Just For Now: How Emergency Decisions Quietly Become Permanent

Every community that lasts long enough will have a night like this.

Something breaks. The funding falls through, or a key person leaves without warning, or there is a safety problem that genuinely cannot wait for the next full meeting. Someone — usually the most capable, most trusted person in the room — makes the call alone. There is no time for the proposal process. It works. Everyone is relieved, and quietly grateful that somebody was willing to act.

Eight months later, that decision is still in force. Nobody has reviewed it. It was never written down as a rule, so it isn't quite a rule — but nobody would dream of acting against it either. Ask why things are done this way and the honest answer is: because of that week in November.

This is the failure the Blueprint calls Emergency Rule Bypass Precedent, and of the twenty-odd scenarios in the RCOS stress test suite, it is the hardest one to see from the inside. Most failure modes have something to point at — a founder who won't let go, a clique that meets before the meeting, money that stopped being visible. This one is built entirely out of good decisions made by people acting in good faith. That is exactly what makes it dangerous. As the stress test puts it: the exception is quietly becoming the rule.

The problem is not the emergency. It's the missing way back.

Communities need the ability to move fast. A group that has to convene a full consent process before it can shut off a burst water pipe is not principled, it is brittle. Nobody at EcoHubs is arguing for a community that cannot act.

The failure is narrower and much more specific than "someone used emergency powers." It is the absence of a return path — no defined end date, no review, no moment where the group looks at what was done under pressure and consciously decides to either let it expire or adopt it properly.

Without that return path, every crisis leaves a deposit. The rules do not get repealed; they get layered over. And because each individual exception was reasonable, there is never an obvious point at which to object. The stress test rates this High severity for a reason: each unreviewed exception is a precedent, and precedent is how a constitution gets rewritten without anyone voting.

This pattern is extremely well documented — just usually at a larger scale

It helps to know that communities are not uniquely bad at this. Societies with lawyers, archives, courts and centuries of practice fail at it in precisely the same shape.

Rome, of all places, got the mechanism right first. The Republic's dictatorship was a real emergency power — genuine, concentrated authority — but bounded by a hard cap: six months, or until the specific task it was granted for was done, whichever came first. It was appointed for a named purpose, not a general condition. Dictators often resigned early; Cincinnatus stepped down after about fifteen days in 458 BC. That constraint held for roughly three centuries.

Then, in 82 BC, Sulla had himself appointed under a new law with a new justification — "for the reconstitution of the Republic" — and, critically, no expiration date. Caesar reused the machinery and in 44 BC took the title dictator perpetuo: dictator in perpetuity. The Republic did not fall because someone invented emergency powers. It fell after someone removed the sunset clause from powers everyone had already accepted as legitimate.

The modern versions are less dramatic and more instructive, because they show the three distinct ways the return path fails.

The renewal that never stops. Egypt's emergency law was in force from 1981 until May 2012 — renewed by the president every two years for three decades. The sunset clause existed. It simply became a recurring administrative task.

The review that nobody performs. The US National Emergencies Act of 1976 was written specifically to stop this. It requires an emergency to lapse after one year unless actively renewed, and it obliges each house of Congress to meet every six months to consider voting an emergency down. By late 2017, not once had Congress met to hold that vote — while dozens of emergencies remained active, including one declared in 1979 that had by then been renewed for the thirty-eighth time. This is the most important lesson in the whole article: a sunset clause with no mandatory review is just a calendar reminder that nobody reads.

The exception that gets promoted. France declared a state of emergency after the November 2015 Paris attacks, extended it five times, and formally ended it in November 2017 — by writing its central powers into ordinary law. The emergency ended. The exception didn't. It was naturalized.

One more finding is worth sitting with, because it undercuts the argument that always gets made in the moment. A study of pandemic responses across 144 countries, published in Social Science & Medicine, found that illiberal emergency practices did not correlate with better outcomes. The shortcut is justified by results. The results often don't show up.

Why small communities are more exposed, not less

The intuition is that a group of thirty people who know each other well is safer from this than a nation-state. It is not, and the reasons are worth naming honestly.

There is a well-replicated finding in organizational research — the threat-rigidity effect, first set out by Staw, Sandelands and Dutton in 1981 — that groups under threat do two things at once: they narrow the information they take in, and they concentrate control higher up. Dissenting or peripheral input gets filtered out precisely when it is most needed, and decision-making quietly consolidates around whoever is already central. This happens at group level, not just organizational level. It happens to thirty people in a room. It is a stress response, not a character flaw, and no amount of shared values switches it off.

Layer on top of that the conditions of a small, high-trust community. There is usually no archive, so the only record of what happened under pressure is memory — and memory reconstructs. The person who acted was almost certainly right, which makes revisiting it feel like an accusation rather than an audit. And in a group built on care, asking "under what authority did we do that?" reads as distrust of a friend. So nobody asks. The review a parliament would at least be embarrassed about skipping is here skipped warmly, and with love.

