Token governance fails in a predictable sequence. There is a burst of participation at launch, a long slide, and then a handful of addresses deciding everything while the rest of the token supply has quietly stopped voting. Delegation is the mechanism most DAOs eventually adopt to reverse that slide — and it works, which is why it is worth understanding how it fails.
Why abstention is rational
Voting costs attention, gas, and time, and a single vote almost never changes an outcome. A token holder who is not a full-time delegate is behaving correctly when they abstain. This is not apathy in the colloquial sense; it is a cost-benefit calculation that the protocol keeps losing.
The response that does not work is adding noise. More forum posts, more snapshot notifications, more governance calls — none of that changes the underlying arithmetic, because the problem is not that holders have not heard about the vote. It is that the expected value of participating is below the cost of learning enough to participate well.
What delegation actually changes
Delegation assigns voting power to someone who will do the work, without transferring the tokens. That separates holding from deciding, and it lets a small number of capable, accountable delegates carry quorum on routine matters. It is the only mechanism that has reliably raised effective participation, and it does so by shrinking the number of people who need to be informed rather than by persuading more people to be informed.
Five design choices determine whether it works or becomes a cartel.
Persistent versus per-proposal. Persistent delegation means one decision covers every future vote, which is convenient and concentrates power. Per-proposal delegation is more democratic and produces more informed outcomes, but requires the delegator to show up repeatedly — which is precisely the behaviour delegation was meant to work around.
Instant recall. A delegator must be able to withdraw support immediately and cheaply. If recall requires a transaction with a delay, or costs meaningful gas, delegators stay with a bad delegate out of inertia.
Caps. A maximum share of voting power any single delegate can hold is the direct answer to a delegate cartel. It is unpopular with the delegates who would be capped, which is why it is rarely adopted.
Cost of delegating. Delegation should be free or subsidised. If it costs gas, you have added a tax to the exact behaviour you are trying to encourage.
Transparency. Delegates should be required to publish their positions and their reasoning. Without it, delegation becomes a blind proxy rather than accountability.
The failure mode: concentration
Delegation concentrates. Well-funded delegates accumulate power because they can afford to be full-time, and a delegate who votes well attracts more delegation, which makes their vote worth more, which attracts more. The equilibrium of an unconstrained delegation market is a small number of large delegates — which is representative government with worse disclosure and no term limit.
That is not an argument against delegation. It is an argument for the constraints above being load-bearing rather than optional. Recall, caps, and disclosure are what keep delegated systems from ossifying into an oligarchy that the token holders technically still control.
The mix most DAOs converge on
The workable pattern has settled into something recognisable: delegated voting for routine parameters, where the cost of a bad decision is bounded and reversible; direct votes for constitutional changes, where it is not; and quorum thresholds calibrated so that a small group cannot pass something significant while the wider holder base is asleep.
Two further details matter more than they look. Proposal thresholds — the amount of delegated support required even to put something to a vote — filter spam and force proposals to build a case before consuming everyone's attention. And timelocks on executed decisions turn a governance capture from a fait accompli into an event that the wider community can see coming and react to.
The honest summary
Liquid democracy does not solve governance participation; it relocates the effort from thousands of holders to dozens of delegates, and pays for that with concentration risk. Whether that is a good trade depends entirely on whether recall and caps are real. Without them, delegation is just a more efficient way to end up with an oligarchy.
References
Next, read Public Goods Funding 101. More in Governance & DAOs.