MeridiansMeridians

The 2026-09-01 pivot — Explorable VN worlds as the consumable artifact

From the Meridians Wiki · Public · Maintained · joint

  • What changed: The lead promise moves from "a maintained Domain you can understand and play" to "worlds you can read" — an explorable visual novel positioned against the book on replayability, immersion, and POV richness. The existing stack is not retired; it is recast as the production line.
  • What is retained: the Domain/System/World/Fate model, the Program, branches and readings, Scenario as the resolution substrate, the authority line ("Domain defines…"), and the maintained-Domain-as-complete- product claim.
  • What is recast: Episode/Stageplay is now the delivery form of the visual novel rather than a parallel narrative product; Script is named as the VN scaffold; "total recall without total attention" becomes a supporting mechanism rather than the headline.
  • Consequences to watch: copy must not imply a shipped reader; adaptation of licensed canon is a rights question, not a capability; per-reader cost claims depend on the no-live-generation property holding in the eventual implementation.

Amendment (2026-09-01) — three mediums, not one artifact

  • What changed: the artifact is not alone. A Domain is inhabited three ways — Scenario (decided), Experience (read; the visual novel), World (wandered) — as peers rather than a promotion ladder. Experience is the name for the VN medium; Episode remains its unit of delivery and Script its scaffold, so no existing vocabulary is renamed.
  • What World now means: the Domain's state rendered as a place, pinned to a chosen moment on a branch and explored freeform by cursor over the location map, with situated roleplay inside it. Choosing a moment restores who was where and what each character could know then. This replaces the vaguer "semi-permanent sandbox that may become livable" identity while keeping economies and MMO scale as deliberately open.
  • Two constraints this creates. First, the cost shapes are opposites: an Experience is capital and serves finished assets, while a World generates as it is wandered, so "does not generate while you read it" is an Experience-only claim and World spend is an ordinary metered lane. Second, presence is derived from canonical Scene.locationId + participantIds rather than stored, which yields last known position rather than a trajectory — the interface must say "not recorded" instead of inventing travel.
  • Consequences to watch: the two senses of "World" (the force inside a Domain, the medium over it) must be disambiguated wherever both appear; nothing said inside a World may write canon; a stored movement register, if ever added, arrives as a typed scene delta on the reducer path.

Amendment (2026-09-07) — Acting is first-class

  • Continuity is evidence, not a fixed characterization. The entity record preserves what can be supported across speech, choices, relationships, knowledge, Frames, Control, and consequences. It does not contain one final actor or one inevitable performance.
  • Acting is the interpretive practice. Production may create and bind immutable State interpretations of entity evidence and Dynamics interpretations of Frame or Control topology. The same evidence may sustain several defensible graphs, and a continuing entity may carry discrete interpretations at different branch moments. In the craft sense, the actor is a person making and validating a strong read that compresses a complex situation into something playable.
  • Actors is the Production home. AI may suggest guidance or draft an attributed graph, but the Director reviews and selects the lens. Chat, Scenario, and Perspectives consume the effective Acting context so the interpretation changes performance rather than remaining a decorative production artifact.
  • Renewal is recurring but manual. New evidence, a changed situation, or a weak prior read should prompt another interpretation. For now the person initiates Suggest, Draft, review, and binding; the system may show staleness but does not autonomously regenerate or select Acting context.
  • The boundary remains intact. Acting is first-class context engineering, not another causal force, canonical psychology, or Will. Evidence remains recoverable; an interpretation binding is attributable; only the governed action path makes consequence history.

First-class clarification 🔒 · Acting reads and manual cadence 🔒

The 2026-08-20 pivot — story cultivation into lived Worlds

Status: active conceptual migration. Anchor: Positioning. Experience contract: Human practice. Concept contract: World.

Refinement record

The 2026-08-20 pivot established Domain states as material for situated, plural-agent Worlds. The 2026-08-27 sequencing decision put audiovisual Episodes through Stageplay first. Founder discussion on 2026-08-28 then clarified the complete relationship and its commercial order:

  • Utility leads. Research gathers current news and sources; the Program maintains interpretation; the Domain and Graph support serious analysis before any game layer is required.

  • Social play extends the utility. Scenario supports light solo or asynchronous friend-group play. Episode develops the same decisions into deeper, perspective-shaped roleplay that can fit around work and life.

  • Scenario is the decision substrate. It can be played directly as the fast abstract card RPG, or a bounded Scenario instance can be wrapped inside a live Episode so roleplay and critical decisions carry the same governed resolution without exposing card-game machinery.

  • Episode has two postures. A catch-up Episode embodies accepted branch/canon material. A live Episode stages context, accepts Will through its wrapped Scenario, and embodies the resolved continuation.

  • The two directions interoperate. A Scenario-developed scene can be rendered as an Episode; an Episode player can advance a branch through Scenario without knowingly switching products.

