Kalsa/workshop/Direction and Constraints.md

title	status	document_type	source	updated
Kalsa Direction and Constraints	governing	project-charter	https://chatgpt.com/share/6a89731d-fe80-83eb-acb4-b2eb9e1fc959 operator brief, 2026-08-22	2026-08-22

Kalsa Direction and Constraints

This vault develops Kalsa as a high-fantasy setting for Ghostlight Dungeon. The exported teenage notes are the seed and historical witness, not a finished canon and not disposable embarrassment. New work must preserve their strange specificity while replacing unsupported assertion with institutions that can actually reproduce themselves.

Product direction

• The setting must support persistent play in Ghostlight Dungeon: places, factions, offices, obligations, conflicts, rituals, material flows, and histories should generate situations rather than merely decorate them.
• Institutional depth should eventually match the neighboring AetheriaLore vault. This is a structural benchmark, not permission to import Aetheria's lore, voice, ontology, or assumptions.
• Kalsa is an enduring generative artifact: author canon feeds world compilation, play creates new pressure, and only reviewed outcomes return through canonization. The vault is not disposable campaign scaffolding.
• Ghostlight ships an opinionated world kernel. Settings generalize by compiling their actors, knowledge, affordances, constraints, resources, and causal rules into that kernel; runtime prose does not invent the rules anew.
• Obsidian is the canonical human authoring surface.
• Quartz is the default reader projection, not a mirror of author truth and not a second source of truth.
• The Public vault is an inhabitant-facing knowledge commons. Players explore what Kalsan cultures teach, dispute, preserve, translate, and misunderstand; they do not receive the author's comparative ontology in a friendlier font.
• Agent memory is a bounded project-memory surface. It records durable judgment, decisions, provenance, open questions, and iteration state; it does not silently become setting canon.

Experience and reveal doctrine

Kalsa must first work as high fantasy. A reader or player can meet dwarves, elves, orcs, ratfolk, gods, shamans, prophets, curses, miracles, priesthoods, relics, ruined cities, and dangerous depths without receiving a lecture about the machinery beneath them. “Vanilla” describes the legibility of the entrance, not generic institutions, and not a lesser or false version of the world.

The intended discovery gradient is:

1. High-fantasy life: the world is encountered through ordinary needs, spectacle, worship, law, danger, kinship, work, and local history.
2. Situated explanation: cultures preserve useful but incompatible accounts through doctrine, ritual, prejudice, testimony, maintenance, and myth.
3. Reconstructible pattern: persistent investigators notice repeated structures, limitations, ruins, inheritances, machine responses, and divine inconsistencies.
4. Divine explanation: a god may know vastly more than mortals and deliberately perform certainty, while remaining bounded, interested, capable of error, and ignorant of the complete author model.
5. Author ontology: the deepest causal account makes every earlier layer more coherent without declaring the fantasy experience unreal.

The old technical vocabulary is primarily author-facing. Kaos, the formal capability axes, signatures, Projections, Beast taxonomy, Ring experiments, and related classifications are not default inhabitant vocabulary. A culture may discover a fragment or invent a locally useful analogue. The author taxonomy constrains what can happen; characters explain those limits from the evidence available to them.

Public explanation must therefore have an epistemic owner. A claim about magic, history, gods, peoples, or ruins should be attributable to a culture, office, school, lineage, text, witness, or legible synthesis of named sources. Neutral editorial connective tissue may help readers navigate, but it may not smuggle author-only categories into common knowledge. Cross-cultural comparison is an act of translation performed by somebody, not the universe speaking in tables.

Inhabitant models deserve the same institutional depth as other machinery. People name recurrent phenomena, found schools and archives, inherit scholarly lineages, train successors, fight over evidence, and build useful theories that remain incomplete. A local term is not decorative camouflage for an author axis: it should carry the observations, interests, tools, exclusions, and historical failures that made the model persuasive.

Mystery therefore depends on epistemic distance, not an exposition embargo. The system can know the truth while the world does not. A peasant, Techpriest, prophet, court, cult, and god can each hold different evidence without any of them becoming a ventriloquist for the setting bible.

Open source and spoiler friction

The deep setting is not proprietary secret material. The repository remains open and determined readers may inspect the complete author machinery. The required boundary is informed friction:

• the default Quartz path contains only spoiler-safe, situated material;
• the Obsidian vault root presents a reader-first index plus one explicitly named Spoilers/ section rather than exposing author domains beside Public;
• repository and author-vault entrances warn before linking into author truth or GM dossiers;
• deep material is labelled as deliberate opt-in reading rather than hidden behind false access control;
• search, sitemap, RSS, backlinks, and navigation on the reader site must not surface author-only material accidentally;
• a human may choose spoilers, but a Ghostlight character agent must never receive them merely because unrestricted retrieval found them.

