VI.0 · The Cognitive Battlespace

Taken here: the decision. Given back: the feeling of having reached it. Reads as: your own opinion.

Part II teaches a working tongue whose privacy value is unproved. This part distinguishes what can be observed from what is inferred, and examines how a decision can be shaped by the information made available. These are problems to investigate, not proof of an adversary behind every difficulty.

A decision can be arranged without a single false item in the briefing. Put the relevant facts in one order, leave the inconvenient comparison outside the frame, and offer a choice whose alternatives have already been narrowed. You may reach the conclusion quite honestly. That is the attraction of the arrangement: it lets you do the persuading yourself. Check the facts, then check what was left out, who chose the alternatives, and what would change the decision. A clean set of exhibits can still describe a carefully cropped room.


VI.1 · What actually reads you

Words are one source of information. Timing, contact, and context may also be available. Their relative importance depends on the system and situation; there is no universal order in which they are read.

An observation might show that two accounts exchanged messages at a particular time. Inferring coordination is another step; attributing a purpose is another again. Metadata can be incomplete, shared, mislabelled, or consistent with several ordinary explanations. Misclassification can still have consequences. Do not present an inference as an unambiguous identification.

Four possible sources of inference, without a universal ranking. Each needs context and an account of what it cannot establish.

Co-presence. Who is repeatedly in the same place, physical or otherwise, as whom. Repeated co-presence may suggest a connection; shared premises, work, or a service queue offer other explanations. It does not establish a plan.

Timing. When traffic happens, and in what bursts. A burst may suggest coordination, routine work, or an interruption. Protecting message content does not necessarily conceal timing. No comparison of ordinary hours with encrypted messages establishes safety.

Volume change. Changes in activity can be compared with an earlier baseline. Missing records, altered schedules, or changed measurement may also explain a difference; neither quiet nor noise identifies its cause.

Position. Where you sit in the graph. An apparent connection between clusters may draw attention, but incomplete records can distort that picture. A connecting role does not establish motive or culpability.

The practical consequence is unromantic. Most of what makes a group legible is produced by ordinary convenience — the group chat that everyone is in, the recurring calendar item, the one person who is in every conversation because they are helpful. None of it was a decision. All of it is shape.

The practical response is to assess the actual information involved, reduce unnecessary collection, and choose access and retention limits deliberately. Ordinary routines, distributed roles, and unfamiliar words do not make a person safe. Do not collect other people’s personal traces simply because this chapter names them.


VI.2 · How a decision is arranged

The formal name is reflexive control, and it was worked out sixty years ago by a mathematical psychologist who noticed that the standard models were wrong about what people optimize. They assumed agents act on the world. Agents act on their model of the world, which includes their model of the adversary’s model, and that regress is the whole opportunity: you do not have to change the world to change what someone does. You have to change what they think it contains.

The catalogued elements are worth naming plainly, because they are recognizable once named and nearly invisible until then.

ElementWhat it does
Distractiona credible threat somewhere you are not, before the real thing
Overloadmore self-contradictory information than you can adjudicate
Paralysisthe illusion of a precise threat to the thing you most value
Exhaustionforcing expenditure on activity that produces nothing
Splittingmaking you act against the interests of your allies
Appeasementlowering vigilance; making the preparation look routine
Intimidationmanufacturing the appearance of unanswerable superiority
Provocationimposing a scenario in which your best option is still bad
Suggestionmaterial that moves you legally, morally, or ideologically
Pressurematerial that discredits, aimed at the people around you

Note what is missing from that list: falsehood. It appears nowhere and is not required anywhere.

Naming a possible technique is a first step. Check the claim and its source, seek contrary evidence, compare plausible explanations, identify the available choices, and locate the actual pressure. A deadline, lost income, or unequal decision right does not disappear when correctly named. Use the Evidence and Decision Record; recognition alone does not establish intent or classify a person as an adversary.


VI.3 · The seven ways a movement is played

Each is given as how it presents, the tell, and the counter. Treat the tells as prompts to investigate, not diagnoses of motive. Test them against evidence and alternative explanations before deciding how to respond.

1 · Overload. A flood of claims, all plausible, several contradictory, arriving faster than the group can check any of them. The tell: the group’s meetings are now mostly about what is true. The counter: a standing rule that unverified claims are held, not debated, and a named person whose job is holding them.

2 · Splitting. The existing fault line — there is always one — is found, named publicly, and treated as the real story. The group is invited to prove it is not divided by expelling someone. The tell: an internal disagreement that has been running quietly for a year is suddenly urgent, and the urgency came from outside. The counter: Part IV rules 6 and 7. Cheap first, in the open, on the thing.