  • Branches cultivate Worlds. Direct Scenario and live Stageplay can develop several main-story timelines. Their places, characters, institutions, history, ontology, and production assets accumulate into the material from which later Worlds may be deployed.

  • World is an experimental, semi-permanent inherited sandbox. It pins a sufficiently developed branch coordinate, may begin as a tourable place, and may later support situated activity, information/financial economies, or MMO-like presence. It is not the product lead or a shipped MMO promise; replayability is plausible but no longer the defining claim.

  • Canon and assets need explicit lineage. Multiple branches may seed different Worlds. A likely posture is one designated official canon branch plus preserved alternatives, but canon designation, branch asset alignment, generated gap completion, and replay semantics remain open.

  • Replayability needs a training loop, not only persistent state. Extraction supplies strong upfront utility and Scenario supplies a real turn, but the product does not yet join Program orientation, Episode immersion and puzzles, Scenario commitment, consequence, debrief, and increasing participant ability into one repeated practice. Exact RPG progression remains open; the durable requirement is that a loop eventually develop both the world record and the participant's capability.

Original multimodal direction 🔒 · puzzle-world pivot 🔒 · live-story correction 🔒 · interpretation boundary 🔒 · Episode/Scenario/World refinement 🔒 · live Episode/branch RPG synthesis 🔒 · Scenario-inside-Episode and World-seeding synthesis 🔒 · private/public Episode perspective synthesis 🔒 · utility/social play/experimental World positioning 🔒 · SFX long-form quality gate 🔒 · replayability and training-loop direction 🔒

The grammar

Domain defines. Branch preserves. Scenario resolves. Episode embodies. World inherits and lives.

  • System supplies programmable rules, structures, constraints, clocks, economies, and tools.
  • World supplies characters, locations, artifacts, information, resources, relationships, and state.
  • Fate holds open questions, possibility, and bearing.
  • Theme recurs — the meaning that repeats across the record; derived, never written.
  • Causality connects — what can lead to what, traced as flags; derived, and its extraction unbuilt.
  • Will enters when a human or agent chooses through the canonical action boundary.

The stored triad (World, System, Fate) is the simulated substrate; Theme and Causality are derived across it — the five together are the pentad (see the 2026-09-09 amendment). Acting is the first-class entity-scale interpretive practice and a post-hoc interpretation of System; an Entity Interpretation Graph is its context artifact. Neither Theme, Causality, nor Acting writes state. Will is not a stored force: it is the attributable authority exercised when a person or agent creates, revises, or selects an interpretation, and when they make the move.

Scenario inside Episode

The common decision path is:

situated context → available or authored moves → commitment → governed resolution → branch consequence

Direct Scenario exposes this as responsive cards, mandates, and rapid simulation. Live Episode Stageplay embeds the same bounded mechanism inside character performance, partial information, strategic puzzles, maps, ledgers, dossiers, and infographics. The player roleplays the key person and makes the consequential choice; they need not know they are operating Scenario underneath.

The Episode is composed per seat: shared public beats establish common reality while private beats expose only the knowledge, memory, permissions, and tools reachable by that role. Wrapped Scenario resolution retains this asymmetry. Distinct seat renditions remain compatible projections of one branch state.

The renderer never makes the move true. Scenario resolution and the one writer preserve authority. A prepared continuation is revalidated against accepted branch state before it is staged.

From branching stories to Worlds

A fictional Domain may sustain many live branch heads and let many players seed divergent timelines. Direct Scenario can develop a scene cheaply; an Episode can later render it. A live Episode can develop the same frontier immersively. A Director may designate one official canon branch while retaining alternate branches as playable history and possible World origins.

Once a branch has developed enough of its story and setting, it may seed a World. The World inherits exact ontology and asset lineage from that branch; bounded generation may complete unstaged connective space. The first posture may simply be a semi-permanent place to tour. Richer postures may add ongoing roleplay, work, trade, information flow, financial systems, social organization, or MMO participation.

World does not silently advance its source branch. Whether it resets, replays, persists continuously, forks, or sends proposals back to story canon remains open.

Amendment (2026-09-01) — the company is the art and science of fate

  • What changed: the framing rises from a product claim to a company one. Meridians is a creative studio whose subject is fate — what a world could still become, and what it costs to find out. The brand tagline becomes "Explore the fate. Live the one you chose.", replacing "Know what matters. Play what follows.", and the supporting premise is because everything is recorded, everything can be replayed.
  • What it adds to the copy: the medium's lineage is named (the choose-your-own-adventure's fork plus the visual novel's craft, over a recorded world that computes what a choice costs); branches are framed as parallel worlds rather than alternate takes, since they differ in recorded variables that can be pointed at; and generatable worlds, the card game, and situated roleplay are described as one gravity rather than a product list.
  • Will injection is promoted to a first-class idea. A Meridians world never moves on its own: Will enters from outside the simulation — a reader at a fork, a seat at the card table, a Director, or an autonomous agent through MCP — as a typed, attributable action, and the world computes the consequence. This is not new architecture; it is the existing "Will makes the move" force stated as a product claim, and it sets the ceiling on automation: exploration proposes and prices futures, it never commits.
  • What did not change: the three-medium contract, the Experience-only no-live-generation claim, the shipped/specified/direction boundary, rights-safe canon language, and authorship attribution.

