This document states a general advocacy method, beyond any single instance of it. That is a large claim, and the claim carries its own limit, stated in §8: history supplies converged instances of the combination; what remains unproven is whether design can reproduce what constraint discovered. The method has no proper name; it is called here, for lack of one, a trimtab method — see §9. This document was developed with assistance from Claude (Anthropic PBC). All strategic decisions and commitments are those of the author.

1. The diagnosis — what existing methods share

A survey of the advocacy field (rebuttal and fact-checking, framing and narrative change, deep canvassing, solutions journalism, social-norms messaging, moral reframing, hope-based communications, entertainment-education, crisis-mapping, twinning and sponsorship) finds four properties shared across nearly all of it:

• Episodic. The unit of work is the burst — one conversation, one article, one exposure, one campaign. Few methods use time itself as the evidence; the ones that did (§8) were not designed to, and are the exception the record supplies.
• Extractive. A story is taken from a subject and carried to an audience. The subject is material; ethics arrives afterward as compliance, not as architecture.
• Persuasive. Even the gentlest modes lead with a vision they want adopted, or aim at the audience’s values. The ask is present, however softly it is held.
• Unfalsifiable. Almost no method states in advance what its own failure looks like. Deep canvassing is the honest exception — it measured, and published its nulls.

Each property is a reasonable adaptation to campaign economics. Together they produce the landscape audiences have learned to tune out.

2. The method — four commitments

Duration as evidence. The unit of proof is the return, not the claim. The method publishes only what can be revisited, and schedules the revisit into the form itself. The same subject, seen again, further grown — duration is the one thing hostile propaganda cannot counterfeit, because a lie must be maintained and a garden only watered. Where other methods argue, this one waits in public.

Membership, not extraction. The person shown owns the showing. Consent names the reach and the permanence edge before anything publishes; belonging attaches to the contribution, and withdrawal remains possible along the whole trace gradient. This is structure, not compliance — the shown person is a member of the thing, never its content.

Demonstration, not persuasion. No narrator, no ask, no concluding caption. The audience completes the meaning or nothing happens. On-ramps exist — follow, twin, fund, visit — but every on-ramp is a door, never a hand on the back. The only conversion that counts is the step the viewer takes as their own.

Falsifiability in the room. The method states its predicted results in advance, including expected near-nulls on single exposure; it carries a hostile-read round that can remove whole content classes before they are built; it names its own failure conditions (§7). A method that cannot say what would disprove it is a belief.

The four commitments are load-bearing together. Any three without the fourth collapse into something the field already has: duration without membership is surveillance; membership without demonstration is a co-op newsletter; demonstration without falsifiability is faith.

Three design rules the commitments imply but do not state. They come from a design model beneath the method — tuning rather than transmission, set out with its transfer ledger in the advocacy reference’s §7 and observed at length in the companion case history on resonance — and they are stated here in the method’s own terms, without that model’s vocabulary:

1. Choose selectivity, and say which. Depth of response and breadth of audience trade off; a narrow, well-defined audience with a deep accumulating response is a legitimate and usually superior choice to broad reach with fast decay. Carriers before crowds is the narrow choice. The choice is named per instance, because it also chooses what is measured.

2. One drop point, in sequence. Inputs out of phase with each other cancel, so the entry point is one and it is chosen: one point, one rhythm, proven or falsified, then the next. Criteria: a rhythm already running; a carrier whose own audience follows them; low damping — the moveable middle, not a band already saturated to futility; a medium that structurally supports revisit; and survivable placement, since a drop point that exposes anyone is disqualified whatever its coupling.

3. Feedback must be able to change the work. What comes back from the audience must be able to alter form, subjects, and schedule — or the method is transmission with better manners. Working with, not for, is the two-way condition stated as design.

3. The root

The method is Goethean observation turned outward.

Delicate empiricism — observing a living thing repeatedly over time, without imposed theory, until the thing shows itself — is a private scientific discipline. This method performs the same discipline in public, with an audience watching the repetition. The revisit pulse is Goethe’s repeated observation; the no-narrator rule is his refusal of imposed theory; the audience completing the meaning is the phenomenon showing itself to the observer rather than being explained at them.

This matters for honesty of lineage: the method is not an assembly of borrowed parts from the field survey. It is one older method, redirected — and the borrowed parts (§4) confirm it from other directions rather than compose it.

4. Lineage — confirmations, not components

Each commitment exists somewhere in the field alone; none exist together, and none carry the falsifiability requirement:

• Duration: sponsorship and twinning models, where the relationship outlasts every campaign around it, and where participation itself converts attitudes over years.
• Membership: participatory action research, where subjects are co-researchers; the toloka, where belonging follows the contribution.
• Demonstration: Fuller — build the new model that makes the old obsolete; prefigurative politics generally.
• Falsifiability: deep canvassing’s measurement culture — the only advocacy literature that published its own failures and got stronger for it.
• Efficacy mechanics: solutions journalism’s finding that demonstrated agency lowers the cost of looking and reduces avoidance — the theory of change, with evidence.