Five signs it is already happening

The stress test lists these plainly. Read them as a self-check rather than a verdict — most communities will recognise at least one.

• During a crisis, normal process was suspended "just for now" — and never fully resumed.
• A decision made under urgency quietly became the new permanent structure.
• Emergency powers were never defined in advance, so their scope and end date are unclear.
• "We didn't have time to do it properly" is becoming a recurring justification.
• There is no log distinguishing emergency exceptions from permanent rule changes.

That last one is the quiet killer. If exceptions and amendments live in the same undifferentiated pile — or in no pile at all — then the difference between "we suspended this once" and "this is how we do things" is decided by whoever remembers it most confidently.

Most communities we talk to sit at L0 or L1. That is not a failing grade — it is the honest starting point, and it is only a problem if it stays invisible.

What the Blueprint actually requires

RCOS does not try to prevent emergency action. It constrains it, in three places that interlock.

Layer 2 bounds the authority. §4.3.5 states that temporary or emergency authority "MUST be explicitly defined, time-bounded, and subject to review." Not "should be reasonable" — defined, bounded, reviewed.

Layer 6 bounds the change. §8.5.3 is the load-bearing clause: emergency changes "MAY be permitted only where explicitly defined, MUST be time-bounded, MUST NOT override Layer 0 invariants, and MUST undergo mandatory post-hoc review and ratification or rollback." That final phrase is the return path, written as a requirement. There are exactly two legitimate destinations for an emergency measure: it is ratified through the normal process as a real, versioned change, or it is rolled back. Continuing quietly is not on the menu. §8.6.3 then requires the Change Protocol itself to spell out emergency provisions "including strict time bounds and mandatory review," so the path exists on paper before it is ever needed.

Layer 0 puts some things out of reach entirely. The Invariants Register is a mandatory artifact, and §2.3.4 is unambiguous: "No decision, role, process, or emergency measure MAY override an invariant." This is the floor beneath the floor. Whatever else bends in a crisis, the things you named as non-negotiable — the right to exit, due process, the commons staying common — do not become available just because the situation is urgent.

Underneath all three sits one line from the RCOS core principles, which is really the whole article compressed: urgency is never treated as justification.

An honest gap in v0.1

We publish what works and what doesn't, so here is a tension we noticed while writing this.

Layer 2's explicitness rules list emergency procedures under MAY be explicit, and Layer 6 lists sunset clauses under MAY be explicit — recommended, not required. Read strictly, the spec is coherent: RCOS doesn't force a community to grant itself emergency powers at all, but the moment such authority exists, §4.3.5 and §8.5.3 bind it tightly.

The practical problem is that a community with no predefined emergency procedure will not calmly decline to act when the pipe bursts. It will improvise — and §4.3.5 then gets applied retroactively to authority nobody ever scoped. The stress test's own advice is to predefine emergency powers now, which sits a shelf higher than where the spec currently files it. Our read: v0.2 should promote "the Change Protocol must state whether emergency powers exist, and if so their scope and sunset" from MAY to MUST. The answer is allowed to be "none" — but the question shouldn't be optional. Disagree? That is exactly the argument this standard is built in the open to have.

What to do in the next month

The remediation steps are small enough to run at one meeting.

List every emergency measure still in effect. Go back over the last two years and write down what changed and never changed back. Expect the list to be longer than anyone predicts, and expect at least one item everyone had forgotten was ever provisional.

For each one: let it expire, or pass it through the normal process as a real, versioned change. Both outcomes are fine. Drift is what isn't. A measure that survives an honest vote has earned its place; one that merely persisted has not.

Predefine your emergency powers now, while nothing is on fire. Who can act alone, over what scope, for how long, and what automatically happens at the end date. The stress test frames the alternative sharply: a bounded, logged path, or a blank cheque.

And then the step that separates L2 from L3 — rehearse it. Run a tabletop scenario on a quiet Tuesday. Invoke the emergency power on a fictional crisis, log it, hit the sunset date, hold the review. A return path that has never been walked is a hypothesis, not a mechanism.

The principle underneath

A community that can act decisively in a crisis is strong. A community that can act decisively and then reliably return to its own rules is something rarer, and it is the actual definition of resilience — stress produces learning instead of drift.

The point of bounding emergency powers is not suspicion of the person who acts. It is the opposite. It is what allows someone to make the hard call at 2am without that call silently becoming a constitutional amendment — and without them carrying, forever, the quiet weight of having changed something nobody agreed to.

Write the way back before you need to go. Then the acting is free.

This is a field note from inside the work. RCOS is a living standard, tested in real conditions and revised in public.

Run this stress test on your own community. Read the full Emergency Rule Bypass Precedent test, or tell us what your group is still doing "just for now" — and whether anyone has noticed. The standard gets better when the field talks back. You can also run a free 5-min Resilience Self-Assessment to check how resilient your community is in other regards.