The north star

# The north star

**Training, learning, immersion.** Meridians builds worlds a person can be dropped into cold, read on the
spot, act in under consequence, and come back from changed. Everything else — the Domain graph, Scenario,
Stageplay, branches, Factory, the record — is how those worlds get made and kept.

The picture to hold: **a spy inserted into an unfamiliar country.** No briefing covers what they will
meet. They have to read a room, work out who is telling the truth, judge what a move will cost, act
before they are sure, and live with what follows. Then they are pulled out, debriefed, and sent in
again, better. A training world is that insertion, made safe, short, and replayable, in a place worth
being in.

## What we sell: training worlds and pocket dimensions

Two things, carved from one recorded Domain.

- **A training world is an hour.** One original world, one question the player must get right, a
  canonical good route earned by right decisions, real diversions, and a reading given back afterwards.
  Solo. Sold by the hour and, in sequence, as the journey ([12](12-the-practice-monopoly.md)). This is
  the Experience (VN).
- **A pocket dimension is a place.** A small kept world that holds its own history between visits,
  admits a few named people as its characters, and can be stepped back into — what happened there stays
  happened, and what another seat knew is there to be found. It is the populated World (VNMMO) in its
  careful, bounded form: a house, a week, six seats, long before it is a server. Sold as a place you and
  yours keep.
- **Agents get the same two.** A training world with a policy in the seat is a test; a pocket dimension
  with agents in it is a bounded arena. The log is the result ([13](13-consequential-test-worlds.md)).

The look is anime — colour, motion, sound; never black-and-white stills — because immersion is the
requirement and a still page cannot put a person inside a room.

Both run on two magics: **the heart**, which commits, and **the mind**, which builds what remembers and
computes what followed — technology is the mind made manifest. A training world is an hour in which the
heart acts and the mind's record keeps what it cost; a pocket dimension is a place that record holds so
several hearts can return to a history they made. The five forces are how we read the ground they meet
on ([Two magics](../../foundations/TWO_MAGICS.md)).

## Five forces, five north stars

The pentad is Meridians' own ontology of worlds ([pentad spec](../../../specs/pentad/README.md)); *wuxing*
is one earlier instance of the same five-fold shape, not its source. The five are also the company's five
north stars — training, learning, and immersion are what they add up to — and our values optimise for the
same five areas, in a person as much as in a world.

The original source is **Fate**: control over one's own probabilities — how a person divines their own
fate rather than receiving it. Effort is continual, but the leverage is not evenly spread; it sits in a
few **causal moments**, the decisions after which what follows is different. Finding those moments, and
acting in them, is the practice a training world rehearses: one hour, one question that must be got right.

- **Fate** — what is still possible, and the ability to shift it. A route not yet closed is a fate still
  open; the work is learning which decisions move it.
- **Causality** — leverage. High causality in a life means what you do leads to what follows and you can
  see the chain; the flag is that chain made visible.
- **World** — the key: the inner world and character development. Who a person is changes with what they
  do, and the protagonist is the one whose acts the World remembers.
- **System** — the boundaries: the rules, institutions, and cadence a world is governed by, and where
  within them a move has purchase.
- **Theme** — the guides: what a history keeps returning to, read after play, pointing at what to learn.

The five are a macro world model and symbolic frame, not a fixed micro theory. No general scene-to-scene
generating or controlling relation is asserted: each record may show its own lagged couplings, or read as
independent. What takes a person further is understanding their own path and goals through the history the
record actually shows, and where their causal moments fall.

Together the five are one lens for self-development, for protagonist-led world development, and for
reading a world's spirit ([World Spirit](../../foundations/WORLD_SPIRIT.md)) — how it is governed and what
it is becoming. The lens stays domain-native: it reads a declared world and a person's own history in it,
never a rank, a diagnosis, or a verdict ([VALUES](../../foundations/VALUES.md)).

Earth has little magic because there the two are pulled apart: what the mind records is rarely what a
person committed, and what a person commits is rarely kept and returned to them. A story has much
because it is fully kept — every choice inside it lands and comes back. That is why what we sell is
places, not content: a world small enough to keep all of it, for a body that stays on Earth.

