A DAO governance model is the rule that decides who holds voting power, how that power is measured, and how individual votes combine into a binding decision. It is the most consequential design choice a DAO makes, because the model is encoded in a contract that runs without a referee and is exposed to anyone who can acquire, borrow, or manufacture voting weight. Every model is an attempt to sit somewhere useful on one central axis: capital-weighted (resistant to fake accounts, but plutocratic) versus participation- or identity-weighted (egalitarian, but fragile to one actor posing as many). This page surveys the families; the voting mechanisms article goes deeper on the social-choice theory behind each, and how DAOs fail collects the post-mortems these models are built to avoid.
The economists Darcy Allen and Aaron Lane of RMIT University's Blockchain Innovation Hub offer a way to compare the families. They treat a vote as a bundle of property rights, each right letting its holder do some things with the vote and not others, and argue that once votes are recorded on a blockchain and handled by smart contracts, the contents of the bundle become a choice written into the protocol: whether a vote can be delegated, whether it can be bought and sold, and whether buying many votes is made prohibitively costly, as quadratic voting does (Cryptodemocratic Governance of Special Economic Zones, 2020). They wrote about special economic zones, areas that run under different rules from the country around them and face high migration and policy uncertainty, and name DAOs as one structure such a zone could adopt. Each model below grants a different bundle.
The on-chain default: voting power is proportional to governance tokens held, and a proposal passes when support clears a threshold, usually alongside a quorum (minimum participation) and a proposal threshold (minimum stake to submit). It is simple, legible, and Sybil-resistant by construction – buying influence means buying the scarce token. The dedicated token-weighted voting page covers its on-chain mechanics in depth; its weaknesses are equally well documented: plutocracy (whales decide), voter apathy (a small holder's vote rarely matters, so turnout runs in single digits), and vote-buying, since a voter who merely borrows tokens has, as Vitalik Buterin argues, "no financial exposure" to the outcome. Forbes summarized a decade of research in 2026 under a blunt headline: DAOs keep centralizing. (Chainlink: governance tokens)
Quadratic voting (QV) lets a voter cast multiple votes on an issue, but the cost rises with the square of the votes – one vote costs one credit, ten votes cost a hundred – so people buy influence roughly in proportion to how much they care, and no single party can cheaply dominate. Introduced to crypto by Buterin and Glen Weyl, QV surfaces intensity of preference rather than just direction. Its fatal on-chain flaw is identity: split your tokens across many wallets and the quadratic penalty evaporates, so QV is only safe atop a working proof-of-personhood layer. Its cousin quadratic funding applies the same math to donations and powers Gitcoin Grants. (a16z governance FAQ)
Conviction voting removes the deadline: voters stake tokens on the proposals they support and their conviction accumulates the longer the stake stays put, so a proposal executes once built-up conviction crosses a threshold scaled to how much treasury it asks for. Because conviction is time-weighted, last-minute rented-capital swings are useless – you cannot buy sustained conviction in a single block. Pioneered by the Commons Stack and first run in production by 1Hive's Gardens, it suits continuous funding decisions but rests on a capital-weighted base and is, by design, slow.
Reputation models grant voting power as non-transferable reputation earned through contribution rather than purchased on a market. Because reputation cannot be bought or sold, it resists the plutocracy and vote-buying that afflict tradable tokens – but it pushes the hard problem onto how reputation is awarded, and it can ossify into an in-group if early members accumulate permanent advantage. Variants appear in Moloch-style share systems and in dedicated frameworks such as the original DAOstack/Genesis design.
Not every DAO votes on everything. Many – especially early or small ones – delegate execution to a multisig (an m-of-n Safe signed by trusted stewards), accepting some centralization for speed and safety. Optimistic governance formalizes this: a council or delegate can act unless token holders raise a challenge within a window, keeping day-to-day decisions cheap while preserving a tokenholder veto. Optimism's Collective runs a bicameral version (a Token House plus a non-token Citizens' House), and Lido's Dual Governance gives stETH holders a safeguard veto over LDO votes.
Futarchy, proposed by economist Robin Hanson, splits a decision in two: "vote on values, but bet on beliefs." Token holders democratically pick a success metric; prediction markets then decide which proposal is most likely to raise it, and the market's verdict carries. It puts forecasts in the hands of those willing to stake money on them, but depends on liquid, hard-to-manipulate markets – which is why it remains more experimented-with than adopted. See the full breakdown in Futarchy.
Independent of which family a DAO picks, vote-escrow (ve) lockups time-weight voting power: lock tokens for up to four years and your weight scales with the lock length, then decays as it runs down. Popularized by Curve's veCRV, the aim is to hand the wheel to long-term aligned holders rather than transient capital – at the cost of liquidity and, in practice, a secondary market in vote "bribes" (the so-called Curve Wars).
There is no universally best rule. The choice trades off decentralization, resistance to manipulation, and broad participation – push hard on any one and the others give. Token voting buys Sybil-resistance and simplicity at the cost of equality; quadratic and reputation schemes buy equality at the cost of needing a trustworthy identity layer; conviction and ve-models buy resistance to rented capital at the cost of speed; and a handful of protocols sidestep the tradeoff entirely by making their core rules immutable, leaving almost nothing to govern (a stance the progressive-decentralization literature calls governance minimization). Mature designs increasingly combine stake with something money cannot cheaply buy – time, participation, or verified identity – and pair the rule with defenses against governance attacks like snapshotting weight at a past block. (survey of DAO governance models)
Caper takes the combining route. Decisions use ranked-choice voting over proposals that can carry real on-chain actions, and weight that is the product of two things: the stake a member holds and the participation they have actually shown, earned as non-transferable tokens minted one per ballot cast and on nothing else – trades stopped minting them on 11 September 2026.
Because those tokens never leave the account that earned them, the participation factor cannot be bought off another holder or borrowed from one, and capital flash-loaned for the length of a block carries the lender's record rather than acquiring one – which closes the flash-loan door without a past-block snapshot of holdings; influence accrues to members who have transacted with the caper themselves rather than to capital passing through. That same earned weight, not a raw token balance, also sizes each member's claim on the treasury at exit – voice and exit are the same number. The specifics live on the linked governance pages.