arscontexta 0.6.0

package/reference/use-case-presets.md

@@ -0,0 +1,341 @@
+# Use-Case Presets Reference
+
+Named points in the configuration space derived from methodology traditions and adapted for specific use cases. Each preset is a pre-validated coherent starting point — the derivation engine uses them as anchors, adjusting individual dimensions based on conversation signals.
+
+---
+
+## Three Use-Case Presets
+
+### Research
+
+**Display name:** Knowledge Research
+
+**Optimizes for:** Deep synthesis, cross-domain connection density, long-term knowledge accumulation
+
+**Closest tradition:** Zettelkasten + Cornell processing phases
+
+**Philosophy:** Full automation from day one. All pipeline skills at full depth, all blocks active. The reference implementation.
+
+| Dimension | Value | Rationale |
+|-----------|-------|-----------|
+| Atomicity | 0.8 | One claim per note maximizes composability and cross-source comparison |
+| Organization | 0.3 | Flat-associative with semantic search |
+| Linking | 0.7 | Explicit typed connections + semantic discovery |
+| Processing | 0.8 | Full pipeline: extract, connect, verify, reweave |
+| Session | 0.7 | Fresh per task |
+| Maintenance | 0.6 | Event-driven |
+| Search | 0.8 | Semantic primary |
+| Automation | 0.8 | Full automation from day one |
+
+**Block configuration:**
+
+| Category | Blocks |
+|----------|--------|
+| Always | atomic-notes, wiki-links, mocs, processing-pipeline, schema, maintenance, self-evolution, session-rhythm, templates, ethical-guardrails, helper-functions, graph-analysis |
+| Conditional | semantic-search (if qmd opted in), multi-domain (if needed) |
+| Optional | personality |
+| Disabled | self-space (goals route to ops/) |
+
+**Key settings:**
+- `self_space: false`
+- `qmd: true` (opted in during onboarding)
+- `personality: "neutral-analytical"`
+- `processing_depth: "full quality gates from day one"`
+
+**Extraction categories:** claims, evidence, methodology-comparisons, contradictions, open-questions, design-patterns, design-dimensions
+
+**Key vocabulary:**
+
+| Level | Universal | Research |
+|-------|-----------|----------|
+| Note type | note | claim |
+| Processing verb | reduce/extract | reduce |
+| Connection verb | reflect/connect | reflect |
+| Navigation unit | MOC | topic map |
+| Collection name | notes/ | notes/ |
+| Capture zone | inbox/ | inbox/ |
+
+**Starter MOCs:** `{DOMAIN:domain}-overview`, `methods`, `open-questions`
+
+**Domain-specific failure mode risks:**
+- Collector's Fallacy (HIGH) — source material is abundant
+- Orphan Drift (HIGH) — high creation volume without mandatory connection
+- Verbatim Risk (HIGH) — source material tempts reproduction over transformation
+- MOC Sprawl (HIGH) — topics proliferate in research domains
+- Productivity Porn (HIGH) — meta-system building displaces actual research
+
+**Example user statements:**
+- "I read 5-10 papers a week on [topic] and need to track claims across disciplines"
+- "I'm building a literature review and need to see connections between sources"
+- "I want an academic knowledge base for my dissertation research"
+
+---
+
+### Personal Assistant
+
+**Display name:** Personal Assistant
+
+**Optimizes for:** Life reflection, pattern detection, personal growth tracking, relationship awareness
+
+**Closest tradition:** Custom — moderate processing for pattern detection, warm voice for personal content
+
+**Philosophy:** All skills available and active from day one, adapted for personal use. Full pipeline with personal extraction categories.