### Meridians is a causal simulator; the visual novel is its interface

Said plainly: **Meridians is a causal simulator.** Have a consequential situation coming — a negotiation,
a hard conversation, a new team, a new country, a hostile room? **Train the pathway before you live it.** The
visual novel is not the category we sell into; it is the highest-bandwidth interface a person has for
being *inside* a simulation — faces, pauses, tone, the temptation to say the wrong thing — and it is
why the consequence lands as felt rather than as a graph. Public copy leads with the world and the
decision ("enter a living world, make decisions, see what happens"), never with the format; nobody has to
identify as a VN player to want to know what they caused.

This changes the priorities and what we monopolise (2026-09-18). **First, causal simulators enabled by
Meridians World Search (MWS) (MCTS)** — the large value, and where value is created today: simulations causally deployed
from a voice interview and rehearsed. **Second, micro sonic analysis** — the value-add that gives the
simulation its felt grain. **Third, the VNMMO** — a stretch goal on a separate roadmap, a persistent universe that relies
on the causal simulator and runs it more automated, built on the side and never instead
([strategy 14](14-the-causal-simulator-monopoly.md) argues it and carries the build-ordered roadmap;
[strategy 12](12-the-practice-monopoly.md#what-we-monopolise-2026-09-18);
[specs map](../../../specs/README.md#priorities-2026-09-21)). The primary roadmap is **Causal MCTS VNs**;
VNMMO is a stretch goal on a separate roadmap. The causal simulator is a sci-fi technology
in earnest: the instrument that lets a person practise and prepare, causally, on known paradigms and
ontologies that describe the world — what the flight simulator is to a pilot, built for the situations
that had no simulator because nobody had an ontology that could run people, incentives, and information
forward. That is why the spacefaring frame stays on the landing: it names the machine by its purpose. And
the machine compounds: what a person tells it by interview and what they do inside its worlds become a
context, theirs alone, from which each next simulation is built more realistic for them.

Development needs consequence, and ordinary life is not causally dense enough to supply it: most
decisions carry nothing, what follows arrives late or never, and the causal moments that would train
judgment are rare and unrepeatable. A training world is causality and consequence built from nothing,
wrapped in an experience strong enough to be felt, so that the practice a life offers once in a decade can
be had in an hour and had again. The spacefaring frame makes the need vivid (a sealed hull is the thinnest
world there is) but the need is Earth's.

**The loop that brings a person back is their own history.** Something happens → you form a model of the
person or the world → you make a commitment, not a flavour choice → the world remembers (people adapt,
doors close, new information surfaces) → you want to know what you caused → you return, because the next
state of the world is now personally meaningful. A conventional story asks *what happens next?*; a causal
simulator asks ***what did I cause?*** — did she trust me, why did the faction turn, was that the cost of
something I did three hours ago? The player investigates their own timeline instead of consuming authored
content. This is what "causally sticky" means, and it is the opposite of adding drama: a quiet
conversation carries weight once the world has taught you that small things become causes (latent
significance). The promise in one line: **your attention changes the world, and the world remembers.**

**A simulator says "try it", not "here is what to do."** Preparation → simulation → consequence →
reflection → retry. Retry is not failure; it is the experiment — *what if I am more assertive; what if I
concede at once; what if I ask instead of argue* — a counterfactual run on yourself, and the branch record
is what makes the comparison honest. The **characters are the interface to the causal system**: their
incentives, memories, relationships, and hidden information decide how the world answers, so playing a
person well *is* reading the system. Two layers carry it — macro, `World → Forces → Causality →
Consequences`, which the pentad already measures as deltas; micro, `Words → Rhythm → Sound → Emotion →
Perception`, the felt grain of a scene, which is stated direction and unbuilt ([pentad
spec](../../../specs/pentad/README.md)). A simulator that only models what happened is not enough; it has
to carry what it felt like while it happened, and the VN is the medium that carries both.

