Reference document — the working specification the v2 site is built from.

Lens v2 — Specification

Working spec. Reconciles Rolling Lens*,* Tables Reworked*, the Tell ruling, the 5×5 matrix and the* Lens Table-Creation Rubric into one contract. Where the drafts disagreed, this document rules; where this document and the Rubric differ on table authoring, the Rubric rules (§B0). All decisions are ruled; the Pressure Clock ruling and its timer proviso are in §B9.

Part A — The player’s face (plain language)

A1. The kit

This book, something to write on, and five dice:

• two d6, different colours — the Yes/No oracle only (one answers, one carries the Twist)
• one d8 — the Pressure die
• one d10 and one d12 — the Result die; you use one or the other, never both

A2. Choosing a table

Ask two questions. The column is what kind of answer you need. The row is what you’re dealing with.

Don’t choose by skill name — the same skill can lead to different tables depending on what you’re using it for. Choose by what your character is trying to do. Each table opens with a one-line “use this when” and a “not this — use that” so you can confirm in a glance.

A yes/no fact (Is the door trapped? Is she lying?) is not a table question. Use your game’s rules or the Yes/No dice. And don’t roll a table just for flavour — its answer must change what you understand, the pressure on you, or your next real choice.

A3. Rolling

1. Pick the Result die. d10 when the answer must stay inside the world’s rules. d12 only when this answer may break them — results 11–12 are the uncanny. Magic, monsters and high tech don’t need a d12 by themselves; only an answer beyond what the world regards as possible does.
2. Roll the Result die and the d8 together. One roll per question. A result may raise a new question and point to another table; it never forces another roll.
3. Read it one of two ways (A4).
4. State the Tell. One concrete sign the character experiences.
5. State the Impression. What the character takes it to mean, in one plain sentence.
6. Record the Resistance — the tag and number printed on the result — and play on. Your game resolves any action that follows.

A4. The two ways to read a roll

Read it forwards when the character has already earned the answer — a check has succeeded, or no check was needed.

Read the Result → read the Complication (left side of the d8) → state the Tell → state the Impression. Result + Complication = the GM’s answer.

The Complication belongs to the thing being perceived. It’s the second fact that makes the answer not simple: ORDERED + CONCEALED — there’s a clear chain of command, and the part that matters is hidden. There is no Impulse, Follow or Resist in a forwards reading. You legitimately know the new information, you record the Resistance, and you choose freely what to do next.

Read it backwards when you know something the character doesn’t, and could use it to pick the best move. It needs a real point of contact — a door being opened, a room entered, a person approached. Don’t roll at a secret the character can’t touch.

Read the Approach (right side of the d8) → turn it into one immediate action → Follow or Resist → resolve the action with your game → then read the waiting Result. Approach + your game’s outcome + Result = what the character experiences.

The Approach belongs to the character. It doesn’t reveal the secret — it decides what the character does, which is what gives your game an honest reason to make its check. Do not read the paired Complication. The waiting Result is then read through what that action allowed the character to notice: a success gives useful understanding in time; a failure gives a partial, mistaken or late one; no check means it only shows through what happens next. A failed check must never hand over what a success would have.

Never read both faces of the d8 in one roll. The Complication is the object’s side and complicates the answer generatively; the Approach is the character’s side and controls what they do before your knowledge can intrude. Which one you read is decided by whether the character has earned the answer (forwards) or you need the firewall (backwards).

The one worked backwards reading: the scenario tells you the maintenance crew moved a shipment during a forty-minute gap; Novak, questioning a freight clerk, doesn’t know. Roll — Result 5 OMISSION, d8 7. Read the Approach first: METHODICAL → take the clerk through the night one log entry at a time, and stop wherever the account skips. Follow. That gives your game an honest reason for its check; it succeeds. Now read the waiting Result: the clerk describes the unloading at 23:40 and picks up again at 00:20, and asked about the forty minutes says Maintenance had the bay but never gives a name. OMISSION adds only the gap and its edges — not what the player already knows.

A5. Follow or Resist (backwards readings only)

Every result carries a Resistance: a Conflict Tag and a number (Control +1, Threat +2). In a forwards reading you record it as printed and choose freely — Follow and Resist don’t apply.

They exist only in a backwards reading, where the Approach decides the character’s action before the Result is known.