+
+| Dimension | Value | Rationale |
+|-----------|-------|-----------|
+| Atomicity | 0.5 | Mixed — per-session reflections, not atomic decomposition |
+| Organization | 0.5 | Light hierarchy for life areas |
+| Linking | 0.3 | Associative connections between reflections |
+| Processing | 0.7 | Full pipeline with personal extraction categories |
+| Session | 0.3 | Continuous context across sessions |
+| Maintenance | 0.4 | Condition-based check-ins |
+| Search | 0.3 | Keyword primary, semantic optional |
+| Automation | 0.6 | Semi-automated with personal touch |
+
+**Block configuration:**
+
+| Category | Blocks |
+|----------|--------|
+| Always | atomic-notes, wiki-links, mocs, processing-pipeline, schema, maintenance, self-evolution, session-rhythm, templates, ethical-guardrails, helper-functions, graph-analysis, personality, self-space |
+| Conditional | semantic-search (if qmd opted in), multi-domain (if needed) |
+| Optional | (none) |
+| Disabled | (none) |
+
+**Key settings:**
+- `self_space: true`
+- `qmd: "user_choice"` (choice during onboarding)
+- `personality: "warm-supportive"`
+- `processing_depth: "full quality gates"`
+
+**Extraction categories:** reflections, relationship-dynamics, goals, habits, gratitude, lessons
+
+**Key vocabulary:**
+
+| Level | Universal | Personal Assistant |
+|-------|-----------|-------------------|
+| Note type | note | reflection |
+| Processing verb | reduce/extract | surface |
+| Connection verb | reflect/connect | find patterns |
+| Navigation unit | MOC | life area |
+| Collection name | notes/ | reflections/ |
+| Capture zone | inbox/ | journal/ |
+
+**Starter MOCs:** `life-areas`, `people`, `goals`
+
+**Domain-specific failure mode risks:**
+- Journal without reflection (HIGH) — capture without pattern detection
+- Cognitive Outsourcing (medium) — pattern detection without human validation
+- Emotional avoidance (medium) — system enables intellectual distancing from feelings
+- Capture without connection (medium) — reflections stay isolated
+
+**Example user statements:**
+- "I want to track my growth and notice patterns in my life"
+- "I need something that remembers what I care about across sessions"
+- "Help me be more thoughtful about my relationships and goals"
+
+---
+
+### Experimental / Build Your Own
+
+**Display name:** Experimental / Build Your Own
+
+**Optimizes for:** User co-design, deep understanding of trade-offs, custom configuration
+
+**Closest tradition:** None — derived from conversation
+
+**Philosophy:** In-depth onboarding with thinking notes surfaced for every design decision. The user co-designs the system.
+
+| Dimension | Value | Rationale |
+|-----------|-------|-----------|
+| Atomicity | null | User-chosen during onboarding |
+| Organization | null | User-chosen during onboarding |
+| Linking | null | User-chosen during onboarding |
+| Processing | null | User-chosen during onboarding |
+| Session | null | User-chosen during onboarding |
+| Maintenance | null | User-chosen during onboarding |
+| Search | null | User-chosen during onboarding |
+| Automation | null | User-chosen during onboarding |
+
+**Block configuration:**
+
+| Category | Blocks |
+|----------|--------|
+| Always | wiki-links, processing-pipeline, schema, maintenance, self-evolution, session-rhythm, templates, ethical-guardrails, helper-functions, graph-analysis |
+| Conditional | atomic-notes, mocs, semantic-search, personality, self-space, multi-domain |
+| Optional | (none) |
+| Disabled | (none) |
+
+**Key settings:**
+- `self_space: "user_choice"`
+- `qmd: "user_choice"`
+- `personality: "derived_from_conversation"`
+- `processing_depth: "full quality gates by default, user can opt down"`
+
+**Extraction categories:** User-defined during init, guided by examples from research and personal-assistant presets
+
+**Starter MOCs:** User-chosen during onboarding
+
+**Domain-specific failure mode risks:**
+- Configuration paralysis (HIGH) — too many choices without guidance
+- Incoherent configuration (medium) — user choices may violate interaction constraints
+- Under-specified system (medium) — skipping decisions leads to gaps
+
+**Example user statements:**
+- "I want to understand the trade-offs before choosing"
+- "My use case doesn't fit the other presets"
+- "I want to build something custom for [unusual domain]"
+
+---
+
+## Five Methodology Tradition Points
+
+Traditions are named coherence points in the 8-dimensional space. They are reference anchors, not templates.
