Governance tokens are described with the vocabulary of shareholding, and the analogy breaks at every point that matters. There is no claim on assets, no claim on profits, no fiduciary duty owed to anybody, and in most cases a small group who can act without a vote at all.
A share in a company is a claim: on the residual assets if it is wound up, on dividends if they are declared, and on a body of law that requires directors to act in the interests of shareholders. Those three are what make a share worth something independently of what anybody will pay for it, and a governance token has none of them.
What it has is a vote on proposals, within whatever scope the protocol's code permits. That is a real power and it is a narrower one, because the scope is set by the code rather than by a duty. A holder cannot compel a distribution, cannot sue for mismanagement in any straightforward way, and has no claim on a treasury that the protocol controls.
The consequence is that the standard valuation question, what is this token worth, has no equity-style answer. Its value is whatever people will pay for the ability to influence decisions plus whatever cash flows the holders have voted themselves, and the second is usually zero and the first is a matter of opinion.
A share is a claim enforced by law. A governance token is a permission enforced by code. Both are real; only one gives you anything if the organisation succeeds and never pays you.
Reading the actual scope of a governance system is more informative than any discussion of it, and it is usually narrow. Typical powers include adjusting parameters such as fees, collateral ratios or supported assets; allocating funds from a treasury; and approving upgrades to the contracts.
The third is the significant one, because a contract upgrade can change anything the contract does, including what future governance can decide. Where an upgrade path exists, it is the real seat of power, and every other proposal is subordinate to whoever controls it. Whether governance genuinely controls the upgrade, or merely advises somebody who does, is the single most important structural fact about any of these systems.
What is typically outside scope is worth listing too, because people assume otherwise. Governance usually cannot compel the founding team to do anything, cannot direct off-chain entities holding trademarks or running interfaces, and cannot access funds held outside the treasury contract. A vote is binding on the contracts and advisory on the people.
Participation in these votes is very low almost everywhere, routinely single-digit percentages of eligible tokens. This is not apathy in any surprising sense: voting costs a transaction fee, requires reading a technical proposal, and returns nothing to an individual holder whose stake cannot change the outcome. The rational response is not to vote, and most holders take it.
Delegation exists to solve this, letting holders assign their voting power to somebody who will use it. It works, and it produces a concentration that the original design was meant to avoid: a small number of delegates end up controlling a large share of the votes, and they are usually funds, protocol teams, or professional participants rather than the distributed holders the token was issued to.
The number worth knowing about any system is therefore not how many tokens exist but how few addresses are needed to reach a quorum and pass a proposal. That figure is computable from public data, it is frequently in the single digits, and it describes the governance far more accurately than the token distribution does.
If voting power is purchasable and the treasury is valuable, there is an arithmetic that somebody eventually does: acquire enough voting power to pass a proposal that transfers more value than the acquisition cost. This is not hypothetical and it has been executed successfully more than once.
The mechanics are cheaper than they sound because the attacker does not need to buy the tokens permanently. Lending markets allow borrowing voting power for the duration of a vote, and some designs allowed acquiring and voting within a single transaction, which reduced the cost of an attack to a fee. Timelocks and snapshot mechanisms exist specifically to prevent that, and their presence or absence is checkable.
The defence that matters most is that the value controlled by governance should not exceed the cost of capturing it. A treasury much larger than the market value of the voting power that controls it is a standing invitation, and it is a condition that can be measured in an afternoon by anybody willing to compare two public numbers.
Almost every protocol that describes itself as governed by token holders also has a multisignature wallet held by a small group, usually with the ability to pause the system, and frequently with the ability to upgrade contracts directly. It exists for good reasons: somebody has to be able to respond to an exploit faster than a governance process can convene.
It also means the honest description of the power structure includes it. Where a small group can act unilaterally in an emergency, and where that group defines what constitutes an emergency, governance is a process that operates with their consent. That is not necessarily bad and it is not what the word decentralised implies to most readers.
The details are public and worth a minute. How many signers, how many are required, whether they are identified, whether their powers are time-limited, and whether a timelock delays their actions long enough for holders to exit. A system with a two-of-three multisig and no timelock is governed by three people, whatever the token does.