Follow the Approach — make it the character’s next real action — and, once the Result is revealed, record its Resistance as printed.

Resist — choose a different Approach — and, once the Result is revealed, raise its Resistance by one under the same tag. Control +1 becomes Control +2; Threat +2 becomes Threat +3. Resistance never loses its tag.

You never pay to rewrite a Complication — that would put you in the author’s chair. Resisting changes only how the character comes at the moment. It never changes the Result, turns a failure into a success, removes an obstacle, hands over knowledge the character hasn’t earned, or decides how the next action turns out.

A6. The Tell

Turn the Result — and the Complication, in a forwards reading — into one concrete sign the character experiences. The Result and Complication say what’s true; the skill, action, equipment or interface decides how the character meets it. Say what they:

• see · hear · feel · smell or taste
• measure, or read from an instrument
• notice moving, changing, or absent

The Tell is evidence, not explanation. State what’s there before what it means, and keep any unseen cause, intention or motive out of it — that’s the Impression’s job.

Weak: Someone planted the chip. Strong: The chip lies above the dust line, on top of the older papers.

If nothing suggests itself, ask: What changes in the character’s senses, body, equipment or surroundings?

Then state the Impression separately.

A7. Tell, Impression, Belief

Keep three things apart.

• Tell — what the character directly experiences.
• Impression — what they take it to mean, right now.
• Belief — what they carry forward as true. Beliefs live on a ledger on the Pressure Card, rated 1–5: 1 a hunch, 2 reason to think it, 3 a conviction. Only a fact moves the number — a Commit, a host result, the source — never a feeling. When you judge the moment has come, test a Belief with the Yes/No dice at its rating; the world answers, once. A thing you’re deliberately not deciding goes on as a Question, unrated, until play answers it.

You may know more than they do. Write the Belief from the Tell, the Impression and your game’s outcome only. It can be right, incomplete, or wrong — that gap is the game.

Tell: The floor tile answers with a hollow sound. Impression: There may be a space beneath it. Player knowledge: a spiked pit lies below. The character gets only the Tell and the Impression.

A8. Playing a written adventure

The adventure stays in charge of everything it establishes. Lens adds and or but — never instead. It may add how the character encounters a thing, the sign it produces, what makes it complicated, the pressure it creates, and what the character presently believes. It may not erase, relocate or redefine an authored fact. If a result flatly contradicts the page, attach it to another open part of the scene; if that’s impossible, reroll that die for free — it isn’t a Resist.

Order of authority: the adventure says what exists → your game says what succeeds → Lens says how the character meets it → Belief records only what they experienced.

A9. More than one character at the table

When more than one player character is present, the character whose action the roll answers is the active character. Read the Tell and the Impression through them alone. The other characters get only what the active character shares, or what is plainly observable to anyone present.

Roll once for the moment, never once per character — a scene has one Lens roll for one question, whoever is standing in it. If two characters’ responses to the same moment are uncertain and both matter, resolve the first, then ask whether a second question remains.

Part B — The designer’s contract

B0. Authority

The Lens Table-Creation Rubric (kept beside this spec in _v2 Design/, complete text) is the governing standard for creating or revising any of the twenty-five tables. §B2–B5 below summarise it for the contract; where they differ, the Rubric rules. A table is locked only when it passes the Rubric’s §12 stress-tests and its Final acceptance checklist.

The test. Every Lens procedure must return the player to uncertainty rather than hand them authorial control over it. Two checks, applied to any rule before it enters this contract: Does this return the player to uncertainty, or give them authorial control over it? — uncertainty wins. And does it run Ask → roll → interpret → return to the character? — nothing in Lens may become an extended conversation with yourself about what would make the better story. Its job is to introduce what you did not choose, not to recruit you as a scriptwriter.

B1. Terms

“Pattern” (from Tables Reworked*) is retired; the Rubric’s vocabulary is Result + Tell.*

The two renderings are parallel and should be taught as one symmetry: Result → Tell and Approach → Impulse. Abstract word on the die; concrete fiction on the page.

B2. The finished table page (fixed — Rubric §13)

A table page carries only these seven things. Universal rules — the roll, the two readings, Follow/Resist, the Tell rule — live in How to Read a Lens Roll and are not repeated on every page.

