Keyboard shortcuts

Press or to navigate between chapters

Press ? to show this help

Press Esc to hide this help

⌈ BARDO DESIGN SYSTEM ⌋

Math-to-Metaphor Translation · v1.0

Making Complex Mathematics Intuitive

Cross-references: 02-widget-catalog.md (33-widget TUI component library with detail_tier property), ../rendering/02-visualization-primitives.md (13 visualization primitives for terminal data displays), ../perspective/00-nooscopy.md (Nooscopy: Golem-initiated decision approval modal, also known as Golem perspective overlay), 00-screen-catalog.md (29-screen summary with screen placements)

Reader orientation: This document specifies the math-to-metaphor translation layer, mapping every mathematical construct in the Golem’s prediction and evaluation systems to a user-facing metaphor. It belongs to the interfaces/screens layer. The user never needs to understand persistent homology or KL divergence; the math determines what happens on screen while the metaphor tells the user what they need to know. Key concepts: Golem (a mortal autonomous DeFi agent), the Spectre (dot-cloud creature whose visual behavior encodes mathematical signals), Grimoire (persistent knowledge store), and CorticalState (the 32-signal perception surface feeding all widgets). For unfamiliar terms, see prd2/shared/glossary.md.


Overview

Every mathematical construct in the system has a user-facing metaphor. The user never needs to understand persistent homology, sheaf cohomology, or KL divergence. The math determines what happens on screen. The metaphor tells the user what they need to know.

This document specifies the translation layer: what each technical term becomes in the user’s experience, how detail hierarchies prevent information overload, and how widgets expose different levels of mathematical depth.


1. Translation Table

Technical TermUser-Facing MetaphorWhat It Actually MeansWidget Context
Bayesian surpriseUnexpectedness scoreHow much a single observation updated the Golem’s beliefs (KL divergence between prior and posterior)FlashNumber flash intensity scales with surprise magnitude
Persistence diagramMarket structure mapScatter plot of topological features showing which market structures are stable vs. transientPersistenceDiagramWidget renders this as a star chart
Wasserstein distanceStructure change rateHow much the market’s geometric structure changed between successive observationsWassersteinRiver ribbon width encodes this value
KL divergenceBelief shiftThe information-theoretic distance between what the Golem expected and what actually happenedSurprise burst cinematic triggers above 3.0 nats
Betti numbersFeature countHow many distinct clusters (β₀), cycles (β₁), and voids (β₂) exist in the observation spaceFlashNumber inline counters on Chain Intelligence tab
Somatic markerGut feelingA learned body-level response to a pattern, formed from past emotional outcomes associated with similar situationsSomaticMarkerPanel shows active markers with strength bars
Hedge weightsSignal trust levelsHow much the system trusts each specialist module’s contribution to the final decisionConfidenceBar per specialist in MIND > Pipeline
Ergodicity gapHidden riskThe difference between time-average and ensemble-average returns, measuring whether past performance predicts future results for THIS specific GolemMortalityGauge variant on stochastic clock detail
Sheaf consistencyAgreement scoreWhether the Golem’s short-term, medium-term, and long-term views of the market agree with each otherMortalityGauge variant; screen color harmony in Portal mode
Phi (integrated information)CoherenceHow tightly the Golem’s subsystems are communicating, a measure of cognitive integrationSpectre cloud cohesion; Φ gauge arc in MIND > Pipeline
UMAP projectionSimilarity mapA 2D projection that places similar things near each other in visual spaceSimilarityLandscape topographic heatmap
Vietoris-Rips complexObservation neighborhoodThe geometric structure built by connecting nearby market observationsNot rendered directly; feeds the persistence diagram
Hodge LaplacianConsistency detectorThe mathematical operator that finds contradictions between timescalesNot rendered directly; feeds the agreement score
HDC cosine similarityPattern resemblanceHow closely two encoded patterns match in the Golem’s memory spaceProximity in SimilarityLandscape; clustering in ForceGraph
Causal DAGCause-effect chainDirected graph of discovered relationships where A reliably precedes BCausalGraphMinimap; ForceGraph in MIND > Grimoire

2. Three-Tier Detail Hierarchy

Every signal in the system renders at one of three tiers. The tier controls how much mathematical detail is exposed. Users navigate between tiers by drilling into widgets (Enter) or backing out (Esc).

