DAO on Trial: Code, Humans, and the Limits of On-Chain Governance

DAOs have reached a point where theory is no longer enough.

The podcast DAO on Trial: Code vs Humans brings together three people who have been directly involved in building, governing, and legally interpreting some of the largest DAOs in DeFi: Michael Egorov of Curve and YieldBasis DAO, Marc Zeller of Aave DAO, and Gabriel Shapiro of Metalex. Moderated by Gerrit, the discussion moves past the question of whether DAOs “work” and instead examines where decentralized governance collides with real-world constraints. This article serves as a summary of this important conversation.

The central tension is simple to state but difficult to resolve: smart contracts can enforce rules on-chain, but DAOs still operate in a world of law, politics, and human incentives.

What people mean when they say “DAO”

One of the first issues raised on the podcast is that the term “DAO” is used to describe very different organizational structures.

Some projects label themselves as DAOs while relying heavily on off-chain coordination, small multisigs, or non-binding votes. Others are closer to what Egorov describes as a “classic DAO,” where governance decisions directly control on-chain logic, budgets, and protocol parameters, such as Curve.

Shapiro stresses that this distinction is not academic. Legal systems respond to how organizations function in practice, not how they describe themselves. When loosely organized projects adopt the DAO label, they risk creating legal precedents that affect more rigorous systems as well.

A useful working definition emerges from the discussion: a DAO is a system whose core logic is enforced on-chain, such that its essential arrangements do not depend on external legal enforcement. That does not mean DAOs never interact with the real world, but it does mean their fundamental commitments are enforced by code.

DAOs as sovereign systems, not companies

As the conversation unfolds, all three speakers converge on an important framing: DAOs do not behave like traditional companies.

Egorov captures this succinctly when he suggests that a DAO is closer to a country than a firm. Zeller builds on this idea by describing Aave governance as a political process in which different actors represent different visions for the ecosystem. In his words, delegates function more like political parties than corporate managers.

This analogy explains several features of mature DAOs that often surprise observers. Disagreement is persistent. Governance is slow at times. Outcomes are not always efficient in the short term. Yet legitimacy emerges from process rather than alignment.

When decisions are debated publicly, voted on transparently, and executed on-chain, participants accept outcomes even when they disagree. That acceptance is what allows DAOs to persist through conflict.

Governance and the importance of real power

A recurring theme in the podcast is that governance only works when it carries real consequences.

Zeller emphasizes that token holders in Aave are sovereign in practice, not just in theory. They decide protocol upgrades, risk parameters, treasury allocations, and even changes to the token itself. The structure distinguishes between routine actions and critical ones, requiring higher levels of consensus for decisions that affect the core of the system.

Shapiro connects this to a broader principle: governance requires both power and skin in the game. If participants can complain but cannot meaningfully affect outcomes, engagement erodes. If voting has no economic consequence, governance becomes symbolic.

This helps explain why some DAOs attract sustained participation while others do not. Curve and Aave differ in design, but both give governance real teeth. Votes matter because they directly change how the system behaves.

Curve and Aave as contrasting models

The podcast offers a useful comparison between two of DeFi’s most influential DAOs.

Curve’s governance is built around incentive design and constraint. Egorov explains that the system controls emissions and parameters, but not everything is mutable. These limits are intentional. Curve’s vote-escrow model (veCRV) requires long-term commitment, making it costly to pursue short-term extraction at the expense of the system. Read more about Curve governance.

Aave takes a different approach. Zeller describes a layered governance system in which the DAO retains ultimate authority, while execution is handled by small, specialized service providers. This allows the protocol to remain decentralized while still moving quickly. Read more about Aave governance.

Zeller pushes back against the idea that DAOs must be slow. In his view, a modular service-provider model can be more efficient than a large, vertically integrated organization that becomes bogged down in internal politics.

Where on-chain governance stops

Despite their differences, both Curve and Aave face the same structural limitation: some assets cannot be governed directly on-chain.

Brands, trademarks, intellectual property, and legal agreements exist outside smart contracts. When these assets are held by off-chain entities, they create points of leverage that governance cannot fully control.

The podcast touches on recent disputes that highlight this problem. Even when a DAO controls the protocol and treasury, unresolved questions about off-chain ownership can undermine sovereignty.

Shapiro argues that this does not invalidate the DAO model, but it does require deliberate design. Ambiguity around who controls off-chain assets becomes a governance risk once a system reaches scale.

Legal entities at the edges

The panel does not advocate a single legal structure for DAOs. Instead, it outlines a spectrum of approaches.

Wrapping a DAO itself in a legal entity can provide clarity and protection, but it may also introduce obligations that conflict with permissionless participation. Egorov raises a pointed concern about what happens when authorities demand identity information about anonymous governance participants with significant voting power.

An alternative approach places legal entities at the edges. Foundations or trusts can manage specific functions such as intellectual property or security operations, with rules that require consultation with the DAO. In this model, the core governance system remains self-enforcing, while legal structures handle unavoidable real-world interactions.

The shared view is cautious. The more a DAO is fully wrapped in traditional legal frameworks, the more it inherits assumptions that may be incompatible with decentralized governance.

Immutability, upgradability, and governance choice

One of the clearest disagreements in the discussion concerns whether core protocol rules should be immutable.

Egorov expresses a preference for immutability, particularly in token economics. If supply cannot be changed, governance cannot inflate away problems or rationalize extraction. Expectations remain clear.

Zeller argues for upgradability. He notes that no team can foresee all future conditions, and rigid systems can fail when mistakes or external pressures arise. He points to Aave’s token migration as an example where modifying supply through transparent governance enabled long-term sustainability.

Both agree that governance actions are never free of consequences. Markets respond, reputations adjust, and abuse carries costs. The difference is not about right or wrong, but about constitutional design.

Applying lessons to new systems

Egorov briefly references how these lessons informed newer projects such as YieldBasis.

Two adjustments stand out. Locked governance positions should allow limited flexibility for operational safety, such as in cases of wallet compromise. Governance mechanisms should also avoid encouraging excessive concentration of voting power in intermediaries whose incentives may diverge from those of individual participants.

These are refinements rather than rejections of earlier designs. They reflect how second-order effects become clearer as systems mature.

The unresolved future of DAOs

Toward the end of the podcast, the conversation turns to the long-term outlook.

Shapiro is clear that DAOs are not fully safe. Their structure challenges traditional enforcement and governance models, which creates ongoing tension with states and regulators. While much attention has focused on securities law, taxation may become a more significant pressure point as DAOs generate meaningful revenue.

Despite this, all three speakers remain cautiously optimistic. DAOs offer a distinct institutional model based on transparency, rule-bound coordination, and credible neutrality. In domains where trust minimization matters, this model has already proven competitive.

Conclusion

DAO on Trial: Code vs Humans does not deliver a verdict in the conventional sense. Instead, it shows that DAOs have entered an institutional phase where design choices matter more than slogans.

The strongest DAOs are not those that deny the existence of law or human incentives. They are those that clearly define what is governed by code, how off-chain realities are handled, and where authority ultimately resides.

The trial is ongoing, and its outcome will be shaped not by theory, but by the governance decisions these systems make in practice.