Blockchain governance is a fight over amendment power. The useful question is not whether a protocol sounds decentralized. The useful question is who can change consensus rules, economic parameters, treasury flows, and upgrade timing, and who can block those changes. On-chain and off-chain governance answer that question in very different ways.
On-chain governance turns amendment power into an explicit protocol feature. Off-chain governance leaves amendment power in social coordination, client implementation, and voluntary software adoption. That sounds like a clean binary. In practice, it is not. Even the most formal on-chain systems still rely on off-chain drafting, testing, and coalition building. Even the most informal off-chain systems still have structured processes, editors, maintainers, and activation mechanics.
The real trade-off is operational flexibility versus governance centralization. On-chain systems can ship predictable upgrades, but they usually concentrate formal authority in token-weighted or stake-weighted voter blocs. Off-chain systems can resist simple plutocratic capture, but they often hide power inside maintainership, coordination bandwidth, and the economic weight of operators who decide what code actually runs.
On-chain governance makes power visible, but not equal
Tezos is one of the clearest base-layer examples of formal on-chain governance. Its amendment process runs through five periods. Each period lasts about 14 days. Only delegates vote. Successful proposals must clear a dynamic quorum and an 80% supermajority of Yea over Yea plus Nay votes, and activation happens automatically at the end of the Adoption period.
That structure gives Tezos something many chains do not have. It gives a legible constitutional path for protocol change. The chain itself schedules the vote, counts the ballots, and flips the new rules live if the thresholds are met. That removes a large part of the ambiguity around when an approved change is supposed to take effect.
But formalization is not the same as broad distribution of power. Tezos delegates participate in governance in proportion to voting power, and that voting power is based on the delegate’s own stake plus delegated stake. In other words, the system makes power measurable, but it also aggregates power into the validator class. Delegators keep economic exposure, yet the ballot is cast by the baker or delegate they empower.
Tezos also shows why “on-chain governance” is always partly a misnomer. The final vote is on-chain. The hard political work is not. The proposal development process is described as code, public documentation, release candidates, testnets, stabilization, and feedback loops with bakers, tooling providers, wallets, indexers, and other stakeholders before on-chain validation. The chain ratifies. Humans still draft, negotiate, test, and package.
Still, the throughput advantage is real. On September 19, 2025, Tezos activated Seoul, its 19th protocol upgrade, through this self-amending process. That is concrete evidence that formal on-chain governance can support repeated base-layer iteration without requiring an ad hoc social fight every time.
Off-chain governance removes the ballot box, not the power structure
Bitcoin is the canonical off-chain case. A Bitcoin Improvement Proposal starts on the development mailing list, then moves into a pull request, and the BIPs repository explicitly describes itself as a publication medium and archive. It also states that publication does not mean a proposal has community consensus or is about to be adopted, and that ultimate acceptance rests with Bitcoin users.
That matters because off-chain governance distributes authority across multiple veto points. Bitcoin Core maintainers have commit access and merge patches, but the project describes that role as janitorial and by agreement of contributors. Bitcoin Core has also stated plainly that its development team does not decide Bitcoin consensus rules, because users decide by choosing which software to run, and the software deliberately does not auto-update.
Miner signaling exists in Bitcoin, but it is not the sovereign layer. Under BIP 9, miners signal readiness for a soft fork with version bits. The mechanism uses 2016-block windows, and a soft fork moves from STARTED to LOCKED_IN if at least 95% of blocks in the window signal support, then becomes ACTIVE one retarget period later. That is an activation mechanic, not a full constitutional theory.
The SegWit episode showed the real politics. Bitcoin Optech’s soft-fork history notes that some users believed miners were illegitimately delaying activation, which led to a user-activated soft fork strategy. Miners, exchanges, businesses, and users all mattered. Even now, Optech notes that debate remains over how much each group contributed to the final outcome. Off-chain governance did not avoid power conflict. It exposed that the decisive coalition was social and economic, not merely hash-based.
This is the core off-chain reality. There is no single wallet button that says “vote yes.” Power sits with maintainers who shape release candidates, miners or validators who can coordinate around activation, node operators who enforce rules, and economically central businesses that can legitimize or delegitimize a path. The model is less mechanically plutocratic than coin voting. It is not apolitical.
Ethereum shows why the binary breaks down
Ethereum’s official governance documentation is unusually direct on the hybrid pattern. At the protocol level, governance is off-chain. At the application layer, many systems built on Ethereum, especially DAOs, use on-chain governance. That split is not cosmetic. It reflects a conscious separation between base-layer constitutional change and app-layer operational change.
The formal process for Ethereum protocol change runs through EIPs. Anyone in the community can create an EIP. A Core EIP is presented to protocol developers, discussed in public forums and AllCoreDevs calls, iterated, implemented, tested, bundled into a network upgrade, and then activated on testnets and mainnet. That is off-chain governance with formal artifacts, not pure informality.
