delegation of authority from token holders to a team that is paid to be 100% focused on the project and whose compensation is dependent on its success. We have done this very well at Infinex; we have a team of 50 people who are all very aligned through Patron allocations.

Let’s now discuss revenue distribution. Over the years, we have seen many examples of revenues bypassing tokenholders and going to an equity company or being completely co-opted by insiders. Friendtech is one of the worst examples of this kind of grift. So to continue optimising our governance system for simplicity, we must agree on how revenue will be distributed up front and enshrine this into governance.

Historically, very few crypto projects have generated meaningful revenues, especially compared with sales of tokens. This has created an incentive for many projects to focus on selling tokens rather than pursuing revenue. This takes many forms, and again, examples are useful here. DYDX held what amounted to a multi-year ICO where they sold tokens in exchange for generating fees on the platform. This looked like a simple volume incentive plan, but the issue was that the fees generated through these incentives did not get returned to token holders; they went to a foundation. This misalignment of incentives is diabolical. Imagine a world where a company had a clause where all revenue generated went to the founding team, and shareholders received nothing. It would be hard to imagine such a structure getting funding. But in crypto, historically, because investors have been able to sell tokens, they have chosen not to call out these structures.

This is where enshrining revenue and fees that flow back to token holders is crucial. Today, the most acceptable structure for this is token buybacks. This is what I propose for Infinex. One of the reasons used to justify teams diverting funds to themselves is funding operations. For Infinex, a solution for this is to do what Binance used to do: periodically buy back directly from the treasury to ensure sufficient operational funds. If these tokens are then locked, they are effectively taken out of the market.

So now we have key rules enshrined, preserving a non-custodial platform. To be explicit, this means that any user can exit the platform without requiring anything from any other party. The second is that all fees and revenues, after any incentives, rebates or other direct costs, must be used for token buybacks.

This is very simple, and I think it would be sufficient in a perfect world. It is almost sufficient for traditional startups, where both users and investors accept that the founder is in charge and has significant discretion to pursue a product strategy of their choosing. However, we have not yet contemplated when things go wrong and a change of management is necessary.

Let's now discuss changes in management. What happens if the founder no longer has the mandate of token holders? In traditional startups, there are two options: the first is that the founders control the company, in which case minority owners can do nothing and are along for the ride. The second is where there have been sufficient sales of equity that the founders no longer have control; in that case, if the owners lose faith in their management, they can replace them. This is rare in startups, but it definitely happens. WeWork is a great example, and so is Apple, but for the opposite reason. But in crypto, it is rare for a founder to control the majority of a project token.

But in crypto, for the last few years, it has been "best practice" for the founder to larp as having given up control of the project, although sometimes it is not a larp (see my abandonment and subsequent return to Synthetix). In these structures, the Founder takes control of a totally separate entity, a "Labs" entity. This is where the majority of meaningful decisions are made. Unfortunately, it is rare for most holders of the token to exercise any control over the Labs entity at all if they bought the token on secondary markets.

Let's assume, though, that there is a single entity. If the founder or manager of that entity has been given discretion over normal decision-making (in the case of Infinex, everything except custody and revenue) and they lose the mandate of the token holders, how do we resolve this? In DAO governance, historically, this process has been extremely convoluted or nonexistent. But I would argue we want to simplify this significantly, such that the bar for exercising this change of control is high. The simplest solution is to allow a group of token holders to propose a new candidate, and if ratified by a vote, that candidate takes over the project. Again, this is similar in process and outcome to how it is done in traditional startups. You don't dissolve the entire company and restart it. You just replace the leader. There are, of course, examples where the leader decides they don't want to be replaced, and they have sufficient social capital both internally and externally that it is impossible to replace them, see OpenAI. Again, feature, not bug.

So let’s practically review how this ought to be done in the case of Infinex. If a group of Patron holders (or the future infinex token) votes to replace me and they reach a pre-agreed threshold of token votes, then the vote passes and I am replaced. This would necessitate a token bond being offered of sufficient size that token holders were willing to vote in this new candidate. This is pretty simple. But what if I refuse to go? In traditional startups, we have the law to rely on, although it is still a somewhat murky question as to who ultimately controls a company, as Matt Levine loves to point out frequently. In Infinex, I propose a simpler solution: if I refuse to give up control, my tokens can be slashed. Now there is a risk here for me, which is that my tokens might be arbitrarily slashed, but tbh it is a risk I am willing to take. If token holders decide to slash my tokens, I will obviously and immediately stop working and start fudding, which would be bad for everyone. If this is no longer a credible threat, then I should definitely be replaced. So we have pretty good incentive alignment in both the failure case and the optimal case.

We have a Patron NFT, which is the governance token today, but it is a pretty retarded form of ownership for a bunch of reasons. Suffice it to say, I think we need to replace it. This change clearly falls outside the bounds of the rules defined earlier. But it is also a pretty critical process in the project evolution. So we need a mechanism to make this decision, and it seems fairly clear this should be a supermajority decision, imo, something like 50% of net voting power in the affirmative. I am proposing adaptive quorum voting, where if there are no votes against, a minimum of 50% is required, but if there are votes against, they are netted off the affirmative votes. This handles low governance participation, but if we have 100% participation, we can handle up to 25% against and still pass the proposal if the other 75% vote for it. My proposal is that a referendum-style vote must be held for any change outside the defined rules of day-to-day governance.

Operationally, these bylaws will be enshrined in the existing Foundation and form part of the constitutional documents that bind its independent director. The Foundation will delegate decision-making to an Association, and both of these entities will ensure oversight and adherence to the governance structure below on behalf of all token holders:

1. A 50% net majority requirement of Patrons voting to enact the new governance system

2. Delegate project decision-making to a project lead

3. All revenue after any direct costs, such as incentives or rebates, must be used for token buybacks

4. The platform must remain non-custodial

5. Modifying governance requires adaptive quorum voting

6. Replacing the project lead with a majority vote by Patron holders

7. Quarterly reports on token buybacks and sales as described below

Another consideration is how to manage the treasury. The simplest version is to operate under the assumption that with these rules in place, particularly the buyback structure, the treasury will be used effectively to drive revenue growth and not raided by the project lead or the rest of the team. I think it is fair to say that this would be a somewhat naive take, given what we have seen in the space historically. Unfortunately, I don't think there is a simple solution for this other than clear disclosures. My proposal is to add a requirement to release audited reports on buybacks and treasury sales on a quarterly basis. If this is not done in a timely fashion for any reason, I should be replaced. This gives token owners oversight of the buyback rate and use of funds without introducing complex mechanisms to manage the treasury.

The final consideration is the legal structure. While this proposal optimises for simplicity and clarity with respect to the onchain project governance, it does not account for legal liability for token holders or the people managing the protocol. While I am personally comfortable taking this risk, it will be almost impossible to find someone who is willing to take this on, therefore making it impossible to replace me should it be necessary. We therefore need a simple structure that enshrines all of the above in a legal wrapper, protecting token holders, managers and counterparties. I believe a Swiss association is the optimal solution for this, and all of the above processes can be incorporated into the association charter, providing meatspace clarity without introducing overhead.

Another consideration is that Infinex needs to be able to sign contracts in the real world. We now have two entities with a clear mandate to enforce governance and act on behalf of the project as a counterparty to agreements and other legal requirements. This ensures we have a mechanism for entering into contracts for services and infrastructure.

If we were launching Infinex today, it is possible I would opt for a traditional company structure. This would have been tempting because of the reduced complexity, and it would have enabled me to have maximal control of the project direction. I'm glad we don't have that option today because I still strongly believe tokenisation is the optimal path for cryptocurrency projects. We are all Patrons now.

Once this proposal has been reviewed by various stakeholders and any feedback incorporated, I will look to formally propose it through our current governance structure.