**MWS exposes the game; it does not play it for you.** MCTS over the branch tree spends its effort
on the high-uncertainty, high-consequence forks and returns pathways with their costs — *demand
aggressively → defensiveness → stall; bring market evidence → the conversation opens; reframe around
scope → a different pathway appears*. What comes back is the structure of the game actually being played —
incentives, information asymmetry, timing, hidden constraints — not the correct answer, and never a
commitment made on the person's behalf ([fate exploration](../../../specs/fate-exploration/README.md),
[branch refinement](../../../specs/branch-refinement/README.md)). "Learn to play the game better" means
seeing the board, then choosing.

**Direction: the interview as the first experience.** The natural front door is not a form but a
conversation — a voice agent that interviews the person like a researcher (goal, situation, actors,
incentives, history, constraints, what they know and might be missing, how they usually respond, what is
at stake, what they want to get better at), notices where the causal picture is thin, and asks. Interview
→ situation model → actor models → causal hypotheses → world construction → MWS → simulation →
play → consequence → reflection → re-simulation. The play is then a second source about the person, kept
beside the self-report and never over it: what they said they would do, and what they did under pressure.
Today the situation-to-world step runs by hand through MCP
([timeline-wisdom](../../../../.agents/skills/timeline-wisdom/SKILL.md)); the voice intake and the
automated pipeline are direction, described as such. The promise the pipeline is built toward: *tell us
what you are preparing for; we build the world, simulate the pathways, and let you rehearse it before
reality does.*

Public copy keeps the sci-fi frame on purpose: it lifts a reader out of realism and keeps them open to a
claim that would sound grandiose said plainly — that this future is far off and inevitable, and that
causal simulators and VNMMOs will be foundational in it, as basic to a confined crew as air and memory.
The frame is a door, not a forecast; the status line beside it (World experimental, no spacecraft, no
transfer evidence yet) stays.

The psychometric reading of the same fact: with enough play, the hours that develop a trait also measure
it. Latent traits are read from what a person did under consequence across many worlds, on a calibrated
scale ([strategy 12](12-the-practice-monopoly.md)), and developed by the next hour aimed where the last
one pointed. Measurement and development are one loop, not two products; the reading stays the person's
own, in the world's words, and is never a rank, a diagnosis, or a verdict
([VALUES](../../foundations/VALUES.md)).

### Pocket dimensions, with care

A pocket dimension is the asset no model regenerates cold: a history made by the people who were in it.
It is held to five rules, in order of what would be lost without them.

1. **Earned entry.** Training worlds first; a pocket dimension is populated by people who have practised
   in that setting, not by a lobby.
2. **One canon, many wills.** Nothing said inside writes the record by being said; consequence is
   computed, attributed, and kept. This is what makes the history worth having.
3. **Seats know less than the whole.** Each seat sees what it could know; the debrief's sharpest line —
   *what the other seat knew* — depends on it.
4. **The person stays private inside the crowd.** Others see your acts and their consequences, never your
   reading. No leaderboard, no trait, no rank.
5. **Bounded before persistent.** A pocket dimension is a run with a beginning and an end until bounded
   runs produce debriefs a solo hour cannot; only then does one stay open.

Small does not mean idle: **training continues in limited space.** A house and six seats is still an
insertion — a question to get right, a cost to see coming, a seat that knew what you did not — and every
visit is read back like an hour. Sealed, scarce settings suit it: a ship, a station, an orbital
datacenter with power, heat, radiation, bandwidth, and time-lag as the rules. At the far horizon, such
datacenters carry humanity's memory and human-agent pocket dimensions beyond Earth, exploring the galaxy
on our behalf and returning evidence to questions no one should answer by assertion: are we alone, what
might intelligence become, and what more does existence contain? They are worlds to build and train for,
not claims about infrastructure Meridians operates today.

What the engine owes both is **coherent continuation**: after any number of decisions, seats, and
returns, the record still says what rules applied, what each seat could know, which Will did what, what
changed, and what remains open. A world that cannot answer those cannot train anyone.

What the product owes the person is **real change, proven** — and a mirror, not a grade. Hours become a
journey aimed at a trait they chose; the reading measures how far they moved, on a calibrated scale with
its uncertainty, in a declared world from a declared seat; and whether it followed them out is shown by a
study we run, not assumed ([12](12-the-practice-monopoly.md)). Never a verdict on them, never a prediction
of their life.

Raw Markdown source