A Plea for Based ZKsync Governance

Hi guys,

I hate to be ‘that guy’ but I see a lot of problems already emerging in ZKsync governance. I’m not a ‘governance community insider’ like a lot of you–I don’t go to the crypto governance conferences and do the mutual back-scratching thing. ZKsync is also the first DAO I’ve decided to be a delegate in. I also may be more of a realist/pragmatist than many others in crypto. I come from the corporate governance world–my background consists of 7 years in BigLaw working for big public companies with very closely scrutinized and carefully structured and regulated governance, and then 7 years as outside counsel or GC to various crypto projects, especially ones backed by more mainstream VCs and therefore with more regimented corporate governance. This gives me a different perspective than many of you which may come across as a bit jarring but I hope the diversity of opinion can be helpful.

No offense intended but I think if you take a deep breath you can see the way DAOs/“governance” have evolved has gotten pretty insidery, dysfunctional and grifty and I really hope ZKsync–with its highly cypherpunk ethos–can avoid falling into the same traps as nearly every other DAO. I’m still optimistic but I’m seeing early warning signs flashing hard and feel like I might as well express my thoughts. I realize it’s very possible they are highly minority thoughts because unlike most delegates/DAO participants I don’t buy into many of the currently popular ‘metas’ of the Ethereum professional governors community–such as ‘optimistic’ governance.

To that end, I want to throw out some concepts and ideas that I think people should be paying much more attention to:

  1. How Much Do we Really Need Intermediaries???

Crypto is meant to disintermediate and to prevent the need for intermediaries. Every new ‘TPP’ that is not a direct grant to an app or other interesting/useful project, but rather is an allocation of money to money-allocators who are supposed to go find/incentivize/pay the real builders, is creating a new intermediary. While obviously there is some need for more nimble grants councils/incubators/etc. for every ecosystem, there should in general be significant caution against and reluctance to empower new intermediaries with large budgets and semi-opaque/trusted internal governance, and there should be significant interest in exploring ways the DAO can invest in / sponsor actual builders much more directly without middlemen.

Keep in mind, we already face one layer of intermediation within ZKsync because Tally/delegated governance was the chosen system and thus most of the voters have power far above their skin-in-the-game and have limited accountability. Intermediaries appointing other intermediaries poses exponential issues.

I’ve spoken to quality teams building elastic chains. In general, they plan to approach the Token Assembly directly to ask for grants/support. I think this can generally be expected, and is a better system (though need not be the exclusive system) for allocating capital.

  1. When we Empower Intermediaries, How Should we Keep them Transparent and Accountable?

DAO-adjacent councils, multisigs, foundations, etc. tend to suffer from a worst-of-both-worlds effect–all of the trust requirements of centralized offchain TradFi entities, with none of the legal checks/balances found with offchain TradFi entities. These issues have been written about extensively both theoretically and anecdotally–for instance, by tokenbrice explaining his experience with a DAO-adjacent council and why he resigned, here.

  1. Monitoring Costs are a Well-Known Principal/Agent Problem that Must Always be Considered in Appointing and Managing Intermediaries. Any Remedies/Protections Against Agent Misconduct are Only as Good as the Willingness, Knowledge and Ability to Use Them on a Timely Basis

Monitoring costs, and other agent costs, arise whenever you appoint intermediaries. One of the best uses of blockchains and smart contracts is to reduce these costs, but they are not a panacea–it all depends on how things are structured. For example, an agent being required to hold money in a multisig is great because the DAO can watch all the transactions, but if the multisig is not rate-limited and there is not a legal rule that actually requires the agent to keep the money in the multisig, this monitoring-cost-reducing technique can be easily bypassed.

A principal may have various remedies against agent misconduct or even simple agent underperformance. In the real world, these often take the form of legal agreements and legal remedies where a principal can sue an agent for breach of fiduciary duty or breach of contract. In crypto, these remedies often take the form of onchain revocation of funds, various versions of ‘hardforking’, slashing, ‘ragequitting’, ‘guildkicking’ and similar onchain remedies.

In both cases, the fact that a potential remedy exists does not mean it will be effective. To effectively exercise a remedy, you must know something wrong has occurred, and this means you both have all relevant information and be actively processing it. This means monitoring costs must be reasonably low (otherwise problem will not be detected because detection methods are too expensive) and remedies must be good. My humble suggestion is that DAOs are very very bad both at making sure they have good potential remedies against misconduct–often just yeeting funds into a multisig staffed by supposedly ‘high-reputation’ people and then ‘optimistically’ hoping they perform well–and at exercising the remedies they do in fact have. This is why they often lose money, endure ‘hostile takeovers’, get rugged by grantees, etc.