This follows the Dwarf Fortress “Hidden Fun Stuff” model: adults may look behind the curtain, but the normal experience does not force the choice on them.

Ghostlight knowledge and play boundary

Kalsa has one body of canon and multiple derived consumers. Those consumers do not become lore owners.

• World truth: author canon and hidden mechanics used to compile causal limits and resolver affordances.
• Retrieval evidence: linked sources, claims, provenance, and relationships used to assemble grounded context.
• Epistemic projection: the facts, beliefs, perceptions, memories, emotions, misreadings, and available actions one character can receive in one scene.
• Reader projection: spoiler-safe common knowledge and situated sources published by Quartz.

The Persona-side character agent receives only the epistemic projection. The adjudicator receives typed affordances, actual state, and the constraints needed to decide what happens. Resolution uses risk, difficulty, state, approach, and persistent consequence; a player can negotiate a different or less risky approach without negotiating reality itself. Ignorance is architectural, not a prompt instruction asking a model to forget what it just read.

Vault compilation is expensive interpretation rather than bulk storage: parse and link the corpus, build semantic and graph retrieval, reconcile references, lower candidate entities and relations into typed affordances and constraints, then validate the world package. This is why imported vaults are bounded. The Kalsa repository does not yet implement that compiler integration; Ghostlight owns runtime projection and adjudication.

Play-to-canon boundary

Persistent play may expose gaps, create events, and generate candidate material. An improvised outcome becomes shared canon only through the same provenance path as any other adoption: preserve the event receipt and branch context, identify affected owners, test contradictions and downstream consequences, then explicitly adopt, revise, or reject it. Runtime persistence is not automatic franchise canon.

Authorial inheritance

The seed carries visible Jewish historical influence, especially the Divided City's contested sacred geography and the Sarxe response to massacre, dispersal, mobile continuity, and the refusal to present one annihilation target. Future work should acknowledge and transform that inheritance honestly. Neither polity is a one-to-one analogue for Jews, Jerusalem, Israel, or a diaspora community, and no Earth identity may be smuggled into Kalsa as racial essence.

Motivating setting direction

The project has adopted this editorial synthesis: Prometheus repeatedly imposes clean formal structures on human systems; history turns those structures into culture. It was distilled from the seed and later language discussion, not stated by the operator as a verbatim setting law.

The operator-owned language direction is the High-Latin/Vulgar-Latin analogy: formal Lojban inheritance should coexist with natural descendant speech. The more specific bullets below are adopted or candidate elaborations from later deepening work, not direct quotations of operator intent:

• Prometheus imposed Lojban as a supposedly more logical, communicative prestige language alongside rationalized measures and timekeeping.
• High Lojban can survive in law, liturgy, treaties, technical manuals, priesthoods, scholarship, and old machine interfaces.
• Ordinary use should produce regional descendants through simplification, reanalysis, substrate influence, sound change, idiom, ellipsis, metaphor, and social signaling.
• Magical practice should exert domain-specific pressure on speech: prophecy on evidentiality, sorcery on hierarchy, Sarxe life on collective reference, and Analyst practice on technical causality are candidate directions, not yet blanket canon.
• Contact after the Fall should produce trade languages, koine forms, mixed prestige registers, and once-condemned constructions that become normal grammar.
• Techpriests may preserve formulae they only partly understand because surviving machinery still recognizes them.
• Elite performance of “correct” High Lojban may function as class power even when elites speak a descendant language at home.

The durable theme is not “logical language fails.” It is that a designed system becomes alive when people inherit, misuse, repair, ritualize, weaponize, localize, and love it.

Fundamental setting question

Kalsa asks what a world becomes when causality is inhabited by innumerable competing oracles, none privileged, whose different powers perceive and steer possibility through different shapes. Conscious Prophecy is only the most legible form. Luck, Sorcery, Shamanism, Analysis, Channeling, Intuition, Projections, Chaos Beasts, and God hierarchies participate without all producing visions or propositions.

The comparison to a singular prescient ruler is diagnostic: timeline collapse becomes possible when one oracle's observation and response dominate the field. Kalsa normally contains too many partially legible competitors for that global monopoly. Luck and Luck Gods come closest through broad, continuous survival pressure. God hierarchies, conquest, submission, ideological assimilation, prophetic custody, reproductive control, and archive seizure can all attempt to manufacture local timeline capture by reducing independent sources of intervention.

