Ask a DAO what its governance costs to run and you will get a number. Ask two DAOs and you will get two numbers that cannot be put side by side. This is not sloppiness. The figure a DAO publishes is a function of three choices it made for unrelated reasons: which legal entity holds the budget line, which currency the budget is denominated in, and which activities count as governance rather than as the work governance directs. Change any one of them and the reported cost moves without a single person doing anything differently.

This page is about reading those figures. It works through three DAOs that publish enough detail to be checked: Optimism, whose season reports give a multi-season series; ENS, which paused its entire working-group structure in 2026 and had to price the pause; and Arbitrum, whose largest governance-adjacent costs sit in places no governance budget reports. The neighbouring pages cover the pieces: what DAOs pay delegates is on delegate incentive programs, what they pay contributors is on contributor compensation, and the bodies that spend it are on subDAOs, working groups and pods.

The reclassification problem: Optimism’s +38%

SEEDGov’s Season 7 report for the Optimism Collective states the Token House’s governance operating costs as 1,055k OP, against 762k OP in Season 6, and calls it a 38% increase. The four components it lists sum exactly to the total:

Grants Council	400k OP	down 34% from Season 6’s 610k OP; the council shrank from 15 reviewers to 7
Milestones & Metrics Council	170k OP	newly independent, previously inside the Grants Council budget
Developer Advisory Board	190k OP	up from 90k OP, grown to 9 members
Security Council	295k OP	first season its budget was public and Token House approved, having moved from Foundation expenses to the Governance Fund

The report flags the cause of the jump in its own words: the rise is expected, it says, because the Security Council’s costs became part of the public budget for the first time. What it does not do is the subtraction. Take the 295k OP reclassification out and Season 7 governance cost 760k OP against Season 6’s 762k OP. On a like-for-like basis the number did not rise 38%. It did not rise at all.

This is the single most common defect in DAO cost reporting, and it is structural rather than dishonest: the same body was funded in both seasons, but in Season 6 the Foundation paid for it and in Season 7 the DAO did. Optimism’s report is unusually good precisely because it discloses the change. A reader who only saw the headline would conclude the Collective’s governance got 38% more expensive in a season where it also cut the Grants Council by a third.

The sequel matters for how this figure should be read. In June 2026 the Optimism Foundation proposed dissolving the Grants Council, the Milestones and Metrics Council and the Developer Advisory Board – the three bodies that made up the 760k OP – on the grounds of governance overhead, while explicitly carving out the Security Council as worth its cost. The cheaper 72% of the structure was the part that closed. See council dissolution and governance unwinding.

The ratio moves even when the cost does not

Governance operating cost is close to fixed over a season. The money governance directs is not. Optimism’s Season 7 Token House approved a single Intent budget of 9.5M OP, down 51% from Season 6’s 19.5M OP. Setting the two published figures against each other gives an overhead ratio, the share of a season’s approved OP that goes to running the process rather than to what the process funds:

	Season 6	Season 7
Governance operating cost	762k OP	1,055k OP (760k like-for-like)
Intent budget	19.5M OP	9.5M OP
Overhead ratio	3.9%	11.1% (8.0% like-for-like)

The ratio roughly doubled on a like-for-like basis, and more than doubled as reported, in a season where governance spending was flat. Every unit of the move came from the denominator. This is worth stating because the two readings support opposite conclusions from the same ledger: governance got no more expensive and governance now consumes twice the share of the treasury it deploys are both true of Optimism’s Season 7, and a DAO arguing either side can cite the report honestly.

The construction here is ours, not the report’s. The report presents operating costs and the Intent budget as separate figures and does not divide one by the other. The comparison is defensible because both are OP outlays approved by the same body for the same season, but a reader should treat the ratio as a derived quantity, and one that a DAO can move by changing the size of its grant programme without touching governance at all.

ENS: what a structure costs, and what pausing it costs

ENS DAO gives the clearest published answer to “what does the apparatus cost”, because in 2026 it had to price the apparatus twice: once running, once idling.

The working groups were funded in the October 2025 window at $959,000 USDC and 15 ETH: Meta-Governance $379,000, Ecosystem $470,000, Public Goods $110,000 plus the ETH. In December 2025 the DAO voted to pause working-group elections so stewards could support an independent retrospective, and the Meta-Governance stewards put a temp check to the community asking whether the groups should keep that full allocation or drop to a continuity budget. The continuity option was $58,500 per month, $234,000 over four months:

Steward compensation (9 stewards)	$40,500 / month	$162,000
Secretary	$5,500 / month	$22,000
Scribe	$2,500 / month	$10,000
Operational reserve (legal retainer, tooling, ops)	$10,000 / month	$40,000

Two things fall out of that table. The first is the composition: $194,000 of the $234,000, or 83%, is compensation for eleven people holding governance roles. Everything else a working group needs, legal retainer and tooling and operations together, is the remaining 17%. A DAO’s floor cost is its roster, not its infrastructure.

The second is the retrospective itself. The executable proposal that commissioned it requested $125,000 to the Meta-Governance multisig, released against milestones and clawable back by social proposal at any point. Over the same four months, the entire working-group structure being studied ran on $234,000. Examining the spending cost 53% of what the thing being examined cost to operate during the examination. That is not an argument against the retro, which produced a published report and a twelve-month roadmap. It is an argument that meta-governance is itself a cost class, and one that no DAO reports as such.

The pause had a further cost that never appears in any budget. By May 2026, with the retro published several weeks past its April target and two structural proposals still open, stewards were described as operating at reduced capacity with no clear mandate and no election timeline. Term 7 was eventually sequenced to begin on 1 July 2026, roughly seven months after elections were first delayed.