Tier 1: Overview (Metaphors Only)

Purpose: Ambient awareness. Glance-readable. No numbers, no labels, just color, rhythm, and atmosphere.

Visual register: ROSEDUST palette shifts, heartbeat rhythm, Spectre behavior, noise floor density. The user develops intuitive pattern recognition after hours of peripheral monitoring.

Signal FamilyTier 1 PresentationNormalAlarming
TopologyWassersteinRiver in sidebarThin quiet streamWide bright flood
ConsistencyScreen color harmonyUnified paletteFractured color temperatures
Coherence (Phi)Spectre cloud cohesionTight, breathing normallyLoose, dots drifting apart
Gut feelingsSpectre body zone disturbancesCalmGut contractions, unfocused eyes
TA patternsBackground activity levelOccasional dim flickersCascade of activations
SurpriseFlash frequencyRareFrequent bright flashes

Screen placement: HEARTH > Overview (Tab 1). The default idle screen, designed for a second monitor.

Tier 2: Drill-Down (Metaphor + Number)

Purpose: Directed investigation. The user noticed something at Tier 1 and wants to understand what is happening.

Visual register: Named widgets with labels, sparklines, gauge values. Metaphor labels appear alongside numeric values. Categories are labeled in plain language.

Signal FamilyTier 2 PresentationScreen
Topology“Market structure map” with feature counts, “Structure change rate” ribbon with numeric annotations, regime tagFATE > Chain Intelligence
Consistency“Agreement score” gauge (0-100%), contradiction count, disagreeing timescales namedSOMA > Epistemic, FATE > Chain Intelligence
Coherence“Coherence” waveform with numeric value, weakest link identified by nameMIND > Pipeline
Gut feelings“Active gut feelings” list with strength bars, pattern name, affect directionFATE > Technical Analysis, SOMA > Affect
TA patternsSignal battery with per-pattern states, “trust level” overlaysFATE > Technical Analysis
Surprise“Unexpectedness” distribution, per-element scoresFATE > Oracle Overview

Interaction: Lock into any Tier 2 pane to navigate elements. Each element is selectable. Enter opens a Tier 3 modal.

Tier 3: Deep-Dive (Full Math)

Purpose: Complete mathematical detail for users who want to see the raw computation. Debugging, verification, parameter tuning.

Visual register: Modals with data tables, formulas, computation traces. This is where the math lives for those who seek it.

Signal FamilyTier 3 Modal Content
TopologyFull persistence diagram with (birth, death) coordinates, Wasserstein distance computation, VR complex stats (simplex count, max dimension, optimal ε)
ConsistencyHodge Laplacian eigenvalues, per-edge disagreement vectors, restriction map parameters, raw observation vectors per timescale
CoherenceMutual information matrix (7x7 subsystem pairs), all 63 bipartition scores, MIB history, signal grouping table
Gut feelingsMarker binding vectors (HDC), retrieval SNR, somatic map saturation, full PAD decoding, prototype similarity
TA patternsPattern hypervector details, fitness history, mutation log, crossover parents, attention auction bid history
SurprisePrior/posterior parameter values, conjugate model type, KL divergence computation, sufficient statistics history

Access: Enter on any element in a Tier 2 pane. Modal system supports infinite nesting (see 03-interaction-hierarchy.md). The user can drill from a topology feature to the observation vectors forming the simplex, to the specific gamma tick, to the market data at that tick.


3. Widget detail_tier Property

Every widget that renders research-derived signals carries a detail_tier property that controls which tier of information it currently displays. This property is part of the widget’s state, not its configuration. It changes based on user navigation.