3 · Exhaustion. Every provocation, request, and bad-faith question is answered, because each one individually deserves an answer. The tell: the group’s calendar is full and its plan has not moved in six weeks. The counter: a budget. Decide in advance what fraction of capacity goes to responding, and let the remainder go unanswered on purpose.

4 · Provocation. A scenario is arranged in which the group’s best available option is still damaging. Usually a person, usually at the edge, usually filmed. The tell: a new participant is unusually keen on a specific escalation and unusually uninterested in everything else. The counter: Part I §I.4, held in advance and out loud, so that the answer is already decided when the moment arrives and nobody has to be brave on the spot.

5 · Appeasement. A partial concession, delivered warmly, costing nothing, and timed to arrive when pressure peaks. Vigilance drops. Eighteen months pass. The tell: the concession is procedural — a review, a working group, a commitment to consult — and there is no date attached to anything. The counter: accept nothing without a date and a named owner. Part IV rule 2, applied outward.

6 · Elevation. The most extreme voice in the movement is promoted as its representative — quoted, booked, platformed — until the movement is what that person says. The tell: the person speaking for you was not chosen by you, and is becoming more extreme rather than less. The counter: choose your own, by lot from those willing, and say publicly who it is. An unchosen spokesperson can only be replaced by a chosen one.

7 · Absorption. The competent are hired. The moderate wing is funded. A seat is offered on the body that decides. This need not feel like an attack, and may not be one — it is an opportunity, offered sincerely, by people who mean it. The tell: the movement’s demands have become things the institution was already willing to do. The counter: there is no clean one. Take the seat or refuse it, but write down first what the seat is for and what would tell you it has stopped being for that, and give the note to somebody who is not on the body.

Afterwards: the five ways the outrage is drained

This section concerns attempts to defuse outrage after visible mistreatment. It does not describe every successful action. Part I also discusses withdrawal of cooperation and other mechanisms. Visible harm neither guarantees support nor has to occur for change to be possible; nobody should seek injury to make the argument legible.

Visible disproportion does not convert itself into pressure. It has to survive a management process, and the process has five moves. They are used together, they are used by institutions that have never coordinated with one another, and they are used because they work.

A note on the names in the right-hand column. The five moves on the left are a named model. The five words on the right are not — they are this handbook’s own paraphrase, chosen to alliterate rather than to quote, and three of the five soften what the model actually recommends. Where they diverge in substance rather than wording, the table follows the substance and says so.

The moveHow it arrivesThe counter
Cover-upthe event is not recorded, not released, not confirmedReveal. Assume the record will not be produced and produce it yourself, at the time, from more than one position
Devaluethe people it happened to are described as the sort of people it happens toRedeem. Name them, ordinary details first. A name and a job undo most of what the devaluation was doing
Reinterpreta misunderstanding, an isolated incident, a regrettable necessity, or provokedReframe. Fix the frame in advance and in one sentence, because the second account to arrive is always the one arguing
Official channelsa review, an inquiry, an independent process, a commitment to learnRedirect. Do not enter the channel by choice. Keep organizing publicly around it; treat an invitation to participate as a tactic to answer, not an obligation
Intimidationcharges, costs, employment consequences, and quiet visits to the people around youResist. Decided in advance, collectively, with the terms written down, because it is not decidable on the night

The fourth is the one to understand properly, because it is the one that looks like winning, and it is the row where this table first drafted the softer answer. An inquiry is not nothing and is occasionally a remedy. Its reliable effect, independent of anybody’s intentions, is to convert an urgent public fact into a pending private process with no date attached — and a group that enters one and stands down while it waits has been answered without receiving an answer. The corrected counter is accordingly a refusal rather than a compromise: stay outside the channel, keep acting, and treat its existence as something to route around rather than something to sit inside politely. Item 5 above describes the same move arriving dressed as a concession. It is the same move, and it gets the same answer.

With the condition Part I §I.5 states. How far that holds depends on whether the channel has any force where you are. An inquiry with powers of compulsion, a regulator that has fined this institution before, an ombudsman whose findings have actually changed something — none of those is the same object as a review announced by the people being reviewed, and the counter is not the same either. Establish which one is in front of you before deciding to route around it. The default here is refusal because the announced internal review is the common case, not because entering a channel is always wrong.

The shape is worth stating on its own. All five work by inserting a step between what happened and what people conclude about it. All five counters work by removing one.