1. Name and family — one-word verb; its column.
2. Use statement — “Use [Table] when the character…”. No example skills: the character’s purpose selects the table, not the skill.
3. Nearest boundaries — the line that resolves the likeliest confusion (“Use Examine for one known physical subject. Use Search when that subject has not yet been found.”).
4. Open question — one question per table, introduced by “Ask:”, that serves both cases (after a successful host check, and when no check is needed). Open, not binary; names the table’s proper focus; never asks for a complete history, secret plan or exact proper noun. (Ruled set-wide: the check/no-check distinction is a universal rule in A3–A4, not a per-page repeat.)
5. Result d10/d12 — d10/d12 | Result | What it reveals | Resistance. Twelve rows. The column is headed Result on every table.
   • 1–6 grounded, +1 under the appropriate tag.
   • 7–10 grounded, +2 — sharper pressure, danger, urgency, exposure or loss of control. Higher is not failure; it is useful development under greater pressure.
   • 11–12 uncanny, Uncanny +2 — only on the d12; must break an understood rule of space, time, cause, identity, awareness, behaviour or physical possibility. Merely magical, futuristic, alien or monstrous within its setting is not uncanny. Word each so it forecloses a mundane reading — state the impossibility in the row (no possible means, before it could have existed, not yet made).
   • Tags need no even distribution, but repeated tags must reflect the table’s subject, not convenience.
6. Tailored Pressure d8 — d8 | Complication | Approach. Names and order are fixed across all twenty-five; only the descriptions are rewritten for the table’s activity: 1 DELICATE / CAREFUL · 2 VOLATILE / RECKLESS · 3 CONNECTED / STRATEGIC · 4 FLEETING / URGENT · 5 CONCEALED / SECRETIVE · 6 OVERT / DIRECT · 7 LAYERED / METHODICAL · 8 EXPLOITABLE / OPPORTUNISTIC. Every Result must work with every d8 entry. An Approach must yield a materially different action when Resisted — if Follow and Resist would behave the same, the Approach is too weak. Never reduce an Impulse to noticing, remembering, recording or thinking.
7. One short worked example — a forwards reading, quoting a real row, in this fixed shape: the host check (if required) → Result → Complication → Tell → Impression → Resistance recorded → the player then chooses what to do. The rolled lines quote the row number (Result 2: ACCESS, Complication 7: LAYERED) and label the value Resistance:, never “Tag”. It shows no Impulse, Follow or Resist, and prints no host skill name or setting-owned proper noun. It demonstrates the table’s distinctive job, not the whole procedure. Because it always shows a concrete sensory Tell, it is the per-page cue for the Tell rule without restating it. The single generic backwards-reading example lives in How to Read a Lens Roll (A4), not on table pages.

B3. Table contract (write before the rows)

Name · Description (the circumstances) · Prompt (the open question) · Boundary (nearest alternatives and when to use them) · Result promise (what every row must provide) · Impulse promise (what followable behaviour it may produce).

Locate the table as an intersection and say so: Examine is Perceive applied to a Focus. Consult is Discover applied to a Will. Invoke is Act applied to Potential. If it could sit in another cell just as easily, its purpose isn’t clear yet.

The name must let a player complete: “My character is trying to ___.” One familiar base-form verb, distinct from every neighbour.

B4. The matrix (designer terms)

Columns answer what kind of GM response is needed: Perceive · Discover · Relate · Act · Endure. Rows answer what shape the engagement has: Field · Will · Focus · Potential · Process.

The same grid as A2, in designer terms:

Each column answers a distinct kind of question:

• Perceive — What does the character notice?
• Discover — What does deliberate inquiry uncover?
• Relate — What develops between intelligences?
• Act — What shape does committed action take under resistance?
• Endure — What does pressure, passage or consequence do to the character?

Each row names the shape of the engagement:

• Field — a surrounding situation, environment, community or system.
• Will — an intelligence, intention, belief or desire (including crowds, AIs, formations, gestalts).
• Focus — one bounded subject, opponent, problem or consequence. Need not be an object.
• Potential — something concealed, undeveloped, inactive or lost, brought into play. Each table makes the unavailable available.
• Process — something unfolding across time, distance or a sequence. A developing course, not merely “takes a while”.

