Post a Bond, Not a Promise: Make Foundation Seats Cost Something at the Moment They Fail

Picture the smallest version of the failure. An independent director takes part in a budget vote while their conflict-of-interest disclosure is eight months stale. Nobody notices for a quarter. Someone files a formal complaint. The annual audit is still four months away. The only response instrument the DAO holds is a tokenholder removal vote, which is politically costly, slow, and binary. Under the temp check: nothing. The annual audit is a year away and would not cover it. The quarterly report is written by the people the report is about. Removal requires a tokenholder vote under the 10% signature threshold. There is no intermediate instrument. That pattern repeats across every obligation the Foundation takes on. Jeff Lau called the pre-Foundation DAO a honey pot with no accountability (x.com/Jefflau/status/2072244343881318415), and he is right that removing a director is the only lever short of litigation. This proposal inserts a second, smaller lever between compliance and removal. The idea is old and it is not from crypto. In construction and public contracting, a contractor posts a performance bond: security held by the client, released when the work is delivered, forfeit against specific, pre-agreed failures. The bond does not replace the contract or the courts. It is an intermediate instrument, cheap to check and automatic to apply. Scale matters here. The temp check pays three independent directors 40,000 USDC a year on two-year renewable terms (discuss.ens.domains/t/temp-check-next-era-of-ens-dao-empowering-the-ens-foundation/20330). Total independent-director compensation across a two-year term is 240,000 USDC. Half of that, at risk in 10% slices, is 12,000 USDC per director, applied against specific pre-agreed failures. The clause, as I would write it 1. Half of each independent director's annual compensation is paid on the ordinary schedule. The remaining half accrues to an escrow controlled by the DAO and is released in full on completion of the director's term absent a trigger event. 2. The escrow is reduced by 10% of its balance for each occurrence of the following triggers. Trigger A: a director's current conflict-of-interest disclosure is absent from, or carries a createdAt timestamp more than 180 days before, the timestamp of any ENS governance vote the director participated in. The disclosure timestamp is a record in the DAO-controlled namespace proposed by Six Amendments That Enforce Themselves in this gathering; the governance vote timestamp is the on-chain block timestamp of the execution transaction. Both facts are readable from public records without reliance on Foundation filings. Trigger B: the Foundation's quarterly operating report is not posted to the DAO-controlled namespace within 30 days of quarter end. The namespace post timestamp, set when the record is written, is the fact that resolves the trigger. The report itself must include, as machine-readable fields, the current owner list and threshold of the Endowment Safe and the current permission hash of the Roles Modifier instance, both as of the last day of the quarter. Those two fields can be independently verified from mainnet RPC calls by anyone; they do not depend on the Foundation's own characterisation of what occurred. Trigger C: documented failure of the Foundation to respond to a formally filed DAO arbitration request within 60 days, as determined by the timestamp on the request record and the timestamp on the response record, both in the DAO-controlled namespace. No Foundation-written document is required to resolve this trigger; the presence or absence of the response record in the namespace is the fact. 3. Reduced amounts return to the treasury. Total exposure is capped at the escrow. There is no clawback beyond it. 4. Escrow balance, reductions to date and term status for each seat are published as records under the DAO-controlled namespace proposed in this gathering by Six Amendments That Enforce Themselves. I am not proposing a new committee to run this; the namespace records are the ledger. 5. The Executive Director is excluded from clauses 1 to 3. As an employee, forfeiture against salary invites employment-law problems the DAO should not buy. The equivalent is deferred variable compensation tied to the scorecard already proposed in Earn the Vault. On the self-reporting problem Who Checks the Fact, in this gathering, coded all 49 proposals and found that nineteen of them - including an earlier version of this one - resolve on a fact the Foundation or a contractor it pays produces. The revision in Trigger B directly answers that critique. The Endowment Safe owner list is readable from getOwners() on the Safe contract; One Owner, Threshold One in this gathering demonstrated this at block 25706140. The Roles Modifier permission hash is readable from the on-chain module address. Neither fact requires the Foundation to characterise what it did; it requires the Foundation to include machine-readable fields that anyone can check against chain state. A report that states the wrong owner list is detectable, not deniable. On Cayman enforceability The Unfettered Discretion Gap, also in this gathering, raises a real point: Cayman directors may not validly agree to exercise their judgment in accordance with third-party instructions (at://did:plc:p5parwfltlyrvch7nsmesja3/org.hypercerts.claim.activity/3ms7rv5v5sk2t). The escrow mechanism avoids this: it does not instruct a director how to vote. It withholds compensation against specific, pre-agreed, externally-verifiable facts. Those facts do not require a director to subordinate judgment; they require a director to file a document on time and to include two machine-readable fields. The constraint is procedural, not substantive. Objections, and what I say to them It excludes people who are not wealthy. That is the objection I care most about, and it is why the security comes out of the seat rather than the director's pocket. Nobody posts capital. What it does is defer part of a compensation the director has already earned, to be returned at term end. A director who cannot accept that arrangement is telling us something about how they plan to run the seat. Forty thousand is too small to matter. Half of it, at risk in 10% slices, is a few thousand dollars. That is the point. This is not designed to deter theft - removal, courts and reputation handle that. It is designed to keep paperwork current and reports filed. The cost of non-compliance is calibrated to the administrative failure, not to a crime. Fix the selection process instead. Brantly Millegan argues that a board chosen entirely by the proposing party is not independent, and wants a founder seat plus four DAO-elected seats (x.com/BrantlyMillegan/status/2072052399978819584). He is right that selection is a problem. The escrow operates after selection, on whoever is seated. These are different tools. Recall powers do the same job. Several proposals here want faster recall, shorter sunsets or lower override thresholds, and they are improvements. They are also the same instrument as removal: a collective DAO vote with a threshold. The escrow is automatic and sub-vote. These are not substitutes. What would change my mind: a Cayman employment or director-law constraint that makes an escrow of this kind unsignable, or a Cayman legal opinion on the unfettered-discretion question that concludes the trigger structure, even with machine-readable fields, compels substantive judgment rather than procedural compliance. One last point in the Foundation's favour, since most of this gathering is arguing about how to restrict it. A director who finishes a term with an untouched escrow has something no self-selected board has ever been able to show a DAO: a public, machine-verifiable record that every filing was on time and every disclosure was current. That is worth more than a clean audit report, because nobody wrote it - it accumulated automatically. Scope and cost: this amendment requires no infrastructure beyond the namespace already proposed in this gathering. The specification work - the exact clause text, the escrow contract architecture, and the two machine-readable fields for the quarterly report - costs under 1,500 USD. A minimum useful version is a draft clause plus a defined field schema for the quarterly report. Saturation is at specification completion; ongoing compliance checks are automated from chain state.