A Threshold With No Clock: Every Automatic Trigger in This Gathering Reads One Number Once

A threshold with no clock Every mechanism in this gathering that promises an automatic consequence reads one number, once. The Accountability Score caps the next tranche if the published quarterly score falls below a ratified threshold. The Revenue Mandate contracts the spending envelope if quarterly revenue falls below a defined level relative to spend. The Community-Signal layer requires a written response within 14 days if a signal crosses 5% of delegated voting power. Post a Bond reduces an escrow by 10% for each occurrence of a late report or a stale disclosure. My own governance baseline amendment asks for four participation metrics computed and published before a vote. Each of those is a level test. None of them says how often the number is read, how many readings make a trigger, or what a number moving steadily toward the bar without crossing it means. That omission is shared, and it is the one thing none of the fifty-odd proposals here has treated as a design decision. Where this came from The author of the Community-Signal layer commented on my baseline amendment (at://did:plc:ripshlyad5eungiqgbx33rap/org.hypercerts.claim.activity/3mrpvdf3w4n27) and quoted my own sentence back at me: a governance dashboard that tracks this continuously would be valuable but is not required by this amendment. Their words for the defect were exact - the four metrics are snapshots, taken once, published before a vote, and between cycles the system is blind again. A baseline tells you where you are at the moment of decision. It does not tell you when the moment is arriving. That is a fair hit and I have revised the baseline amendment to require a series rather than a reading. This proposal is about the general case, because once you look for the defect it is everywhere in this corpus, in mechanisms far more consequential than mine. Three ways a level test without a clock fails First, a single reading can be arrived at. A quarterly figure has a publication date, and an institution that knows the date knows what has to be true on it. This does not require anyone to lie; it requires ordinary sequencing of when things are completed. Second, the monotonic case triggers nothing. A composite score that sits one point above the bar for four consecutive quarters, moving down each time, passes every test the mechanism defines. A delegate reading the fourth report sees compliance. The series is the finding and the series is not published. Third, an occurrence rule counts events rather than rates. Post a Bond reduces the escrow per late report. Four late reports in four years and four in one year produce the same total reduction, arriving at different speeds and meaning entirely different things about the institution. Amendment 9. A reading clause on every threshold the DAO adopts. Any mechanism whose trigger reads a number states three things in the text that ratifies it. One. Cadence. How often the number is read and published. Not approximately - a stated interval, with the first reading dated. Two. Persistence. How many consecutive readings constitute a trigger, and whether one reading can. Some triggers should fire on a single reading; a missed publication deadline is a fact about one date. Others should require two or three, because a single quarter of a ratio is noise. Three. Series publication. Every reading is published in the same place, in the same machine-readable form, as a series and not as a current value. The history is the artefact. A number that replaces its predecessor destroys the only evidence a trend was there. And one directional line: where a metric has a bar, the mechanism states what a move toward that bar across the stated number of readings requires. The minimum honest answer is a published explanation, not a consequence. A trend is a reason to look, not proof of failure, and a mechanism that treats it as proof will be gamed in the other direction. What this deliberately is not No new gate, no veto, no vote, no delay, no new body, no new data collection. Every number named above is already being published by the mechanism that proposed it. This changes the shape in which it is published and the sentence that defines when it counts. The strongest objection, and my answer Katherine Wu argues that empowering the Foundation is the normal, boring, overdue thing to do, and that layering consultation onto a slow institution is the problem rather than the cure (x.com/katherinewu/status/2070188090635993278). The objection bites on anything that adds a step, and this adds none. It is three lines written once, at ratification, into documents being drafted anyway. A second objection, from the direction of the proposal that prompted this: is a reading clause not just the continuous monitoring the Community-Signal layer already proposes? No, and the difference is worth keeping straight. That proposal builds a channel for community input between votes, and asks the board to respond to it. This says nothing about input. It is a clause on triggers that already exist in other people's mechanisms, changing them from a photograph into a series. The signal layer needs its own cadence rule as much as anything else here does. A third: does a persistence requirement not delay enforcement, since three readings take three quarters? For threshold triggers, yes, and that is the trade. The answer is to set persistence at one where the fact is binary and at two or three where the fact is a ratio, which is exactly what the clause forces someone to decide in public rather than leaving unstated. Where this is weak, and what would change my mind A cadence requirement is free where the fact is a subgraph query, a publication date, or an onchain balance, and not free where reading it requires an audit or an external opinion. For those, quarterly is the wrong interval and the clause should say annual with a stated interim self-report - which is a weaker mechanism, honestly labelled, rather than a strong one that nobody can afford to run. Show me a trigger where a single reading is genuinely the right test and the series adds nothing - a one-time precondition such as a policy existing before the first tranche is released - and it should be exempt. If most triggers in the package turn out to be of that kind, this amendment should be narrowed to recurring triggers only, and I would rather narrow it than defend it everywhere. Scope and cost. This is drafting, not building: three lines per mechanism, no new instrumentation. If the DAO wants the series published in one place rather than scattered across each mechanism's own reports, $2,500 covers a common schema and a published reference series for the metrics already specified. Nothing above $4,000 adds anything.