#![allow(unused)]
fn main() {
pub enum DetailTier {
    /// Metaphors only. Color, shape, rhythm. No text labels, no numbers.
    /// Used in ambient/overview screens and Portal mode (F4).
    Overview,

    /// Metaphor label + numeric value. Named categories, gauge fills,
    /// sparkline traces. The standard operating mode for most screens.
    DrillDown,

    /// Full mathematical detail. Data tables, formulas, raw values.
    /// Activated by Enter on a DrillDown element, rendered in modals.
    DeepDive,
}

pub trait TieredWidget: Widget {
    fn detail_tier(&self) -> DetailTier;
    fn set_detail_tier(&mut self, tier: DetailTier);

    /// Returns the metaphor label for this widget's data.
    /// Used in Tier 1 and Tier 2 rendering.
    fn metaphor_label(&self) -> &str;

    /// Returns the technical label for Tier 3 rendering.
    fn technical_label(&self) -> &str;
}
}

Tier transitions

Navigation actionTier change
Navigate to HEARTH > OverviewAll visible widgets → Overview
Navigate to a detail screen (FATE > Chain Intelligence, etc.)Widgets → DrillDown
Enter on a DrillDown widgetOpens modal at DeepDive tier
Esc from DeepDive modalReturns to DrillDown
F4 (Portal mode)Forces Overview tier on all non-modal widgets
F4 + F2 (Portal + Golem Perspective)Overview tier + floating annotations from CorticalState

Implementation notes

  • Tier 1 widgets render using PAD-modulated color and rhythm only. Text rendering is suppressed. The widget’s metaphor_label() may appear as a dim text_ghost watermark.
  • Tier 2 widgets use metaphor_label() as the visible label, not technical_label(). “Agreement score: 72%” rather than “Sheaf consistency: 0.72”.
  • Tier 3 modals use technical_label() as headers. “Sheaf Consistency — Hodge Laplacian Eigenvalues” is the full formal heading.
  • Widgets that do not implement TieredWidget default to DrillDown behavior on all screens.

4. Metaphor Consistency Rules

The same mathematical phenomenon always produces the same visual metaphor, regardless of which screen or context it appears in.

Mathematical patternVisual metaphorExamples
Fragmentation (increasing separation)Spatial dispersalHigh β₀ scatters PersistenceDiagram points. Low Φ disperses Spectre cloud. Lifecycle degradation scatters fringe dots.
Contradiction (conflicting signals)Color temperature splitSheaf inconsistency splits screen palette. Somatic conflict splits gut/chest response. TA vs. Daimon disagreement flashes warning.
Understanding (new connection)ConvergenceCausal chain discovery draws fragments together. Dream integration merges replay frames. Consistency recovery reunifies palette.
Danger (threat detection)ContractionNegative somatic markers contract gut zone. Low health constricts position displays. Crisis register tightens borders.
Health (stable operation)Breathing rhythmCalm heartbeat, steady cloud oscillation, smooth waveforms, regular atmospheric noise.

These mappings are not decorative. They are computed from specific data sources. The crack in the screen IS the β₀ transition. The color split IS the consistency score. The cloud dispersal IS the Φ decline. The metaphor and the math are the same thing rendered at different resolutions.


5. Invertibility Guarantee

Every visual metaphor in the system is traceable back to its data source. The user who sees a wide river (Wasserstein distance spike) can press Enter to see the numeric distance value (Tier 2), then Enter again to see the full persistence diagram comparison (Tier 3). At no point does the system discard information. It layers presentation over it.

No hidden state: every signal that affects the user’s view is visible somewhere in the system. If topology_signal is modulating the Spectre’s surprise behavior, the topology_signal value is visible on FATE > Chain Intelligence. The user who notices something strange can always trace it.


⌈ the math is real. the metaphor is honest. the user decides how deep to look. ⌋ ║▒░ BARDO ░▒║