Hello Aragon Community,
At Delphi Digital, we’ve been researching the governance space for a while. As part of this, we’ve taken a deep dive into the Aragon Ecosystem with a particular focus on Aragon Court. We were very impressed with what we’ve found, and decided to take a position in ANJ. This week, we also published a research report on ANJ for our institutional subscribers.
We were impressed with the quality and traction of Aragon’s governance-as-a-service DAO tools and its potential to become a vertically integrated, cross-chain digitally sovereign jurisdiction. We see Agreements and the Court as picks and shovels DAO infrastructure,with agreements enabling subjective contracts and the court enforcing them. Combined, they can enable use cases such as secure off-chain “Optimistic Voting”, inter-DAO collaboration and more. In the near term, the recent proposals on the Yearn and Aave forums, and the support they’ve received, are evidence of early PMF and represent incredible opportunities for the Court to prove its worth.
While, as we wrote last week, we believe most of the concerns pointed out on the Yearn forum are non-issues, security is a legitimate issue that must be dealt with. With a PoS like consensus algorithm, the Court is vulnerable to 51% attacks, and security is directly correlated to the Court’s market cap. At $6M, the Court is simply not secure enough to handle much larger DAOs, as value at stake in disputes can quickly far exceed the Court’s market cap, opening it up to attack.
With this in mind, we’ve put together a proposal outlining three ideas we believe will help bootstrap the Court’s security and market cap. We included these in our report to subscribers, but thought we’d repost them to get the Aragon community’s feedback. I will now describe them in turn:
Improved Fee Market
Currently, fees are set at a fixed rate, meaning an organisation submitting a dispute with $100 at stake pays the same as an organisation submitting a claim with $100M at stake. This is problematic because Court users are not paying for the security they use. This means the market cap, and consequently the Court’s security, will not grow alongside the value of claims adjudicated, leaving the platform open to attack.
It also leads to flawed value capture, in that the protocol extracts too much from those with smaller disputes and far too little from those with larger disputes. With the current numbers of 10 DAI subscription fee and 30 DAI dispute fee, our suspicion is the Court is likely to be undercharging almost everyone.
In order to test out these assumptions, we forked Placeholder’s ANJ model, altering the assumptions and built two separate discounted cashflow models: one using the current fixed fees and another using variable fees that scale based on the amount of value at stake. We go over these models below and on the next page.
Current Fixed Fee Model
The original Placeholder model assumes a world where DAOs target the long tail of SMEs, with 1000 DAOs in Y1 and an aggressive growth rate leading to 22,000 DAOs by Y5. While we believe in this thesis in the long-term, in the short-term we think it’s more likely that we have a smaller number of more economically significant DAOs. The data seems to bear this out, as there are currently 1600 DAOs on Aragon, of which less than 100 are active each month. Assuming a generous 50% conversion rate for the Court, this is still only 50 DAOs subscribed in Y1. As such, we kept the growth rate the same but lowered the DAOs subscribing in Y1 to 50. We also assume the number of cases and costs per case go up as the DAOs become more economically meaningful and the Court proves its usefulness.
As we can see by the model, juror profits rely solely on the number of DAOs and average number of cases per DAO. This means there is no relation between the amount of value at stake in the court and juror profits. This results in a low NPV, meaning high value cases are likely to either pose a security risk to the Court or choose not to use it altogether. In addition, juror profits derive primarily from dispute revenue rather than subscription revenue. This is problematic since, as with insurance an effective court is one that’s used as little as possible, with the possibility of dispute acting as a deterrent but subscription fees driving the majority of the revenue.
Variable Fee Model
In the variable fee model, we use the same base assumptions as in the fixed fee model, but add in some assumptions around DAO size as we expect fewer, larger DAOs to adopt the Court. We assume an average DAO treasury size of $10M and that this grows quickly as the ecosystem develops. We assume the Court will begin by securing 20% of these treasuries, growing 10% per year as trust in the court increases. Average dispute size is $52k or 1% of the $5.2M average dispute size in arbitration courts in the US, growing 10% per year. Crucially, we assume cases per DAO actually decline 10% per year, reflecting the intuition that as the Court proves itself, it is used less and less.
Rather than the current fixed pricing, we assume variable security based pricing for both subscription and dispute fees. In the case of subscription fees, we assume these are based either on the percentage of the DAO’s treasury that is secured by the court or on the total amount of collateral at stake on a particular agreement (i.e. for each type of action an agreement allows, calculate number of potential actors * collateral at stake per action). We set this fee at 20 bps/year. In future, part of this fee could also be refunded to DAOs that don’t submit any disputes to encourage efficient use of the Court.
In the case of dispute fees, we assume these are priced based on the amount of value at stake in that dispute. In the case of disputes that don’t have monetary value (e.g. the wrongful update of a Github repo), these could be priced in terms of the collateral at stake by the actor. We set this fee at 10bps per value at stake in a dispute.
As we can see, variable fee based pricing vastly increases profits to jurors and the NPV of ANJ. This makes the Court significantly more secure, allowing it to safely service higher value claims and develop its network effect. In addition, unlike with the fixed fee model, variable fees make up the vast majority of juror revenue which is the only sustainable long-term equilibrium.
These are just a few ideas on how security based pricing could be implemented, and we can imagine different models emerging with time such as insurance based models based on a DAO’s history, expected frequency of cases and value at stake. The key takeaway is that the level of fees must be related to the amount of value at stake in order for the Court to be long-term sustainable.
