This is a great post @drnick !
The overall approach of TPPs is attractive to me, and the key concern for me is how we can create effective mechanisms as a socio-technical system (multiple stakeholders + mechanisms) that designs and is impacted TPP mechanisms and other parameters of the system (governance design, etc).
With this framing, allow me to push back on a few things and emphasise others, so we continue to refine our understanding of the possibilities and best approach to take here:
-
Being “data-driven” sounds good but in practice often fails to acknowledge the complexity of reality, becoming counterproductive. Innovation leaps happen through contrarian bets on outcomes that can’t be demonstrated in the short term. Leadership being the act of inspiring others to march into the unknown (and a fundamental component of organisational systems we shouldn’t dismiss). Data driven also has a data availability problem, where what’s easy to measure is often preferred to what’s more representative but harder to measure (a rather problematic bias). We have both worked on inter-subjective mechanisms for a while and I think it’s critical to bring said thinking and research around sense-making and decision making mechanisms beyond majority voting and objective oracles here. Complexity theory provides us ample grounding for this, by having proven time and again that if we try to compress too much complexity into a single metric, said complexity shows up somewhere else in the form of unintended side effects and (negative) externalities).
Let’s avoid creating perverse incentives. Data-informed good, data-driven (most frequently) bad. -
The incentives of the actors involved need to be navigated carefully. DAOs are multi-stakeholder constructs by nature and zkSync is no different. We have:
- users of the infra (protocols, organisations, and individual users),
- those building the infra and ecosystem
- token holders (with overlap on the other groups + airdrop hunters and the like),
- delegates
from this perspective, the question of who designs mechanisms and for whom’s benefit is critical. Other DAOs have quickly realised that delegate systems lack sufficient incentives for effective participation, and have started to devise a series of complimentary incentive programs to increase governance participation. The risk is one of very low throughput and hence becoming irrelevant and other ecosystems outpace this one. I have often suggested Citizens Assemblies (or the web3 version Multi-Stakeholder Assemblies) as a better form of governance that avoids principal-agent problems and includes focused deliberation and incentives for participants. Irrespective of the system, we currently have a gap in incentives and a higher barrier to the creation of programs that can allocate capital (need to design, audit, get approval, deploy, etc). That makes zkSync more reliant on the foundation or charitable contributions by delegates (related to next point).
-
Designing mechanisms is very hard. Few are skilled at thinking through how socio-technical systems can change with a given mechanism. And DAOs have largely failed to date to design highly effective mechanisms (the exploration and refinement continues). Traditional allocations of capital (pots, instead of pipelines) are easier to manage and operate off-chain, allowing for rapid prototyping and managing ambiguity and uncertainty thanks to human actors that can very rapidly adapt. Here we’re largely devoid of said ability for rapid prototyping. Rafa pointed out how AI tools enable virtually anyone to code solidity, so the barrier is not so much technical, as it’s about the necessary level of clarity that engineering an automatic mechanism requires. Then all those engineering decisions need to be audited and validated through governance and that can give space to even more debate and even longer and more painful governance approval cycles (i.e. low throughout).
Low throughput is already a major issue in other ecosystems with lower mechanism design constraints, and compounded with the limitations around incentives and the requirements for usage of onchain data or otherwise bias for data-driven as opposed to data-informed… and we have quite a challenge here.
Now, I’m not saying this to be all doom and gloom (otherwise I would just be allocating attention somewhere else instead of posting here). I believe there’s an interesting path forward IF we can proactively address the limitations mentioned above.
In that, what I’m thinking moving forward (and would love to know everyone’s thoughts!):
- Double down short term on the idea of capped minters + multisigs with accountability frameworks.
- Use the above to
- enable R&D in this space that’s not reliant on charitable contributions (also so we can move to more advanced TPPs as soon as possible)
- and resolve incentive alignment (more on that below)
- Align with the foundation on strategy to know how much the DAO can focus on evolving TPP mechanisms without short term ecosystem development needs, or otherwise whether the DAO should also support ecosystem development short term. Depending on the conclusion, also create some short term capped minters + msings with accountability frameworks focused on ecosystem development strategy and execution.
NOTE ON INCENTIVE ALIGNMENT