tgame/docs/19-concepts-player-and-ui.md

Player & UI Concepts — 2026-05-12 brainstorm

Continuation of 17-foldins and 18-expedition-engagement. Format is a scannable concept glossary — one short paragraph per idea. Categorize / prioritize later.

Player identity

Omnipresent observer. The player has no avatar, no name, no in-fiction position. They watch the cast, occasionally intervene. Closest reference is Black & White's spatial framing (third-person, disembodied) — but not the deity-role implications. The cast doesn't worship or defer to the player.

Player as character volition. When the player makes a decision for Brom, they are Brom in that moment. No "override" friction, no compliance dialogue, no resentment system. The character just does it. Player intervention is a lever on the cast's mind, not a command from above.

Three engagement levels are all valid. Lean-in (read every bark, weigh every fork). Check-in (drop in periodically, intervene on what matters). Walk-away (launch and read the report). The party self-completes regardless; the player picks their depth per session. Honors the morning-after test.

UI pivot

Context-switched screens, not persistent HUD. Each screen is a complete composition: hall, workshop, world map, expedition view, library, etc. No always-on portrait bar across all screens — UI elements belong to the screens they serve. Notification glow on the hall scene signals "there's something to look at" without forcing constant visibility.

Eye-of-the-Beholder portrait bar. Persistent UI element within the expedition view. 4–6 framed portraits along the bottom: status overlays, expression shifts (4 fps), speech banners. The portrait is the character to the player's mental model. Same vocabulary as the hall's roster wall.

Journey graph (Oregon Trail × FTL). Expedition view's central artifact. Branching node graph: round = flavor, square = decision, diamond = leg-end. Party-marker advances node-to-node, stepwise. Branches visible — decisions are forks, geographically committing. Past nodes stay on the graph as visited history.

Not a card game. Cards-the-genre signifier is overloaded. The game uses illustrated artifacts (portraits, journal pages, dispatches, maps, item cards in inventory) but never has cards "in hand," a deck to draw from, or a discard pile. No mana, no hand management. The art format is honest; the gameplay framing is not card-game.

The "alive hall" stays first-person. Existing 09-art-style discipline preserved. In-fiction reframe: the observer hovers close to the hall (peers into it as a doll's house) rather than standing in it as guildmaster. Minor revision to 08-ux-ui language needed.

Voice & content

Character voice is the personality channel. The latent matrix (the MBTI-adjacent 4-axis system from 18-expedition-engagement) expresses through how each character speaks. Bark library + journal voice + decision phrasing all flow from the character's latent profile. Players develop favorites by mouthfeel, not by stats.

Mixed addressing dynamic. Characters chatter among themselves, narrate their own observations, occasionally address the player during decisions. Loose, not rigid. Texture difference between flavor (overheard) and decisions (directed) is natural.

Chronicler is a party role. A character with the right traits (Lettered, Scholar, Cartographer) chronicles the run — produces journal entries. No chronicler in the party = sparse, garbled updates. Knowledge of the run is a capability gate, not a free service. The chronicler's voice shapes the journal's prose style.

Mechanism

Auto-quests with loot tables. The whole expedition system reduces to a quest catalog with conditional triggers and weighted loot tables. Each auto-quest is a small state machine — nodes, decisions, branches, outcomes. Bounded, simulatable, AI-batchable. 90s CRPG mechanism dressed in modern surfacing.

In-character play is rewarded by access, not bonus. No "stay in character +10" score. Decisions just route which content branches actually fire. Brom-played-Bold walks into the fight chain; Brom-played-Cautious walks past it. The triumphant-fight loot table doesn't roll if there was no triumphant fight. The system doesn't measure coherence; it routes content through actual decisions.

Out-of-character runs feel thinner, not punished. The texture flows either way — a Cautious-Brom run still has flavor and outcomes, just less signature texture. Player learns experientially that letting characters be themselves produces richer runs. No tooltip teaches it.

Sim-tractable shape. The Python sim (17-foldins § Data pipeline) can now model expeditions concretely: party + decision policy + quest catalog → run outcomes (drops, character state, narrative arcs). Monte Carlo runs measure content variety, loot balance, in-character outcome gaps. Tunable before game code exists.

Content envelope

Quest catalog target. ~200–500 auto-quests across all zones/tiers for v1. ~5 nodes each, 2–3 branches per fork. Tagged with axis affinities, capability gates, zone+leg constraints. Loot tables hang off the existing materials taxonomy.

Bark catalog target. ~50–100 lines per character, latent-vector-shaped, context-tagged (route start, encounter, decision-pending, ration-low, wounded, etc.). AI-batched per character at crystallization.

Illustration target. ~1000 quest-node images, AI-batched against the palette/dither pipeline. Plus per-character expression sets. Plus per-zone backdrops.

Glossary format trial

This whole document is the proposed format for the broader idea glossary. One concept per blurb. Short. Cross-linkable. Sortable into categories (must-have / nice-to-have / parking-lot / future). If this format works, the next move is to back-fill the existing 00–18 docs by lifting their concept anchors into a single index file.