Amendment (2026-09-09) — the pentad: a visible triad and an invisible dyad

  • What changed: the trinity is now the visible triad of a five-force pattern ordered by visibility — World → System → Fate → Theme → Causality. Theme and Causality form the invisible dyad: forces derived across the record rather than stored in it. Causality — what can lead to what — is named for the first time and is the most invisible force, because its only trace is the flag: a Fate question collapsed to a certain value on one branch. A branch is therefore an assignment of flag values over one question set; parallel worlds differ in their flags.
  • What is retained: every stored force, its extraction and measure; the reducer as one writer; Will injected from outside; the flag ontology at reader grain (now stated to be the same object at world grain); Acting and the spatial map unchanged in code.
  • What is recast: Theme goes from "auxiliary instrument over the trinity" to an invisible force that reads and never writes — the non-causal protection is kept as a non-writing rule. Acting and spatial mapping are post-hoc interpretations of System. Questions become a post-hoc synthesis over all five forces; prose, plans, scripts, and Episodes are generated from them through a reconfigurable pipeline of typed stages rather than one fixed run.
  • Consequences to watch: the branch diff over flags, the question synthesis stage, and the pipeline configuration record remain unbuilt; Activity and Signature now aggregate the five channels while the pentad grade keeps its separate currencies. Copy must not imply every planned pentad surface ships. The pentad spec owns the frame, the status ledger, and the open decisions.
  • Measurement follows visibility. What the scorecard measures today is activity — per-scene increments over all five channels. Longitudinal readings remain distinct composition diagnostics: relations across the record (recurrence, open mass, weighted collapse and reach). Candidate measures, the surfaces each force lives on, the post-hoc settings that keep the interpretations of System current, and a six-step roadmap with falsifiers are planned in the pentad in the app; none of it is built.
  • Amendment (2026-09-10) — visibility is the ladder, not the theory. The visibility order is the order the forces were discovered in, because it is the order each becomes computable from a record (counts and constraints → a measure → a difference across scenes → a sum across scenes). The theory is the two cycles the five phases of wuxing carry — a generating ring World → System → Causality → Fate → Theme → World and a controlling star in which each force bounds the one two steps ahead. "Visible triad / invisible dyad" is to be read as stored / derived, an implementation fact, not a rank: the five are peers. Pentad spec §2.4 and white-paper Principle 8 own the statement.

Direction, not implementation

Fixed Stageplay production is active incubation, not a shipped Studio capability. Live Episodes and World deployment are not shipped. Current runtime invariants remain true until intentionally migrated: one writer, attributable actions, the filesystem record as canon, IndexedDB as a projection, and the existing MCP operation surface.

The incubation has an evidence gate rather than a schedule. SFX Studio remains experimental until it can reliably produce engaging long-form animation and storytelling for online audiences across more than one production. The lab should continually hill-climb the clearest audience-quality constraint—acting, movement, identity, staging, editing, sound, comprehension, continuity, or production economy—through a bounded proof that improves a reusable layer. Valid output, higher resolution, additional generation, and one successful short do not by themselves move the capability boundary.

The first interactive proof should support both entry routes over one scene: play it directly in Scenario and render the accepted result as an Episode; then play an Episode-wrapped Scenario from a situated role and stage its accepted continuation. The first World proof should inherit a small developed location and cast from an exact branch coordinate, fill one labelled gap, and make the result tourable without changing the source branch.

Amendment (2026-09-19) — 五行 as macro symbol, not micro mechanics

  • What changed: the generating ring and controlling star leave the engine as mechanics. 五行 remains the macro symbol for the five-force world model — element names and glyphs, the pentad mark, pentagon geometry, and stable display order — but Meridians asserts no theory of how the forces relate scene to scene.
  • What is measured: Constitution is now a fitted, per-record projection of unconstrained lagged couplings: which force pairs move together or apart, at lag 1 or 2, with their sign and evidence. Couplings are readable only with at least six pairs and |t| ≥ 2; an empty fitted result means the record reads as independent, while a too-short record is unread.
  • Why: tests across nine hosted records found ring/star cycle fidelity at or below chance in all nine. Generation and prompts therefore remain aware of a record's observed couplings without steering toward any presumed relation; Will stays external and attributable.