Read the multisig before reading the proposals. If a small group can upgrade the contracts, the votes are advisory and everything else is presentation.
Many protocols hold substantial treasuries, and governance's most consequential real power is allocating them. This is where the incentives concentrate and where the failures cluster, because a treasury is a pool of money controlled by a process with low turnout and concentrated delegation.
The recurring pattern is not theft but drift. Grants to entities connected to the proposers, budgets that grow without review, payments in the protocol's own token that transfer dilution to holders who did not vote, and spending decisions taken by a handful of delegates on behalf of thousands who did not participate. Each individual decision is defensible; the aggregate is a treasury spent by people who did not fund it.
The number that describes this is the rate at which the treasury declines relative to what it produces, and it is computable from public transactions. A protocol spending its treasury faster than its revenue replenishes it has a finite runway, exactly like any other organisation, and governance discussions rarely frame it that way.
Many governance tokens carry an implicit promise: the protocol generates revenue, and holders could vote to direct some of it to themselves. That possibility does substantial work in how the token is valued, and in most cases it has not been exercised.
The reasons given are usually sound. Directing revenue to holders may change the token's regulatory characterisation in several jurisdictions, it reduces the funds available to grow the protocol, and it can make the protocol less competitive against rivals that charge less. Each is a real argument and together they explain a deferral that has lasted years across most of the sector.
The honest reading is that a token valued on a cash flow that governance has repeatedly chosen not to enable is valued on a decision that has not been taken and might never be. That is a legitimate thing to hold a view about, and it should be held as a view rather than as an entitlement, because nothing obliges anybody to enable it.
Five things, all public. Whether governance controls the upgrade path or merely advises somebody who does. How many addresses are needed to reach quorum and pass a proposal, which is usually a very small number. Whether a timelock separates a passed vote from its execution, giving holders time to exit if they disagree. What the multisig can do without a vote. And whether the value controlled exceeds the cost of acquiring control.
The pattern in the answers separates systems where a vote decides something from systems where it ratifies. Neither is inherently wrong, and the second is far more common than the vocabulary suggests. Knowing which one you are holding is the difference between participating in a process and providing legitimacy to one.
No, and the differences are the substantive ones. A share carries a claim on residual assets, a claim on distributions if declared, and a body of law requiring directors to act in shareholders' interests. A governance token carries a vote within whatever scope the code permits, and none of those three.
Typically parameters such as fees and collateral ratios, allocations from a treasury, and upgrades to the contracts. The third is the one that matters, since an upgrade can change anything including what future governance may decide. Whether governance controls upgrades or merely advises somebody who does is the key structural fact.
Because voting costs a transaction fee, requires reading a technical proposal, and returns nothing to a holder whose stake cannot change the outcome. Not voting is the rational response and most holders take it, which is why delegation concentrates power in a small number of professional participants.
It has been done. Voting power can often be borrowed for the duration of a vote rather than bought outright, which reduces the cost substantially. The structural defence is that the value governance controls should not exceed the cost of capturing it, and that comparison uses two public numbers.
A wallet held by a small group, usually able to pause the system and frequently able to upgrade contracts directly. It exists so somebody can respond to an exploit faster than a vote can convene. It also means governance operates with that group's consent, which is a different structure from what the vocabulary implies.
A delay between a proposal passing and taking effect. Its purpose is to give holders who disagree time to exit before the change applies, and its absence means a passed proposal executes immediately. Whether one exists, and how long it is, is among the most informative facts about a governance system.
Because directing revenue to holders may change the token's regulatory characterisation, reduces funds available to grow the protocol, and can make it less competitive. Those are real arguments. The consequence is that a token valued on that cash flow is valued on a decision that has not been taken.
Check whether governance controls the upgrade path, how few addresses are needed to reach quorum, whether a timelock separates passage from execution, what the multisig can do without a vote, and whether the value controlled exceeds the cost of capturing control. All five are public, and together they take about twenty minutes to establish for any given protocol. That is a short exercise against a decision people frequently make on the basis of a discussion forum and a token price, and it is the only version of the question that has an answer rather than an opinion.