The plain-language labels in A2 are the player-facing rendering of exactly these; the two grids are one object in two vocabularies.

The Endure inversion (rule). In Perceive, Discover, Relate and Act the character acts on the world. In Endure the row-shape acts upon the character — which is why Tempt is the character’s own will where Read, Consult, Negotiate and Deceive are another’s.

Numbering. L1–L25, column-major. L-number mod 5 gives the row (L1 / L6 / L11 / L16 / L21 are all Field); the block of five gives the column. This is the live order; the earlier list in Tables Reworked is retired.

Naming constraints.

• Navigate = social/structural terrain only; physical passage is Travel. The one naming risk in the grid — its boundary line leads the table.
• Bond develops a relationship; reading its present state is Read.
• Perform is an unfolding act before an audience (song, speech, command, rite), not making an object.
• Invoke brings a dormant capability into active effect.
• Reckon makes one particular consequence immediate, not general fallout.

The grid as diagnostic. Every unanswered question lands in exactly one cell. A question that doesn’t fit identifies an overloaded cell — it does not justify a 26th table. (Crafting → Examine / Reckon / Invoke, or a plain host roll.)

Row and column labels are an index aid, never a procedure. The player-facing questions live in the reference index (A2). The roll instruction remains “choose the table matching the question.”

B5. Authoring limits

Mature fiction, simple interface. A table passes only if a player can choose it from its name and one sentence, understand the result on first read, turn it into one concrete detail without a word puzzle, see what the character is pulled to do, see how resisting differs, and keep playing without an explanation. If an example is needed to rescue a result, the result isn’t clear enough.

Form. One-word name · one-sentence description · one short question · common words · no design terminology in rows · no proper nouns or assumed setting · Impression ≤ two short sentences · Impulse one immediate action, clear verb first, about three to eight words · at most one interpretive step from result to fiction. Free indirect discourse is a technique, not a term the player meets.

Every Result must: state one main truth · use familiar words and a short sentence · stay recognisable across different examples · produce a concrete Tell · create an actionable difference · stay distinct from every other Result in the table · work with every d8 entry.

Avoid: vague themes (mystery, importance, danger) without saying what creates them · proper nouns or setting-specific content · hidden motives outside a table whose purpose is reading intent · deciding whether a host action succeeds · dictating damage, conditions or other host mechanics · solving the character’s next action · restating one result at several severities. A Result may establish evidence of agency — a cut, alteration, arrangement, signal or mark — when it is physically apparent; it may not establish why an unseen actor did it. Treat wanted, intended, in order to, meant for with suspicion: motive belongs in the Impression or a later Yes/No question. No Result may restate a d8 entry (CONNECTED, CONCEALED, OVERT, LAYERED, DELICATE, VOLATILE, FLEETING, EXPLOITABLE) — the Result is the primary finding; the d8 carries it beyond its subject.

Lock gate (Rubric §12 and the acceptance checklist). A table is locked only after it passes: the combination test (every Result × all eight Complications — 96 pairs — each coherent; none contradictory, repetitive, unrelated, laboured or inert); the approach test (all eight Approaches on at least three situations — each concrete, usable before hidden information, resistible, and leaving success to the host); the genre test (mundane/historical, fantasy, science fiction, horror); the subject test (person, coordinated group, creature, place, object, responsive system, gestalt); the system test (several host skill vocabularies — tested in design, never printed); the scenario test (generative play, a read scenario, a known trap, an answer of “No”, no host check); and the sixteen-year-old test. Then the Final acceptance checklist.

B6. Integration rules

• Host compulsions come first. Where the host system already dictates behaviour (Pendragon Traits and Passions), Lens does not roll to contradict it. Lens answers only where a GM-facing question remains.
• Uncanny is relative to the setting. A familiar spell or Discipline is not a d12 event. d12 opens only when the answer may exceed what that world treats as the established working of magic, science or power.
• No Lens roll to recover a failed host check.
• A coordinated group is one subject while it acts with a shared disposition or purpose. If its members stop acting together, narrow the focus or use a Result such as DIVIDED.
• Reputation records how the active character is regarded: their standing within a group, or the trust, respect, recognition, legitimacy, or suspicion directed toward them by another intelligence. It is never a third party’s standing. (Supersedes R8’s one-line gloss; validates Navigate, Confront, Bond, Perform and Negotiate; re-tags Canvass DIVIDED → Pressure +1, Consult CONTESTED → Clue +2, Question ADMISSION → Exposure +2.)
• Multiple active characters — ruled in A9 (the active character; one roll per moment). It lives in How to Read a Lens Roll, never on a table page.

B7. Retired and renamed

B8. Blast radius (what changes in the shipped set if v2 is adopted)

• Ch1 — the kit line; the “Impulse is your character’s gut, made of dice” metaphor now describes one of two readings.
• Ch3 / Quick Start / R9 — Ignition is three lines; Hook and Opening Beat become blank-page tools; the Aim and Stake sit on World Pressure (B9).
• Ch4 / R4–R7 — the Belief ledger (B10): Facts and Beliefs as separate objects, ratings, Questions, the test; the Commit no longer writes a Belief.
• Ch3 — the first roll (First Sign, Impulse, the d6).
• Ch4 — Beats / Resist cost / how Resistance feeds the clock (B9).
• Ch5 — the L11 First Sign citation.
• Ch7 — the tables teaching, tag order, toggle.
• Reference Layer — R-sections on the roll, First Sign, the loop; the R10 stems; R8’s Reputation gloss widens to the B6 definition.
• Glossary — add Result (as term), Complication, Approach, Resistance, forwards/backwards; remove First Sign and Pattern; redefine Tell, Impulse.
• Quick Start — kit, loop.
• All six Palettes — Translation Guides are mapped row-by-row to the old 25 names: full remap, not an edit.
• Part II — the 25 old table files are replaced.
• The five fixed tables (L1–L5) — one header fix: Assess, Read and Examine head their result column Tell; per the Rubric it must be Result (as Search and Watch already have it).

Keep v2 in _v2 Design/. Nothing here touches the canonical set.

B9. RULED — the Pressure Clock

Resistance is the Beat value. It feeds the Pressure Clock exactly as in v1. The clock, the Dominant Tag, the Commit Procedure and World Pressure are unchanged.

Proviso — World Pressure is the timer. In play, World Pressure’s job is the end point: the confrontation timer. Every local Commit advances it by one, carrying that clock’s Dominant Tag up; when it fills, the crisis arrives whether the character is ready or not. Playtesting confirmed this is what gives a session its shape and its deadline. v2 changes nothing here — keep it explicit in the teaching that the clock’s reach (local → World) is the thing that makes the session finite.

RULED (Sept 2026) — World Pressure carries the Aim and the Stake. Ignition is three lines with a table behind each: Aim (source, or Stakes Prompt Need), Stake (the thing on the sheet the case can cost — Loss d6 points at where to look; this absorbs the old “detail with bite”), Stage (first lead + what presses, as a question — Lead d6, Stage-constraint d12). Lines are written bare so World Pressure reads as one sentence. The Aim and the Stake are written on World Pressure; the Stage is the first local clock, and is never the Aim as a question (first playtest wrote the case twice). When World Pressure fills, the Stake comes due in the shape of its Dominant Tag, and every open Belief is tested (B10). Hook and Opening Beat become blank-page tools (R11). Friction is not a line — it is the without / before / while that turns the Stage into a question.

Two rulings added (Sept 2026). A serious host failure left without a consequence is a pressured moment: roll the table for what the character was doing and read it through the failure — Reckon if they were looking or asking; the tag is the price, never the withheld answer (How to Read, When to roll). And a tag that reaches +3 on a clock is named in the world on the Card — the thing that will Commit, the thing you would Target (R5, R6).

Resist cost stands as ruled in A5 — +1 under the same tag, never untagged, and only in a backwards reading. This sharpens the Dominant Tag where v1’s untagged Beat diluted it. Accepted as the intended feel.

B10. RULED — The Belief Ledger

Ruled 4 Sept 2026. Deltas applied to R4, R5, R7, R11, R13, R14, the Glossary and How to Read a Lens Roll. Replaces the Fuse draft. Nothing here changes the roll, the clock, the Dominant Tag, the Commit, or World Pressure. B9 stands. It changes one thing: the Belief stops being a line and becomes a ledger.

The premise. The Card holds two kinds of thing, and they are separate objects. Facts are the world’s — Commits landing automatically when a clock fills, host results, what the source reveals. Beliefs are the character’s — what they take to be true, from any source, written only from what they felt, saw, or rolled. Neither is written about the other. v2 gives the second kind a rating and a life. Any number of Beliefs, each rated 1–5 — and any number of Questions, unrated. They go up and down as the world speaks to them. When the player judges the moment has come, a Belief is tested and the world answers. Pressure comes at the character automatically, through the Commit. Belief comes at the world when the character is ready, through the test. Blade Runner’s hypothesis management, on Lens’s dice.

What a Belief is

One claim the character holds that the world has not confirmed — a theory (the animoid ties to Vice’s case) or a prediction (if Vice hears of this, they pull my case). Written in the character’s voice, on the Card, with a rating. It must pass the Belief test as now: could the character have known this from what they experienced? The rating is how far the record on the Card supports it — Commits, Results, host outcomes — never how far the player’s knowledge does.

The rating is the Yes/No likelihood, 1–5. The Yes/No has no 6: a certainty isn’t asked, and a 6 on the answer die never says yes — one rule for the whole system. A Belief is by definition unconfirmed, so the ceiling is 5 — 5-in-6, the same ceiling as Blade Runner’s D12/D12. No new dice.

What a Question is

A line the character has not leaned on. Is the sire lying? — no rating, because there is no inference yet, only a thing deliberately left unresolved. Write it on the ledger with a question mark and no number. It becomes a Belief the moment the character leans — the sire is lying · 1 — and until then it is answered only by play: a Result, a Commit, the test of some other Belief. Never off-screen, never in your head. This is how a solo player keeps a secret from themselves — not by pretending not to know, but by declining to decide.

Four states of knowledge, one card — observed is the Tell, inferred is a Belief, unresolved is a Question, confirmed is a Fact: a Commit, a host result, the source, or a tested Belief. Facts and Beliefs are separate objects. A Commit is what a filled clock does to the world. A Belief is what the character thinks, from any source — a host roll, a Lens roll, a look across a room. The Commit Procedure does not write a Belief; it produces a Fact, which may move one.

Writing one

Whenever the character forms one. Usually from an Impression that matters — the Impression is already the candidate held felt-not-fixed until a confirmer agrees, and the ledger is where the candidates live — or from a host result, a Commit, the fiction itself. Written from what the character felt, saw, or rolled, never from what the player knows. Start it at 1 for a hunch, 2 when there is reason to think it, 3 for a conviction. Your call, and it is the only judgement in the mechanic that isn’t the world’s.

Not every Impression becomes a Belief. Write the ones with narrative backbone — the ones the character would act on, or fear.

The three grades, in one case. A hunch: the steward looked away when the seal was mentioned — the steward knows something · 1. Reason to think it: two witnesses have since put the steward at the gate that night — · 2. A conviction: your detective is at a desk with a piece of black-market tech, and the host’s lead table has just made a fact of it — the device is tied to an arms smuggler and a case another agency is running. The fact is the host’s — confirmed, no clock filled, not a Commit — and it goes on the Card as a Fact. The Belief you build on it — if this goes in my report, they take the case off me — is the character’s, and it starts at 3, because it rests on something confirmed, not on a look.

Moving a rating

Only a Fact moves it — a Commit, a host result, the source, a Result’s fact, a tested Belief. One step per Fact, up if it supports the line, down if it contradicts it. There is no review schedule and no Shift end: the ledger is event-driven, and the event is always the world speaking.

Two rules keep it honest:

• A Fact that bears on a Belief moves it. Up if the world just supported it, down if it contradicted it. You may leave a Belief alone — but not when the world has just spoken to it.
• Nothing else moves it. Not an Impression, not a Tell, not a scene ending, not a change of heart. Confidence is not evidence.

A Belief that falls below 1 is struck. A Belief the story has stopped feeding can be struck at any time. So can a Question.

Test — when you’re ready

The player calls it: a confrontation, a key piece of evidence verified, a sudden event or realisation in which the truth is laid bare. The test comes when the story, the investigation, and the rating have converged. Does it feel inevitable? Then test.

Yes/No dice at the Belief’s rating. Twist die alongside. The test cannot be repeated. Nothing on the ledger is ever certain — at 5 it is still one in six to be wrong, and that is the point.

Who answers. The Yes/No dice, always. A test asks whether something is true about the world, and that is the Yes/No’s job by definition; the host RPG answers whether the character’s action succeeds and has no roll for whether the steward is the informant. That is why the rating is the Yes/No likelihood — no new procedure, and nothing that depends on the host. The host keeps two jobs around a test: it gets you to the moment (the confrontation, the search, the interrogation — if that action fails, the moment hasn’t come and there is no test), and it resolves whatever the answer starts (the arrest, the fight, the flight). One override: if the source already answers — the module names the informant — there is no roll. The source is the confirmer.

Read the throw in this order: answer die against the rating — yes or no; doubles — decisive; twist die on a 1 or a 6 — and or but. Three readings, the same two dice.

The corners compose: 1-1 is yes-and-decisive — true, more than you knew, and it closes. 6-6 is no-but-decisive — wrong, but a Lead, and the Lead is whoever led you. Doubles is read on Belief tests only; on a plain fact question the Yes/No stays a two-second tool.

Either way the Belief is resolved. It comes off the ledger and its answer joins the Facts — wrong is a fact too: the claim is false, and the world has said so. It does not go back to 1 to be built again; the test is never repeated. If the character still believes something afterwards — they just haven’t noticed yet — that is a new Belief, at 1, from the wreckage.

Convincing others. When the character argues a Belief to someone — selling the theory to a superior, asking for help on the strength of it — test it. Confirmed: advantage on the host roll. Wrong: disadvantage. A test made to convince does not resolve the Belief; it stays on the ledger at its rating.

The ceiling, and the one forced test

At 5 it stays at 5. Further support changes nothing but the character’s certainty. The world answers when you ask it.

• When World Pressure fills, every open Belief is tested, in the order they were written, as part of the confrontation. Nothing is left hanging. The countdown ends with a reckoning of everything the character believed.

Otherwise nothing forces a test. Facts move Beliefs; only the player tests them, and the player calls the moment, always. That preserves the legwork: if everything is riding on a Belief, build it before you test it.

The firewall, restated

A Fact is true whether the character knows it or not. The Belief is the character’s and may be wrong. The ledger makes the gap visible and rated: I’m through at 3, while the Fact beside it says the inspector will realise. The player can read both lines. The character can read one. Every Fact that moves the rating down is the world closing the gap; the test is where it closes on screen.

A false Belief never becomes true by being believed. The rating measures how far the world has supported it; the answer is still the world’s — the source’s in a module, the dice’s in generative play, at odds that never reach certainty. Confidence is not evidence. This is why the ledger stops at 5.

The Card, revised (R7)

What this changes and what it doesn’t

Unchanged: the roll, the two readings, Resistance, Beats, local clocks, Dominant Tag, Targeting, Turn Spark, the Commit Procedure’s shape, World Pressure’s size and reach, the timeline guard.

Worked continuation — the planted lure

At the checkpoint, before the clock fills, Mara’s Impression on TESTED — this one’s paying attention, she wants to hear me say it — has backbone. It goes on the ledger: She was expecting someone · 1.

The clock fills. Dominant Tag Control. Commit: the inspector waves Mara through but holds her papers a beat too long — she’s clearing the courier she expected, not the one she got. A Fact, and it bears on she was expecting someone — the held papers — so it moves: 2. Mara forms a new Belief from the wave-through: I’m through — she bought it · 3. World Pressure ticks.

Two scenes on, at the inn, another clock fills. Commit (Dominant Tag Exposure): a rider in the checkpoint’s colours dismounts in the yard and asks for the stable-master. A Fact that bears on both lines: I’m through contradicted — 2. She was expecting someone supported — 3.

Mara is in the loft with the trapdoor open. The player judges the moment has come — the rider is asking questions ten feet below — and tests I’m through at 2. d6 shows 5: wrong. Twist 6 — but: a Lead. The rider asks for “the courier with the black seal.” Mara hears every word. Resolved: the answer joins the Facts — the checkpoint has circulated the black seal. Off the ledger. New Belief, from what she now knows: They know the seal, not my face · 1.

She was expecting someone stays on the ledger at 3, untested — Mara has no way to test it yet. If the countdown fills first, it is tested in the confrontation with everything else.