Every deepening wave that touches magic, gods, state capacity, war, law, kinship, trade, migration, scholarship, infrastructure, or play must therefore ask:

• which differently shaped oracles enter the causal system;
• what each can perceive, steer, conceal, or only infer about the others;
• which institution controls observation, material levers, communication, and response;
• whether the arrangement preserves competing futures or routes them through a dominant hierarchy;
• what that coordination protects, what it forecloses, and who bears the transition; and
• how inhabitants explain the result without possessing the author model.

Oracle ecology and timeline capture are author-facing working terms. Living cultures require their own models, evidence practices, scholarly lineages, interests, and characteristic mistakes.

Institutional depth standard

A mature institution should answer, in setting terms:

1. What material or social need does it satisfy?
2. Who possesses authority, and what makes that authority legible?
3. How are members recruited, trained, ranked, paid, disciplined, and replaced?
4. What resources, records, places, tools, and supply chains keep it alive?
5. What formal doctrine does it claim, and what informal practice actually governs it?
6. Who benefits, who pays, who is excluded, and who can refuse or exit?
7. What other institutions constrain, fund, fear, imitate, or parasitize it?
8. How does it differ by class, region, lineage, language, and magical capacity?
9. What historical crisis created its current form?
10. What internal contradiction generates playable conflict now?
11. What changes when a player helps, harms, captures, exposes, or reforms it?
12. Which claims are established canon, attributed belief, present inference, or unresolved design?

Lists of offices and proper nouns do not count as depth by themselves. Depth means causal machinery: authority, reproduction, material support, contested legitimacy, and consequences.

Iterative critique and repair

Each deepening pass must leave an inspectable record:

1. Select one bounded institution, region, practice, or cross-cutting system.
2. Inventory source facts, later additions, inferences, contradictions, and unknowns.
3. Model owner, inputs, outputs, dependencies, beneficiaries, costs, failure modes, and interfaces with neighboring institutions.
4. Critique generic fantasy defaults, missing material constraints, false consensus, absent subordinate perspectives, and consequences that stop at the page edge.
5. Repair the canonical notes through the smallest coherent set of changes.
6. Connect the repair to people, places, events, resources, language, magic, and adventure pressures.
7. Falsify it with at least one hostile or marginal perspective and one historical stress case.
8. Record decisions, provenance, unresolved pressure, and the next highest-value gap in project memory.

A pass is complete only when the repaired material produces more specific downstream consequences without creating duplicate owners or silently erasing the seed.

Provenance and canon

• Preserve the raw exported pages as immutable source witnesses.
• Distinguish seed, adopted, provisional, contested, rumor, and deprecated claims.
• Do not launder an assistant suggestion into canon by copying it without status and provenance.
• Contradictions should become explicit design questions, perspectival disagreements, or deliberate historical uncertainty—not accidental parallel truths.
• Public-facing pages may omit workshop apparatus, but publication filters must never decide canon.

Guardrails

• No generic medieval-Europe paste with renamed nouns.
• No institution without material support, succession, enforcement limits, and an affected population.
• No culture treated as a personality type.
• No magic system isolated from labor, law, warfare, kinship, disability, trade, education, ecology, and state capacity.
• No formal language frozen outside history once ordinary people speak it.
• No repair loop that rewards page count over causal density and usable conflict.
• No agent memory, task ledger, or critique note published as lore by accident.

Current unknowns

• The canonical present date and temporal distance from the Fall.
• Which seed contradictions should remain contested in-world rather than be resolved.
• The knowledge annotations and compiler artifacts Ghostlight will require beyond the physical public/author boundary established here.
• Which revelations are common, learned, restricted, character-specific, or campaign-current in each future play package.
• The final custom-domain binding for the Quartz site.

Transcript evidence map

The governing claims above come from the complete embedded shared discussion, not only its rendered final branch. The decisive operator turns are 292 and 295 (a curated, familiar fantasy preset rather than an AI-worldbuilding vertical), 342 (causal hard-SF/fantasy seed and missing institutional depth), 388 (hidden vocabulary, archaeological reconstruction, and gods with incomplete knowledge), 391 (open-source Hidden Fun Stuff), 394 (vanilla high-fantasy entrance), 400 and 403 (oracular spectacle plus hidden capability limits), 406 and 409 (character projection, typed and dice-based adjudication, and negotiated approach), 416 and 419 (opinionated kernel and world compilation), 423 (Jewish influence), 437 and 440 (Aetheria-level iterative deepening and the vault as an enduring artifact), and 530 (High/Vulgar Lojban inheritance).