Arbitrum: the largest lines sit outside “governance”

Arbitrum makes the boundary problem concrete, because its two biggest governance-adjacent costs are not counted as governance costs by anyone.

The strategy firm. Entropy Advisors works exclusively with the Arbitrum DAO under a two-year mandate running 1 September 2025 to 31 August 2027, at a base fee of $3M per year, claimable as monthly payments of $250,000 from the Arbitrum Foundation. That is an increase of roughly $500,000 a year on the first term, which ran at $205,834 a month. The engagement also carries a 5M ARB vesting allocation on a one-year cliff and three-year vest, with a further 10M ARB held by the Foundation for milestone-based alignment mechanisms to be negotiated separately. The firm publishes monthly updates to the forum and was still doing so in June 2026.

The operating company. Arbitrum’s OpCo is a DAO-adjacent legal entity created to execute strategy. Its second bi-annual transparency report, covering activity to 30 April 2026, states cumulative expenses of $1,148,778 and 100,000 ARB, of which $568,179 was incurred since the previous report six months earlier. The breakdown is the interesting part:

Admin & Operations	$561,907
OAT compensation	$367,500
Incorporation and legal	$192,316
Event hosting	$13,784
Firestarter grants	$13,270

$192,316 of it is the cost of existing as an entity at all, before the entity does anything. That is the price of the boundary itself: a DAO that wants a legal counterparty to sign contracts, hold funds and employ people pays a six-figure setup, and every cost that then moves inside that entity leaves the DAO’s own governance budget. Optimism’s +38% is the same mechanism running in reverse.

Arbitrum’s third line, the delegate payroll, is the one that is reported as governance spending, and it is by some distance the smallest of the three. Its July 2026 quarterly review repriced the whole schedule downward as ARB fell, cutting per-proposal rates 28–40%. That programme is covered in detail on delegate incentive programs, and the commercial relationship behind the first two lines — how a provider is selected, renewed and terminated — on DAO service providers.

Reading a governance cost figure

Four questions make two DAOs’ numbers comparable, or establish that they are not:

1. Which entity holds the line? A cost paid by a foundation, an operating company or a service provider does not appear in the DAO’s governance budget. Optimism’s Security Council cost the same in Season 6 and Season 7; only the payer changed. Ask what the foundation covers before comparing anything.
2. What is it denominated in, and what is it paid in? A budget set in USD and settled in the governance token spends the most tokens exactly when the token is cheapest, and a budget set in the token reports a flat number while its real cost swings with the market. Optimism reports in OP; ENS reports in USDC; Arbitrum sets delegate rates in USD and pays in ARB. None of the three series is directly comparable to the others in any year.
3. What is inside the scope? Councils and stewards almost always count. Strategy firms, operating companies, legal entities, security audits and tooling subscriptions usually do not, and they are frequently larger.
4. Did anything get reclassified? A multi-season series is only a series if the boundary held across it. When a body moves between payers, the year-on-year change is an accounting artefact until you net it out.

A useful discipline for a DAO publishing its own figures is to report the cost twice: once on this year’s boundary and once on last year’s. Optimism’s report comes closest of the three by disclosing the reclassification in the same paragraph as the headline, which is what makes the correction possible at all.

How Caper approaches this

Every cost class on this page traces back to the same root: governance in a token DAO is work done by people who must be found, elected, compensated and audited, and the apparatus that does the finding, electing, compensating and auditing is itself an ongoing expense. Caper does not have a cheaper version of that apparatus. It does not have the apparatus.

There is no delegate, steward or council class in the contracts. A search for delegation primitives across all three crates returns nothing, so there is no standing body whose seats need funding and no delegated weight to subsidise or retain. The two things a member can do in governance, propose and vote, are both priced as fees paid into the caper’s treasury rather than as budget paid out of it: create_proposal requires an XRD payment equal to the configured proposal fee, and the core contract deposits it straight into the treasury vault; vote requires an XRD payment equal to the configured vote fee, and deposits it into the same treasury even on the path where a ballot arrives outside the voting window and is rejected. Both fees are settings an admin can change through update_fees, not constants, so the schedule is a policy rather than a property of the protocol.

The honest statement of the tradeoff is that this moves the cost rather than removing it. A caper’s governance has no payroll because it has no roles, and the participation it gets is whatever members will pay a fee to supply. That buys a treasury whose governance line is structurally positive and a cost series with no boundary to reclassify, and it gives up the specialised review that a funded council is meant to provide. Which side of that trade a given organisation wants is exactly the question the ledgers above are trying to settle. Background on the model is on what is a caper, and the flow those fees ride on is on proposals and voting.

References

• Token House participation and incentives: Season 7 (Cycle 31a-38) — SEEDGov, gov.optimism.io. Season 7 operating costs, the S3–S7 budget evolution, and the Intent budget.
• [Temp Check] Metagov Update: Working Group Operations During the ENS Retro — discuss.ens.domains. The continuity budget and the October 2025 approved allocations.
• [Executable] ENS Stakeholder Analysis and Retro — discuss.ens.domains. The $125,000 retrospective budget and its milestone terms.
• Path forward on Working Groups for Term 7 — discuss.ens.domains. The election-pause timeline and the Term 7 sequencing.
• Entropy Advisors: Exclusively Working with the Arbitrum DAO, Y2-Y3 — forum.arbitrum.foundation. Base fee, payment schedule and the ARB alignment structure.
• OpCo/OAT 2nd Bi-annual Transparency Report, May 2026 — forum.arbitrum.foundation. Cumulative OpCo expenses by category to 30 April 2026.
• RAD Budget Quarterly Review 2Q2026 — forum.arbitrum.foundation. The repriced delegate payroll.