The combination is the contribution. The parts are not claimed as new.

5. Design constraints that follow

Stated generally; each travels with the method into any instance:

• Capture design. Every output is built to survive hostile re-use; when in doubt, it stays unpublished. Reach multiplies exposure, so reach is a design input, not a distribution afterthought.
• The vulnerable are not material. Classes of subjects whose exposure cannot be made safe are excluded by design, and no gate opens the exclusion. In wartime this includes children, precise places, and schedules tied to named people.
• Proportionate traces. No act of participation carries a trace more legible or more durable than the act itself requires; every step up the gradient is the participant’s own named step. Cryptography may touch artifacts, never people.
• No fabricated subjects. No composites, no invented detail, no placeholder presented as real; scaffolding is always flagged as scaffolding. An empty region is the correct answer until real consented material exists.

6. The instantiation test

Something is an instance of the method only if all five hold:

1. A real, revisitable subject — something that will still be there, changed, when the observer returns.
2. An owner-member, not a source — the shown person holds the showing.
3. A form with no narrator — the audience completes the meaning.
4. A stated failure condition — the instance says in advance what would show it is not working.
5. A named blocking belief — the audience’s actual obstacle, diagnosed, not assumed.

A worked example — a living map of places where something is growing: the point is the revisitable subject; membership follows the point; the sixty-second no-narrator story is the form; a room exercise carries the predicted near-null and the hostile-read round; futility — not opposition — is the named blocking belief. Five for five, which is what would make such a map a designed instance rather than an application.

7. What the method declines

Declined by name, with reasons — not merely omitted:

• The rebuttal mode. Correction repeats the frame it corrects; the correction literature backs the refusal.
• The manifesto mode. Declared identity invites contest; demonstrated existence does not argue.
• Fictional carriers. Entertainment-education works, and is refused anyway: the method’s authority rests entirely on nothing being invented. Serialisation transfers; fiction does not.
• The urgency register. Urgency is episodic by nature and spends trust the durational form needs. The method loses every news cycle by design and accepts this.
• Aggregate spectacle. As instances multiply, the single subject must remain the unit of encounter — compassion collapses at scale, and a map that becomes a statistic has left the method.

8. Honest limits

• First designed instance, not first instance. The durational-membered-demonstrative combination emerged independently on at least four continents — the Stolpersteine, the Green Belt Movement, the Madres de Plaza de Mayo, the Treatment Action Campaign, the AIDS Memorial Quilt — almost always in movements too poor to buy reach. That convergent evolution is stronger evidence of generality than any single designed instance could be. What remains a hypothesis is narrower, and stated plainly: whether deliberate design can reproduce what constraint discovered, and whether the fourth commitment — falsifiability, which none of the converged cases carried by intent — adds what this method claims it adds. A first deliberately designed instance is the test of both. And the convergence is among winners, not a guarantee: Esperanto ran more than a century on three commitments and never tipped. The full record is held in the advocacy methods reference published alongside this document.
• Attribution is unsolved. The method cannot prove the crack it causes — a weakness shared with the entire field, named rather than hidden. The falsifiability commitment bounds the claim; it does not close the gap.
• Slow by design. Durational evidence cannot answer an episodic emergency. The method is wrong for crisis response and does not pretend otherwise.
• Scale is bounded by consent. Membership architecture does not compress. Growth is limited by the pace of real consent conversations, and any shortcut breaks the second commitment. This limit is accepted as a feature — it is also what makes the trust legible.

9. Open questions

• The name. Unnamed by design; “trimtab method” is a placeholder that names the mechanism, not the thing. The lived concepts of mutual work — toloka, dugnad, meitheal, harambee — suggest the name may already exist somewhere rather than needing coinage.

• A second designed instance. Convergence supplies historical instances; deliberate design has yet to prove one. Which domain tests reproduction first belongs to whoever picks the method up.

• The measurement. The instruments are stated in the advocacy reference, §7; no instance has yet run them from a T-zero baseline to a published null. Until one has, the fourth commitment is a design, not a result.

Provenance

Derived from whatif_method_general_EN_v0_4.md (v0.4) by tools/build_oer_edition.py on 2026-09-05. Removed for this edition: the first designed instance (a specific map project) and the decisions reserved to its team; the source’s changelog and drafting notes. The method, the record, every figure, and every source are unchanged. This edition is licensed CC BY-SA 4.0 and is written to be taken, changed, and outgrown.

This document was developed with assistance from Claude (Anthropic PBC). All strategic decisions and commitments are those of the author.