Ethereum also documents the cost of that openness. Working groups for serious protocol changes often take several months and sometimes years. The same page says there is no single metric like a coin vote that can gauge community consensus, and no single stakeholder group can coerce the others through brute force. That is slower. It is also a meaningful defense against a simple “buy the governance token, change the protocol” attack surface.
There is an important counterweight here to overly cynical readings. EIP editors are not sovereigns. EIP-1 says editors do the administrative and editorial part and do not pass judgment on proposals. The same pattern appears in Bitcoin, where the BIPs repository says editors should be liberal with publication and not decide on behalf of the community. The administrative gate is real, but it is not the whole power structure.
Tezos, Bitcoin, and Ethereum compared as power systems
| System | Primary governance mode | How changes move | Who holds decisive leverage | Main trade-off |
|---|---|---|---|---|
| Tezos | Base-layer on-chain | Delegates submit and vote through a five-period amendment process. Successful proposals activate automatically after Adoption. | Delegates and bakers, because voting power follows staked and delegated tez. | High procedural clarity and upgrade throughput, but formal authority concentrates in stake-aggregating validator entities. |
| Bitcoin | Base-layer off-chain | BIPs are discussed socially, published in a repository, implemented in software, and adopted only if users run the code. BIP9 miner signaling is one activation method for soft forks. | Maintainers, miners, node operators, and economically important businesses in shifting coalitions. Users retain final adoption power. | Stronger resistance to mechanical token plutocracy, but slower coordination and more ambiguous power centers. |
| Ethereum | Hybrid: off-chain at protocol, often on-chain at app layer | Core protocol changes pass through EIPs, public discussion, client implementation, testing, and coordinated network upgrades. Many DAOs on Ethereum then use on-chain voting for application-level decisions. | Client teams, validators, node operators, app developers, and users. No single metric determines consensus. | Broad stakeholder inclusion at the base layer, but long coordination cycles and persistent ambiguity around contentious cases. |
Participation and speed are where the trade-offs become measurable
On-chain governance makes participation measurable. It does not guarantee that participation is broad. Tezos governance limits protocol voting to delegates, not all token holders directly. More broadly, an empirical study of 21 DAOs found high concentration of voting rights, significant hidden monetary costs, and a large amount of low-value governance activity. That study is about DAOs, not every layer-1, but the concentration warning generalizes well to token-weighted systems.
On-chain governance usually wins on timing certainty. In Tezos, five periods of roughly 14 days each imply an amendment path of about 70 days from process start to automatic activation if everything passes. Off-chain systems rarely offer that calendar certainty. Ethereum states directly that serious EIP working groups can take months or even years, because security review, multi-client implementation, stakeholder feedback, and voluntary adoption all take time.
Speed, however, is not a free lunch. Faster amendment paths make it easier for the winning coalition to move parameters quickly. That is useful for treasury management, emissions tuning, or emergency response. It is also dangerous if the electorate is narrow, borrowed, apathetic, or captured. Vitalik Buterin has argued that coin voting over-empowers wealthy holders, can separate governance power from economic interest through lending and custody, and is better suited to applications than base layers. That is an argument, not a law of nature, but it is a strong description of the capture surface many token systems create.
Off-chain governance has the opposite profile. It is harder to acquire decisive power with a single open-market purchase because there is no universal ballot to buy. The cost shows up elsewhere. You need engineering credibility, implementation capacity, distribution, coordination with major operators, and enough social legitimacy that others will follow your release. Capture still exists. It is just less legible and less reducible to wallet balances.
The practical design question is not on-chain or off-chain. It is how much constitutional power should be liquid
For token economy design, the cleanest framing is a control map. Which parameters can change. Who can propose changes. Who can approve them. What quorum and majority apply. Whether there is a veto. Whether activation is automatic. Whether users have time to exit before a decision takes effect. Those questions matter more than the label.
On-chain governance is strongest when the decision surface is narrow and operational. Treasury disbursements, emissions bands, incentive budgets, or bounded parameter adjustments fit better than wholesale constitutional rewrites. The more of the protocol you expose to token voting, the more you turn governance into a liquid market for control. Timelocks, limited scopes, and credible exit windows reduce that risk, but they do not erase it.
Off-chain governance is strongest when neutrality matters more than rapid iteration. That is why Bitcoin keeps consensus change socially expensive, and why Ethereum keeps protocol governance off-chain even while much of its application layer experiments with on-chain voting. High friction is inconvenient. It is also a defense. It prevents a protocol from being too easy to steer by the largest balance sheet or the best turnout operation.
At FinDaS Tokenomics, this is the starting point for tokenomics consulting and token economy design work. Governance is not a branding layer. It is a power-distribution system with explicit winners, veto holders, and attack surfaces. Any serious tokenomics expert or token economy advisor should map those power paths before talking about decentralization. The hard part is not choosing between on-chain and off-chain as identities. The hard part is deciding which powers should be formal, which should be delayed, and which should be so difficult to exercise that no coalition can casually rewrite the protocol.
