Berlin Blockchain Week 2025 - Governance Takeaways

During Berlin Blockchain Week this June, the ZKsync Governance Team was able to meet with governance facilitators, governance researchers and Delegates from across the Ethereum ecosystem. @rafa and I attended various events and were inspired by some of the governance-focused talks. Below are a few notable takeaways from these discussions and talks.

Overview

Decentralized governance is at a major crossroads - do we keep optimizing for a broken system or is it time for something radically new? If you’ve listened to Vitalik speak recently, you may have heard his call to 'burn DAOs/governance to the ground’ - a push to rethink and redesign governance models in response to the clear shortcomings and pain points of current systems. While a lot of suggestions for change are often just band-aid solutions for current models, there is growing discourse around more compelling and transformative ideas that aim to replace existing structures entirely.

While new governance design experiments are beginning to emerge (e.g. EigenLayer council model, ZKsync token mechanics), there is a lot of room for innovation. As we enter the next era of decentralized governance, a clearer consensus is emerging around a few specific areas outlined below.

Rethinking Purpose of Governance

A recurring theme in the discussions the Gov Team participated in was the comparison between the original intent behind governance and its design decisions, and whether those purposes still hold true today. A few key takeaways:

  • The initial rationale for adopting DAO-style decentralized governance - dispersing power to prevent capture, lowering platform risk, bolstering technical security, and enhancing transparency - should be re-evaluated in the reality of participation challenges and today’s shifting regulatory landscape.
  • Rethinking entity structures: Having a Labs / Foundation model became the standard for most ecosystems. There are a range of tensions tied to different aspects of the current entity structures (e.g. communication, power/responsibility separation). Given the changing regulatory environment some are considering merging these entities back together. (See a16z article “The end of the foundation era in crypto”)

Desired System Changes

We also noted various desired system changes during discussed with different parties. At a very high level, these include:

  • Token-based governance models: Most projects chose the token governance / delegation route when designing their governance systems. Now, nearly a decade after the original DAO hack, these models are being questioned and many are starting to advocate for alternatives (e.g. move away from a Delegate system).
  • Governance surface minimization: What is the right governance surface area? What facets of a project/organization really need to be decentralized (technical vs monetary vs strategy vs social/community). Within the verticals of ares of decision making, how far down the stack does decentralized governance need to reach (e.g. deciding the logo color vs removing entire council)? There seems to be trend towards governance surface area minimization (e.g. see Optimism Season 8 Governance Update).
  • Participation Challenges: Participation remains a persistent challenge in current governance mechanisms. Rather than incentivizing Delegate engagement through compensation, it may be more effective to explore entirely new governance models that sidestep these issues altogether. (e.g. EigenLayer council model / optimistic governance + veto systems).
  • Re-aligning multi-stakeholder participation: In many governance systems, the stakeholders best positioned to contribute - such as builders with real skin in the game - are often the least engaged. Meanwhile, more active participants can be those with minimal exposure or even misaligned incentives. Governance design must evolve to better attract and retain meaningful participation from key contributors while minimizing the influence of extractive actors.
  • Automation and AI in Governance: There is a long-term orientation towards automation and the use of AI in governance.
    • AI: Going beyond just using AI to synthesize forum discussion/community sentiment, could we create AI agents to replace Delegates or councils, executing decisions based on a specific set of values/conditions - or could governance become one agent that is managed by Delegates/token holders?
    • Automation: There is a strong desire to be able programmatically enforce processes rather than have to manually manage them. Any council or steerco’s mandate should be to automate their work to make their required participation be obsolete or extremely minimal. This highly aligns with ZKsync’s token mechanics model - which aims to reduce multisig responsibility & coordination for token distribution.
7 Likes

Great Stuff Shelby! Thanks for writing this up.

I have some thoughts on this here: https://x.com/raphbaph/status/1896538021081583882

And recently published, see Jeff’s amazing paper here: [2505.04136] Delegation and Participation in Decentralized Governance: An Epistemic View

TL;DR: More participation can be bad if participation isn’t qualified. DAOs should develop a toolbelt of decision making modalities and match them to
a) the kind of decision taking place (e.g. financial vs social)
b) the decision-makers involved (low-context shrimp vs high-context contributors)

Some argue this introduces a meta-problem of how to decide on how to decide. While this is one, quite cynical way, of seeing the space, I would argue it opens up a design space and can actually circumvent Arrow’s theorem by offering a multi-dimensional space for preference elicitation.

2 Likes

Disclaimer: I am completely in favor of changing / improving stuff.

Question: Who could make such a decision? Can “zksync” introduce changes unilaterally? Is the TA the own “entity” that would have to vote on this?

1 Like

Thank you for sharing this @rspa_StableLab ! Looking forward to diving into that paper more.

I agree it is worth getting into this nuance when it comes to governance design! As you already hint at, this is heavily tied to governance surface area. First question is what are the necessary verticals of governance surface area that should be governed by Delegates, next is how far “down the stack” within each vertical should that surface reach for decentralized governance (vs what can/should be made by internal teams), and who should be involved within those verticals/at those different levels within each vertical. It may be a headache, but this is where I see the governance design space moving towards.

It also begs the question of how certain decisions are made (e.g. approval vs optimistic + veto). Technical decisions are the easiest example. Most Delegates voting on protocol upgrades (across all DAOs, not just ZKsync) don’t have the technical expertise to be able to sufficiently review / understand technical changes to the protocol included in an upgrade. This challenge also prompts the question if Delegate approval is actually the most effective way to approve protocol upgrades generally. e.g. what if protocol upgrades were optimistic and TA / SC / ZK Chains could veto the upgrade?

Bottom line is, the gov systems/mechanisms have been using the past ~5 years are not working the way we expected / need them to. Really exciting to hear more and more discussions that try to dive deeper into that nuance.

2 Likes

Any changes to the ZKsync governance system can be proposed by any party, but any changes to governance parameters on any of the 3 Governors will have to be approved by the Token Assembly.

1 Like