FALSIFIER — the fourth move. The other four are hard to test and this one is not. If cases where a movement entered an official review and stood down can be shown to end in remedy at the same rate as cases where it treated the review as a delay and kept acting, then the channel is a remedy, redirect is wrong, and this row should be struck from the table rather than softened.


VI.4 · Nobody has to be doing it

Everything above describes an actor. The uncomfortable part of the present arrangement is that the same effects are now produced without one.

Selection pressure can do the work without a single director. Of the several things a movement says, the versions rewarded with attention can travel furthest, whether or not they express its position accurately. The angriest extract comes back as the official view; the qualifications arrive later, too long to quote. Check whether that is happening here. No coordinated plot is required, but choices about ranking, revenue, and platform design still have makers and beneficiaries. The absence of a secret memo does not make the arrangement ownerless.

A second mechanism now runs underneath the first and it is considerably faster. Variants of a position can be generated and circulated more quickly than any group can answer them, so the correction arrives after the variant has been seen. What circulated then becomes part of the material the next round of generation is built from, and a drift that began as noise is fed back in as evidence of what the position is. Three parts: speed, a bias toward whatever was already dominant, and feedback. None of them needs an author either. It is the same actorless problem with the gap between drift and correction shortened to less than the time it takes to notice.

Vocabulary revision may support learning, cultural expression, or a useful discussion about stale meanings. It has not been shown to outrun modelling or prevent distortion. A public glossary, context, or repeated use can disclose a revised mapping. Keep Part II’s distinction between comprehension, cultural value, and unproved privacy claims.

The measurement that catches it is simple and almost nobody runs it: compare what you meant to say with what actually spread. Take the three positions the movement most wants understood and find where each ranks in what circulated last month. The gap is the drift. It will be larger than expected and it will be in a consistent direction.

One caution, and it is not a small one. What circulated is measured with the same machinery that produced the drift — its metrics, its resolution, its idea of what counts as circulation. The measurement is taken inside the system under test. A movement whose defensible position travelled well in places the machinery does not count will read a drift that is partly an artefact of where it looked.

Ask people with different relationships to the group—including people who left—what they understand its positions to be. This is purposive qualitative feedback, not a representative sample or conclusive confirmation. Record how people were invited, seek contrasting experiences, and compare their accounts with other evidence. Agreement may reflect a shared source; disagreement is a reason to investigate, not a defect to average away.

FALSIFIER — the actorless case. If a movement’s drift can be shown to track the activity of identifiable actors rather than selection pressure — that is, if removing the actors removes the drift — then VI.4 is describing an operation after all, the four instruments in VI.5 are aimed at the wrong thing, and this section should be a subsection of VI.3.


VI.5 · Four instruments

Small, unglamorous, and effective, mostly by making a fast decision slower.

Latency. Nothing decided on the day it arrives. Part III rule 2. A conviction that appeared immediately after new information is not yet yours.

Paraphrase. State the position again without any of its favourite words. Most capture lives in vocabulary, and an argument that will not survive translation into plain language was carried by the phrasing.

Independent calibration. Someone outside who owes the argument nothing, asked what is wrong with it, and thanked when they say. Not a friendly reader. A disinterested one.

Inventory. Write down what you believed, when, and what changed it. Drift is invisible in the moment and obvious in a list.

Three of the four can be run alone. That is the weakness of three of them.

FALSIFIER. If groups running latency, paraphrase, calibration, and inventory can be shown to drift from their stated positions at the same rate as groups running none, the instruments are ceremonial and this part is decoration.

Three filters before response

Use three filters before turning a concern into action. The action filter asks what changed in the world, who can see it, and what would disconfirm the interpretation. The source filter asks whether the material is direct evidence, interpretation, proposal, commitment, adaptation, or fiction. The cost filter asks what response would spend: attention, trust, labour, legal exposure, money, or a person’s name.

Apply the data rule with particular meanness. State the purpose before collecting information. Collect the least needed to decide the next lawful step. Set a deletion date. Do not make a private archive of other people’s ordinary lives because the archive gives the archivist a small administrative thrill. The repair desk does not need a map of every donor’s politics to decide whether a cupboard key needs a second holder.

At the repair desk, “three complaints arrived after Mira changed the intake form” is evidence of complaints, timing, and a change. “Mira is undermining the desk” is an interpretation. “Restore the old form for one week and compare error rates” is a proposal. “We will run the test on Saturday and publish the result” is a commitment. Keep those names apart. Otherwise the minutes will record an agreed finding where the meeting merely had a feeling about Mira.