In my opinion, ZKsync should do everything possible to reduce its agent costs / monitoring costs, (1) by avoiding intermediaries as much as possible; and (2) when it has intermediaries, by having an agreed upon standard way of trust-mitigating them, involving a mix of onchain (e.g. rate limitations on SAFEs, DAO ability to remove/appoint SAFE members, etc.) and offchain controls (e.g., legal agreements establishing what intermediaries can and cannot do) with real teeth. This means not being totally ‘optimistic’.

In discussions I’ve had with a few of you, it has been pointed out to me that the problems under categories #1 and #2 are not as bad as I claim and not worth dealing with in the granular ways I have pushed for, because the DAO can always kill a “capped minter”. However, some of the same people in other conversations I’ve had with them have also said the DAO should/will almost never do this, as it is a nuclear option, would be a distraction, would be negative for the community, would be drastic, etc. Moreover, it is an imperfect remedy because:

(1) It does nothing to claw back ZK that has already been minted and was potentially misused. Thus, that misused ZK basically becomes ‘free money’ to the rogue agent.

(2) It is an all-or-nothing ‘nuclear option’ meaning nothing of value from the project can be easily preserved (e.g. by installing new council members while otherwise maintaining continuity) and thus there is heavy pressure not to use it due to the sunk costs fallacy.

(3) Most importantly, because of high monitoring costs, a problem may not be detected long enough before ZK is minted in order to kill the minter in time to mitigate most damages. Most agent-principal scandals–be they from crypto (rugs, FTX-like frauds, etc.) or in tradfi (Enron, Theranos, etc.) are only detected after they have become incurably disastrous.

ZKsync’s current practice of having TPP recipients propose their own handpicked supervisory council–as was done with both Ignite and Catalyst–is not a good solution to monitoring cost issues. There is a huge conflict of interest in this regard, as the TPP recipient is incentivized to propose ‘friendlies’. And this council itself is then just another intermediary posing all the same principal/agent problems as the proposer itself does, and there need to be ways to handle the agent costs from that. Just in general, it generally does not make sense to let people pick their own supervisors nor does it make sense to have a new panel of supervisors for every TPP.

Needless to say, all these issues with monitoring costs compound the more intermediaries you appoint. A DAO may be reasonably able to monitor the performance and remedy the underperformance of a single DAO-adjacent grants council, but its resources may be overwhelmed if there are 2, 3, 4, 5, 6 different grants councils or “programs” all functioning like high-trust intermediaries.

  1. Bundling is a Recognized Adverse Governance Practice, and Un-Bundling is a Recognized Good Governance Practice

In corporate governance, ‘un-bundling’ refers to making sure that initiatives that are unrelated or only loosely related, are separately proposed to and voted upon by governance, in separate votes, so that each can be considered on its separate merits. Unbundling is well recognized as a good governance practice, and in the U.S. is legally required for public companies, enforced by the SEC.

By contrast, in the U.S. Congress bundling is prevalent, and it is widely considered a dysfunctional, bad governance practice resulting in grift and ‘pork barrel spending’, deservedly despised by the American public.

Bundling appointment of financial intermediaries through TPPs with requests to fund governance tech development skunkworks is something that has been pushed for from the top-down by ZKsync Association coordinators in this DAO. I think it is a huge mistake for the same reason that bundling in general is a mistake. While in theory the proposals could be related as the idea would be the for the TPP intermediary to be constrained by the tech, when the tech does not yet exist, it means the program will be running for quite some time without the tech anyway, and therefore despite any apparent relationship, they are actually only loosely related at best. Moreover, even if the tech is developed and in a timely manner, there is no way for the DAO to be sure that the financial intermediary will ultimately fully retrofit their operations to be constrained by the tech once it is in place.

Instead of this, the DAO should consider governance research tech R&D proposals separately from financial intermediary proposals. Financial intermediary proposals should come with proposals to use existing trust-reduction tech (yes, like my company MetaLeX’s, or like Hats, or many others that are already out there and functional) from day 1 , priced like hiring a standard contractor rather than like funding an entire new governance startup. For the more ambitious long-term governance tech R&D (which I agree we should do), the DAO should receive and consider separate proposals on that topic. This is good governance practice 101–separation of concerns, separation of considerations, and not letting intermediaries operate on a completely trust-maximized basis while we spend potentially years figuring out a 100% automated solution that may turn out to be impossible.