Removing the Bonding Curve
What is clear from both the above models is that the fee market in the first few years will be too low to sustain jurors and bootstrap the Court’s security. Given the incredible opportunity of being used as part of Yearn’s governance, we believe it’s extremely important for the Court to implement well-designed supply-side subsidies in order to incentivise an active juror base and ensure security is sufficient to take advantage of opportunities. Before we go into this, it’s important to understand how ANJ is currently issued and why this poses a problem.
We are generally fans of bonding curve implementations as per our recent Nexus Mutual report. However, in this case we feel that the bonding curve doesn’t serve the Court’s goals and may in fact be hindering its progress. While in Nexus’s case the bonding curve is used to build up a capital pool which directly relates to Nexus’s ability to issue claims, in the Court’s case the capital serves no purpose. In addition, denominating the bonding curve in ANT will continue to slow supply-side adoption as jurors must not only take price risk in ANJ but also passthrough exposure to ANT.
While we understand the need to allow ANT holders to capture value from the Court, we believe this can be better accomplished via a fee share later on. In the short to medium term, our position is that it makes more sense to optimise for adoption and growth versus ANT value capture as the network effects behind the Court mean it’s a winner-take-most game. As such, It’s important that Aragon Court take the lead here, especially given higher market cap (and thus more secure) existing competitors such as Kleros. After all, there will nothing for ANT holders to capture value from if the Court doesn’t scale.
As such, our suggestion, laid out in full in the next page, is to suspend the bonding curve altogether, using the locked capital to create supply-side incentives to reward active jurors and bootstrap the Court’s security. An alternative suggestion could be merging ANJ into ANT, but we focus mainly on the former as feel the operational complexity of an ANJ buy-out may make it unfeasible, as well as preventing the development of the Court’s community.
What is clear from both models is that, as with almost all early stage crypto projects, fees will not be sufficient to bootstrap supply-side adoption and must be supplemented with issuance. This is especially critical given the opportunities that have emerged for the Court to be used within large DAOs such as Yearn and Aave. These are massive opportunities with arguably the two most significant DAOs in the crypto space and it is crucial that the Court is secure if it is to be able to take advantage of them.
In order to bootstrap adoption, we propose the implementation of supply-side incentives. We understand there are currently $22M worth of ANT in treasury that can potentially be used and have structured our suggestion accordingly. We suggest that a majority of this ANT is converted into ANJ on the open market and paid out to ANJ stakers (jurors) who are active in the court over the next 2.5 years. A smaller portion is used to allow DAO’s to use the court system for free by covering their fees to jurors, and the residual is set aside to incentivize a long term liquidity pool since there is no longer a bonding curve to purchase from.
Since the bulk of these expenses are used to compensate Jurors, that’s where we’ll focus our attention. The two primary goals of this distribution are illustrated in the charts above. One objective is to incentivize more value to be locked up by jurors by rewarding them for staying active through an enhanced yield. In order to avoid the pitfalls of other liquidity mining schemes, we suggest that the yield Jurors receive comes with long-term incentives which we describe on the next page. Like most supply side incentives, they’ll start off highest in the beginning when sustainable fee revenue is at its lowest. That’s where the second objective comes into play, to provide enough compensation through the initial 2.5 years so that the court can function at scale without actually having the necessary amount of DAO’s as clients to sustain it. These types of systems require trust, security and a successful track record before they can be more widely adopted, but this lengthy validation phase also helps create a long-term moat as other projects must invest heavily to bootstrap security
In order for a long term incentive scheme of this nature to be successful, there need to be parameters in place that ensure it rewards the appropriate actions and long-term stakeholders while minimising the amount of tokens distributed to short-term or yield-focused stakeholders. As such, we suggest deploying the majority of incentives to jurors to begin with until the Court reaches a critical mass of market cap and security that will allow it to serve larger communities like Yearn and Aave.
Once the Court reaches a sufficient market cap and security has been achieved, we suggest pivoting the incentives to the demand-side, incentivising DAOs to subscribe to the Court and submit cases. To prevent the gaming of incentives that happened with other yield farming initiatives, we suggest the demand-side incentives are paid out through jurors by offering partial discounts to DAOs and subsidising jurors for the difference. This helps encourage demand-side adoption by DAOs while maintaining a cost to using the Court in order to prevent sybil attacks in the form of spurious cases.
To further enhance these goals, we suggest that rather than a 1 year token lock-up on rewards a la Synthetix, rewards are locked and a withdrawal fee is implemented. This fee starts off extremely high (<90%) and gradually reduces to 0 over 12 months, continuously punishing short-termism. Withdrawal fees collected go back into the reserve, allowing them to be redeployed as rewards. This means tokens are continuously redistributed from short-term oriented stakeholders to longer-term ones, ensuring the supply-side incentives are not wasted but instead deployed to creating the stickiest, strongest and most long-term oriented community of jurors possible.
These types of mechanics help significantly reduce the execution risk of this project. Our initial discounted cash flow model presented the financials without the incentive infusion (as seen below in the white). Although the NPV isn’t materially different with the capital distributions incorporated, the front-loaded rewards help subsidise the low initial fees, reducing the barrier to entry for jurors and overall execution risk for the project.
We welcome feedback from the Aragon community regarding these proposals and look forward to working together on refining them. We like to take an active role in projects we’re invested in, and are open to working closely with community members or the Aragon team to implement any of these changes.
Edit: For those interested, we’ve also recorded a video walk-through of our proposal which you can see below: