Positioning — product identity and boundary

# Positioning — product identity and boundary

> **This is the anchor.** Every other product document agrees with this page or it is wrong.
>
> **Meridians is a creative studio whose subject is fate — the art and science of possibility. It records
> worlds completely enough that their futures can be run forward rather than imagined, searches those
> futures instead of authoring them, and returns the timelines worth having as decision-driven
> experiences — explorable visual novels with many routes and many endings, prepared once and replayed
> for as long as the world lasts.**

## The north stars: training, learning, immersion

The worlds are built so a person can be dropped in cold, read the room, act under consequence, and come
back changed — a spy inserted into an unfamiliar country, made safe, short, and replayable. What we sell
is **training worlds and pocket dimensions**: the **training world** is the insertion, solo, an hour (the
visual novel); the **pocket dimension** is a small kept world a few people step back into as its
characters (the populated World, bounded and with care); agents get the same two.
[Strategy 01](../strategy/01-the-north-star.md) owns the statement; the
[training worlds spec](../../../specs/training-worlds/README.md) owns the unit.

**The theme beneath the name (2026-09-22):** Meridians are the fateful, causal pathways open to a life or a
world, and the product is about learning to navigate them. Public copy says *train the pathway*, not
*simulate it*: the causal simulator is the category, training the pathway is the promise. That one thread
grounds everything else — the tree, MWS, the five forces, the training — so copy should read as a single
through-line rather than a list of features.

**The lean product, and what stays nuance (2026-09-22):** front-loaded, Meridians is immersive visual novels
that demonstrate real decisions under consequence; the causal tree is shown up front so the person knows the
divergences are planned, not generated live. Fictional worlds for personality development, business
training worlds (sales, strategy, marketing, interviews), the psychometric and cohort value of choice data
on a Rasch scale, and a marketplace of shareable VNs are nuance to the business, argued in
[strategy 14 §3a](../strategy/14-the-causal-simulator-monopoly.md#3a-the-lean-product-and-the-nuance-behind-it-2026-09-22),
never the pitch.

**The category is the causal simulator; the visual novel is its interface.** Meridians turns a
consequential situation into a world with memory: you act, the people in it interpret what you did, the
state changes, and the consequence comes back to you — so the question that brings you back is *what did
I cause?*, not *what happens next?* A simulator says "try it" rather than "here is what to do": run the
situation, see the cost, retry as an experiment. Meridians World Search (MWS) exposes the game being played — incentives,
information asymmetry, timing — and never chooses; the reading of a person stays their own; and whether
rehearsal carries into life is a hypothesis proven externally, not a promise. Public copy leads with the
world and the decision, not the format; nobody has to identify as a VN player. Strategy 01 owns the full
argument and the directional pipeline (interview → world → MWS → simulation → reflection).

**Have a problem? Train the pathway before you live it.**

The primary roadmap is **Causal MCTS VNs**; VNMMO is a stretch goal on a separate roadmap.

The intended far horizon is the **space traveller, human or artificial**, but space travel is the framing
story rather than a technical premise Meridians must prove. A long voyage makes the need visible: a vessel
may sustain life and work while offering only a narrow slice of the places, people, choices, and surprise
that came from moving through a whole world. Training worlds supplement that experience through
simulations, visual novels, and stories; a pocket dimension lets a small society and its history travel
with the traveller. A memory-bearing datacenter is the clearest image of both forms joined, not
infrastructure Meridians operates. It could explore where human bodies cannot yet go and carry questions
larger than one mission — whether we are alone, what intelligence may become, and what else existence
contains. Those remain Fate questions, not discoveries or infrastructure claims.

## The value proposition: better books

If one line has to carry the company, it is this: **better books.** Not longer, not faster, not
machine-written — better along the four axes a printed page has to fake, and which a recorded world can
actually keep. Every roadmap argument, spend decision, and spec trade resolves against them:

- **Consequence.** A choice moves typed world state and the world computes what follows, so a second run
  down the other branch is a different story rather than the same pages again. A fork whose branches
  differ only in prose is a version, not a decision — and refusing to offer it is the single discipline
  that separates this from generated choose-your-own-adventure slop.
- **Meaning.** The world is interpretable, and its lessons persist: a theme is a pattern the record keeps
  producing — what this world rewards, exacts, forgives — shown with the events supporting it and the
  counter-example admitted, rather than a moral asserted at the end. Meaning that survives replay is the
  deepest thing the medium can offer.
- **Perspective.** POV genuinely withholds. A seat reads only what it could know, remember, reach, and
  afford at that moment, so a route is a vantage rather than a colour filter, and the same events read
  differently because someone else was standing there.
- **Immersion.** Sprites, plates, artifacts, staging, and the discipline of a scene, prepared before
  reading so nothing stalls on a model call — the medium's craft, not a chat window with a background.

The three mediums are how those four are delivered, not four more features: you **decide** in a Scenario,
**read** an Experience, **wander** a World. Peers over one Domain, and each one is judged by how much
consequence, meaning, perspective, and immersion it actually buys the reader.

## The company

Meridians is a creative company, not a tools vendor that happens to ship stories. Its subject is the one
force every narrative form is actually about and none of them model: **fate** — what could still happen,
how likely it is, and what it costs to find out. The craft side is the art of it (a fork must land, a
revelation must be timed, an ending must be earned); the engine side is the science (recorded variables,
priors, calibration, deltas, a search over possibility with a price attached). Neither half is decoration
on the other.

What is unusual is the premise underneath: **when every variable is recorded, possibility can be
simulated forward rather than asserted.** A Domain types its rules, cast, artifacts, relationships,
stances, and open questions, so a counterfactual is *run* against the same world rather than imagined
beside it. That is what makes two branches genuinely parallel worlds — same rules, same cast, different
history — and what makes the difference between them something you can point at rather than something a
writer claims.

## The forms it works in

The medium is a convergence, not an invention. The **choose-your-own-adventure** contributed the fork and
the reader's authority over it; the **visual novel** contributed the craft — sprites, plates, routes,
endings, the discipline of a scene. Meridians keeps both and adds a world underneath that knows what a
choice costs. The generated world, the card game, and situated roleplay are the same gravity from
different angles:

- **Generatable worlds** — a canon, corpus, or premise becomes a recorded Domain that can be run forward.
- **The card game (Scenario)** — the decision substrate, where a move is priced and resolved.
- **Scenario roleplay (Episode, World)** — inhabiting the moment rather than reading past it.
- **The prepared novel (Experience)** — the timeline that earned production, served without live generation.

All four orbit the same question, which is the company's whole subject: *what could this world still
become, and which of those futures is worth living through?*

## Will is injected from outside the simulation

A recorded world can compute consequence, but it must never supply the commitment. **Will enters from
outside**: a reader at a fork, a group at the card table, a Director steering a Program, or an autonomous
agent acting through MCP as a typed, attributable action. The engine holds the rules, the memory, and the
price of a move; the move itself is injected.

This is the difference between a simulation and a generated story. A model continuing a narrative produces
another plausible page; an injected Will produces a **divergence with an author** — the record says who
moved, from what vantage, under which rules, knowing what. That is why a branch here is a parallel world
rather than an alternate take: it exists because someone external to the world chose, and the world
computed the rest.

It also sets the ceiling on automation. Exploration may propose futures and price them, the Program may
maintain interpretation, and generation may embody a scene. Those artifacts are not Will; the attributable
act of creating, revising, or selecting an interpretation can be. Anything that would have the engine
quietly select a lens or make the commitment on a participant's behalf is a bug in the product, not a
feature waiting to ship.

## What makes a world worth returning to

Branching is cheap to generate and easy to make weightless. Four properties are what separate a world with
something to teach from a generator of plausible pages, and they are product commitments, not flavour:

- **The world is interpretable, and its lessons persist.** A **theme** here is not a label attached to a
  story but a pattern the record keeps producing: what this world rewards, what it exacts, what it forgives,
  and who pays. Because the variables are recorded, a theme can be *shown* — the events that support it
  named, the counter-example admitted — and a reader who runs a second route is testing a reading of the
  world rather than sampling a different mood. Meaning that survives replay is the deepest thing the medium
  can offer, and it is the reason the engine's job is to compute consequence honestly rather than to
  deliver a satisfying page.
- **Time changes the rules.** A System is not a fixed rulebook but institutions, clocks, permissions, and
  constraints with their own histories: adopted at a moment, binding for an interval, eroding, and
  rewritable by what happens under them. A world where the rules can change under you is the only kind
  where a long game means anything, and the moment cursor makes that legible — the same place under
  different law, two hundred scenes apart.
- **Actors can be reinterpreted without losing their record.** Continuity preserves the evidence that
  makes a betrayal land and a loyalty cost something: commitments, debts, relationships, knowledge,
  speech, choices, and consequences. But evidence does not contain one final character. Acting makes the
  chosen reading explicit, so the same continuing entity may perform through different defensible State
  or Dynamics interpretations at different branch moments. The change is declared and attributable;
  neither a fluent model call nor a production preference rewrites the evidence underneath it. New reads
  should recur as evidence and situations change, but a person initiates and validates them; Acting is not
  an autonomous maintenance loop.
- **The aspiration is cinematic.** Staging is meaning — place, hour, who is present, what is withheld,
  where attention falls, what the cut leaves out. The prepared Experience should feel composed rather than
  narrated, and a World should feel inhabited rather than described. This is the quality bar the asset,
  scene, and pacing specs answer to; it is aspiration, and the honest word for how much of it ships today
  is *little*.

## Who it is for

**The roleplay-minded reader**: someone who would rather decide inside a story than watch one resolve. A
novel hands them a resolution already reached; Meridians hands them the fork and holds them to what they
chose. That is what makes this an alternative to reading rather than a variation on it — the second run
down the other branch is a different story, not the same pages again.

The same instinct is what a serious analyst brings to a live situation, which is why the analytical
configuration and the narrative one are the same engine and not two products wearing one name.

## The artifact is the novel

The consumable form is an **explorable visual novel**, and the competitor is the book. A book gives one
path through a world; a prepared world gives the world — many routes, many endings, POVs that see
different truths, and illustration at the moments that earn it. Replayability, immersion, and POV richness
are the three axes the claim rests on, and each is a property the book cannot have rather than one it
happens to lack.

**A prepared world does not generate while you read it.** Text and images are produced during production
and reading serves finished assets. That is a product property before it is an economic one: a page turn
has no latency, the artifact is fixed and reviewable, it can ship offline, and no reader watches a model
fail live. It is also why the economics work — production is capital, reading is free.

Two supply modes feed it. **Adaptation** experiences an existing canon as a visual novel — a work we own
or have the rights to, with public-domain canon as the case that ships. **Exploration** takes the same
world into its counterfactuals: alternate realities, multiple endings, branches scouted cheaply and
produced only where they earn it.

A scoped world can also carry a lesson without teaching it: the knowledge sits in the causal structure a
reader must navigate rather than in a paragraph explaining it.

## Fate is explored, not authored

What separates this from a branching story someone sat down and wrote is that the branches are **found**.
A Reading cuts a fan of candidate futures from an as-of vantage, without hindsight, and prices each by
prior rather than offering them as equals — so a timeline is interesting because the world's own causal
structure made it so, and a rare ending is rare because it was improbable and taken anyway. Exploration is
cheap and disposable; production is expensive and selective; only the timelines that earn assets get them.

This is the claim underneath the artifact: **not infinite generated content, but a search over possibility
that returns the few timelines worth living through.**

## Three mediums over one Domain

The novel is the lead artifact, not the only one. A Domain can be inhabited three ways, and they differ in
what is prepared before you arrive:

- **Scenario** — you *decide*. The card table: the situation is prepared, the move is not. It is the only
  medium that writes canon.
- **Experience** — you *read*. The visual novel: everything is prepared, so it generates nothing while you
  read it and writes nothing back.
- **World** — you *wander*. The Domain as a place, held at a chosen moment and explored freeform by cursor
  rather than in order, with situated roleplay inside it.

These are peers, not a promotion ladder; a Domain may support one without the others. Their cost shapes are
opposites and the distinction must be kept honest in copy: an Experience is **capital** — generated once,
then served — while a World **generates as you wander it**. The "does not generate while you read it"
claim belongs to Experience alone.

World's defining property is its clock. It is a static artifact of state pinned to a chosen **moment** on a
branch, so entering it restores who was where, what had happened, and what each character could know then.
That makes the same world re-enterable at any point in its own history — you can walk an earlier hour and
speak to someone inside what they were living through, rather than only meeting the story at its tip.

Everything below this section is the production line that makes such a world possible — and remains
useful on its own. A maintained Domain is a complete product for someone who never opens a novel.

> **Status.** The world model, branches, readings, the Script scaffold, and image generation ship. The
> dedicated reader interface, budget-bounded exploration, and asset production for promoted branches are
> specified in [the visual-novel specs](../../../specs/visual-novel/README.md) and are **not built**.
> World's explorable map, moment cursor, and situated roleplay are **direction** — the
> [World concept contract](../../../specs/world/README.md) is a working concept, not an implementation
> contract.

Meridians turns a premise, corpus, or maintained source into an inspectable **Domain**: a causal
composition of **System** (rules and constraints), **World** (characters, locations, artifacts,
information, and state), and **Fate** (open questions and bearing). Branches preserve official and
alternate story states without becoming competing sources of truth.

The commercial promise begins before simulation. Research aggregates current sources and relevant news;
the Program keeps an evolving body of evidence and interpretation coherent; the Domain makes the actors,
systems, relationships, commitments, and uncertainties available for serious analysis. That is a useful
individual product even if the user never plays a Scenario or deploys a World.

The experiential promise is **total recall without total attention**. Meridians preserves the source,
structure, provenance, and history needed to recover how a Domain reached its present state, while the
Program brings forward only the movement that materially bears on the live question. “Total” means the
world remains re-enterable and its load-bearing detail recoverable; it does not mean every token is held in
working context, every source deserves equal attention, or the model is infallible. The world remembers
more so the person can attend to less.

- **Scenario** is the light, abstract decision substrate at a branch frontier. Played directly or
  asynchronously with a trusted group, its card grammar makes situated Will, uncertainty, and consequence
  responsive without demanding a long live session.
- **Episode** is the immersive Stageplay presentation of story, and the unit of delivery inside an
  Experience. A live Episode can wrap Scenario so key players roleplay consequential choices without
  knowingly entering the card game.
- **World** is the explorable medium: a branch's state rendered as a place at a chosen moment, wandered by
  cursor and inhabited through situated roleplay. It can make story-built places tourable and test ongoing
  activity, economies, or social presence. It is not the headline product or an MMO promise.

> **Domain defines. Branch preserves. Scenario resolves. Episode embodies. World inherits and lives.**
>
> **You decide in a Scenario, read an Experience, and wander a World.**

## One decision substrate, two entry modes

Scenario and Episode are interoperable:

```text
play Scenario → commit branch scene → optionally render it as an Episode

play live Episode → reach critical choice → resolve wrapped Scenario → stage the result
```

This is the key interaction contract. Direct Scenario optimizes for rapid simulation. Episode-wrapped
Scenario optimizes for roleplay, narrative engagement, situated information, and felt consequence. Both
use the same governed action boundary and branch lineage.

## The missing training loop

Durable world memory is necessary for replayability, but it is not sufficient. Extraction can create
substantial first-session value and the Program can keep the Domain current, yet a replayable practice
also needs the participant to become more capable over repeated returns. The working loop is:

```text
recall → orient → inhabit → commit → consequence → debrief → adapt
```

This loop is the purpose, and its evidence status is stated rung by rung — *designed*, *measured in fiction*,
*shown to transfer* ([strategy 12](../strategy/12-the-practice-monopoly.md#the-proof-programme)); none of it
ships today as integrated progression. The Program and Memory
Palace can recall and orient; an Episode can make a situated problem legible and felt; Scenario can accept
the consequential commitment; the branch or later World can preserve what followed; Stories, Positions,
and the Program can support a debrief that changes the next read. A complete turn should eventually leave
two legible records: **what changed in the world, and what the participant learned to notice, judge, or do.**

Progression should mean increased ability—stronger comprehension, calibration, causal judgment,
coordination, perspective-taking, tool use, or creative command—not an abstract level detached from the
declared Domain. Roles, resources, permissions, quests, economies, scores, spaced practice, replay, and
tool upgrades remain candidate mechanics. The training loop must be useful before any MMO-scale World
exists and must not turn a bounded simulation result into a verdict on a person.

Repeated over a **journey** — weeks of hours aimed at one trait the person named, each read back on a
calibrated scale and the next shaped by the reading — the loop leaves the
[practice record](../strategy/12-the-practice-monopoly.md) and, with the proof programme, the evidence that
the change followed the person out ([Human practice](human-practice.md#the-training-loop-is-the-product)).
The boundary in the previous paragraph governs it.

## Episode: one medium, two postures

A **catch-up Episode** embodies an already accepted scene, branch move, or canon event. Playback introduces
no new Will.

A **live Episode** stages a branch-tip situation for one or more participants, opens a bounded Scenario
decision, resolves their Will, and continues from accepted state. Players may issue orders, compose plans,
coordinate, trade, disclose information, allocate resources, or set processes in motion.

Single-player and multiplayer are valid. Perspective-specific passages are optional and used where the
scene benefits. A humane, possibly daily cadence is a design hypothesis for working professionals, not a
fixed mechanic. A user may follow several ongoing stories across divergent branch heads.

## Strategic puzzles and visual explanation

Episode puzzles concern direction rather than object locks: people, orders, timing, routes, evidence,
capital, incentives, permissions, institutions, information, contracts, and scarce tools. Maps, ledgers,
graphs, dossiers, boards, and infographics make those systems legible. The side-scrolling pixel-art stage
is the embodied home; other views are instruments chosen for the scene.

Episode 006, *The Supply Tendon*, is the catch-up reference. A live version would leave the logistical
plan open, resolve it through wrapped Scenario, and stage its consequence.

## Story branches seed Worlds

Direct Scenario and live Episode cultivate the main story or several divergent stories. Over time they
establish the ontology, cast, places, institutions, history, rules, information, resources, and production
assets from which a World can be deployed.

A World pins one sufficiently mature branch coordinate and inherits its exact lineage. It may initially be
a semi-permanent place to tour or revisit. Bounded generation can fill unstaged gaps under inherited
ontology, style, geography, and provenance. Later configurations may support situated human and agent
seats, financial and informational economies, social organization, and MMO participation.

Multiple branches may seed different Worlds. A likely posture is one designated official canon branch
alongside preserved alternative timelines, but canon designation and asset alignment across branches are
open design work. A World never silently advances or rewrites its source branch.

Replayability is also open. Resettable runs, forks, comparison, continuous presence, and persistent
activity are possible configurations rather than the current definition.

## The five forces and their interpretations

A Domain is held by the **pentad** ([spec](../../../specs/pentad/README.md)): five peer forces, three
stored in the record and two derived across it.

- **World embodies** — entities, places, artifacts, resources, relationships, information, and state. *Stored.*
- **System constrains** — rules, institutions, economies, cadence, and tools. *Stored.*
- **Fate remains open** — unresolved possibility, pressure, and consequence. *Stored.*
- **Theme recurs** — the durable meaning that repeats across events; reads, never writes. *Derived.*
- **Causality connects** — what can lead to what, traced as **Flags**: turning points that signal Causality
  into action. A **Thread closure** is a Flag that also settles a Fate thread; it is a derived projection.
  *Derived.*
- **Will makes the move** — a human or agent chooses through the canonical action boundary. Not a force.

**Acting is the first-class entity-scale interpretive practice** and, in the pentad, an interpretation of
System: it turns attributed evidence and current Frame or Control dynamics into ordered, selectable State
and Dynamics graphs. The Entity Interpretation Graph is Acting's artifact. First-class does not mean a
writer of state: Theme, Causality, and Acting read the record without changing it. Will is attributable
authority over authoring or selecting the lens and making the move, never a stored force. Stageplay
presents and embodies; it never makes an action canonical by depicting it.

## Delivery truth

**What ships now:** source extraction and Research, structured Domains, the three stored forces
(World/System/Fate) with their measured snapshots, Theme analysis, the deterministic Thread closure projection,
Production → Actors, Acting through State/Dynamics interpretation storage and selection, entity-chat
context substitution, canon and branches,
generation and revision, reasoning graphs, Positions, the Program, Stories, Studio, Scenario
primitives, the canonical record, operations, Hosted, MCP, and configurable setups over that substrate.
Together these support news aggregation, maintained Domain analysis, and light Scenario play.

**What is being built and proved next:** catch-up Episodes through Stageplay and the shared seam that can
render a Scenario scene as an Episode.

## Stageplay and SFX Studio quality gate

Stageplay and the SFX library remain experimental until SFX Studio can **reliably** produce engaging,
long-form animation and storytelling that is competitive in online viewing contexts. Technical validity,
short clips, isolated polished shots, or one strong Episode do not cross that gate. Reliability requires
repeatable evidence across productions: sustained narrative comprehension and attention, coherent identity
and movement, expressive acting, purposeful staging and editing, strong sound, continuity, reusable
production assets, and viable repair economics.

Progress toward that gate is a sequence of bounded hill climbs. Each experiment should identify the most
visible audience-quality failure, repair the smallest reusable production layer that owns it, deliver a
viewable proof, preserve accepted and rejected evidence, and name the next constraint. “YouTube-level” is
a delivery and competitive context, not a fixed duration, imitation target, or claim that the current lab
already produces release-ready long-form work.

**What follows directionally:** live Episode-wrapped Scenario, strategic decision hand-offs, optional
perspective delivery, governed continuation, and branching multiplayer story development.

**What remains open World direction:** maturity and deployment criteria, tours, generated gap completion,
canon and asset lineage across branches, persistence, replay, economies, social activity, and MMO scale.
No real-capital game, investment product, or public World economy is promised.

The maintained concept contract is [World](../../../specs/world/README.md); the experiential relationship
is owned by [the human and agent experience](human-practice.md).

Raw Markdown source