  1. Pluralism Can be Good, Redundancy is Not

If we are going to fund multiple governance R&D skunkworks, they should all be meaningfully different and should offer real diversity of options to the DAO, rather than all trying to fit the governance philosophy of the current ZKsync Association. I don’t understand the purpose of getting 5-6 or however many governance startups to participate in building DAO solutions, when you insist that they basically all build the same ‘optimistic automated governance’ vision that the ZKsync Association is pushing. Having multiple initiatives only makes sense if they are meaningfully different in philosophy, mechanisms, potentially pricing and other commercial terms, etc.

  1. Skepticism, Pessimism and Realism Can be Okay

A lot of really bad things have happened in crypto, including in DAOs. It’s okay to be skeptical, it’s okay to ask tough questions, it’s okay to want more granular, continuous feedback/accountability loops. A lot of people talk about Optimism’s retroactive public goods fundings. While I have my qualms about it (I don’t like getting people to work for free when they don’t know what they will be paid), one good aspect of it is that, ironically, it is fundamentally pessimistic, as you don’t trust anyone with any money until they deliver a result. I think we too should be pessimistic, but that there may be better mechanisms for doing so than RPGF, and we can hack on those and figure them out.

  1. ZKsync is For Builders and Users, Not Governors

We should view the delegate system as a brief necessary evil before ZKsync’s DAO politics become much more driven by builders, apps and users. We should not view ZKsync as a way to enrich/sustain a professional class of governOOOrs. As delegates who are able to do that work because we are governance nerds and chronically online and interested in this stuff, we should view our main job as to make ourselves irrelevant as quickly as possible because ZKsync is taken over by builders and users and they govern as directly as possible. We should try to arrange that the DAO aka ‘Assembly’ has as direct as possible a relationship with ZKsync builders, rather than needing numerous governance and finance intermediaries.

  1. Reputation Staking/Slashing Is Not Enough

When appointing intermediaries, most DAOs implicitly rely on a system of reputation staking and reputation slashing. The idea is that if you just appoint high-reputation people, you don’t need other protections against their misconduct, as they will not burn their reputation. This is a seriously naive view, as we have seen over and over again in crypto as previously high-reputation people like SBF, Su Zhu, and many others (whose reputations are now so bad it is almost comical to say they were once high-reputation, but they were in fact) were willing to burn their reputations for the right price. The issue of course is that offchain ‘faults’ are often hard to detect, hard to attribute, and morally ‘muddy’, thus the risk calculus is not the same as staking ETH and getting it slashed for a double-sign. There are also always those–as we’ve seen from many influencers like ‘ben.eth’–that are willing to fully and deliberately burn their reputation for the right price in a last blaze of glory. Thus, other kinds of check’/balances on intermediaries, beyond just implicit reputation staking/slashing, are necessary.

That is all for now, I know my views may rattle and piss off some of you, but if they resonate with even one of you, it was worth it. Within the next few days, I will be making a separate proposal that aligns with the above thinking and outlines how I think all intermediaries of the DAO should be trust-minimized, starting immediately and not 12, 18, 24 months from now, in order to qualify for large ‘capped minters’ from the Token Assembly.

7 Likes

Hey @lex_node, I generally agree with the idea of minimizing intermediaries - or even eliminating them entirely.

However, what about situations like KYC? For example, if a project applies to the DAO for funding (via TPP), how would KYC/KYB be done to prevent issuing funds to e.g. a North Korean actor?

Or would KYC not be required because it’s a decentralized/DAO decision? What do you think, given your legal expertise?

1 Like

Agree. Didn’t read all of it tho.

We should maybe take a moment to entertain the idea that “the assembly”, the skeuomorphic model (proposal/law, delegates/representative, progressive decentralization [a16z]/decentralization) might be what is kepping the governance space hostage in a local minimum instead the “1 token 1 vote model”. (tokenbrice conclusion)

The “professional goverooor class” issue you raised might just be a marrige of convenience. Looking forward to not reading most of what you will propose.

1 Like

I see that ZKGPS (ZKGPS: Advancing Accountability in Proposal Execution) solves the challenge with who does KYC/KYB :+1:

3 Likes