+
+| Dimension | Zettelkasten | PARA | Cornell | Evergreen | GTD |
+|-----------|-------------|------|---------|-----------|-----|
+| Granularity | Atomic | Coarse | Medium (per-session) | Atomic | Task-sized |
+| Organization | Flat | Hierarchical (4 folders) | Temporal | Flat | Hierarchical (contexts) |
+| Linking | Explicit, bidirectional | Minimal (folder membership) | Implicit (cue columns) | Explicit, contextual | Minimal |
+| Processing | Heavy (formulation + linking) | Light (progressive summarization) | Heavy (5 Rs structured) | Heavy (continuous rewriting) | Light (capture + route) |
+| Navigation | 3-4 tier (emergent hubs) | 2 tier (folder browsing) | 2-3 tier (index-based) | 3 tier (link browsing) | 2 tier (context lists) |
+| Maintenance | Continuous | Condition-based review | Spaced review | Continuous | Condition-based review |
+| Schema | Moderate | Minimal | Moderate (cue/notes/summary) | Moderate | Dense (action metadata) |
+| Automation | Convention | Manual | Convention | Convention | Automation-friendly |
+
+**How traditions relate to presets:** Use-case presets are derived from traditions, adapted for specific domains. Research draws primarily from Zettelkasten + Cornell. Personal Assistant is a custom configuration combining moderate processing with warm personality. Experimental derives everything from conversation, using traditions as reference points during design decisions.
+
+---
+
+## Preset Selection Algorithm
+
+The derivation engine maps conversation signals to the closest preset, then adjusts individual dimensions.
+
+### Step 1: Signal Collection
+
+Listen for signals in the user's description and follow-up answers. Each signal maps to one or more dimensions with a confidence level:
+
+| Signal Type | Examples | Maps To |
+|-------------|---------|---------|
+| Volume indicators | "5-10 papers/week", "2-3 books/month" | Volume projection, navigation, maintenance |
+| Processing verbs | "track claims", "remember reactions", "document decisions" | Granularity, processing intensity |
+| Connection words | "across disciplines", "between projects", "patterns" | Linking philosophy, semantic search need |
+| Frequency indicators | "a few times a week", "whenever I have a session", "after each project" | Maintenance trigger signals |
+| Domain markers | "research papers", "personal growth", "relationships" | Closest preset match |
+| Emotional register | "feel seen", "like a friend", "professional" | Personality dimensions |
+| Design curiosity | "understand trade-offs", "custom", "unusual domain" | Experimental preset signal |
+
+### Step 2: Preset Affinity Scoring
+
+Score the user's signals against each preset. Each matching signal adds affinity:
+
+```
+For each preset:
+  affinity = sum of (signal_weight x match_strength)
+  where match_strength is:
+    1.0 = signal directly matches preset characteristic
+    0.5 = signal partially matches
+    0.0 = no match
+```
+
+**Example:** "I read papers and want to track claims across disciplines"
+- Research: "papers" (1.0) + "claims" (1.0) + "across disciplines" (1.0) = 3.0
+- Personal Assistant: low scores
+- Experimental: low scores (no design-curiosity signals)
+
+Select the preset with highest affinity as the starting point. If no preset scores above 2.0 and the user shows design curiosity, route to Experimental.
+
+### Step 3: Dimension Adjustment
+
+Starting from the selected preset, adjust individual dimensions based on specific signals that diverge from the preset:
+
+```
+For each dimension:
+  if user_signal contradicts preset_position:
+    adjust dimension to signal-indicated position
+    check interaction constraints
+    cascade adjustments to dependent dimensions
+```
+
+**Example:** Research preset selected, but user says "I don't need deep analysis, just capture my reactions"
+- Processing: heavy -> light (user signal overrides preset)
+- Cascade check: light processing + atomic granularity -> WARN (atomic needs processing to recreate context)
+- Resolution: suggest moderate granularity to match light processing, or confirm user wants atomic + light with the trade-off acknowledged
+
+### Step 4: Coherence Validation
+
+After all adjustments, check the final configuration against interaction constraints (see `interaction-constraints.md`). Flag hard violations (BLOCK), warn on soft violations (WARN with compensating mechanisms).
+
+### Step 5: Vocabulary Derivation
+
+Using the adjusted preset as the base vocabulary, override with any domain-native terms the user provided during conversation:
+
+```
+For each universal term:
+  if user provided a domain-native equivalent:
+    use user's term
+  else:
+    use preset's default term
+
+  Verify: would this term feel natural to the user?
+```
+
+---
+
+## Novel Domain Handling
+
+When no preset cleanly matches the user's description, the Experimental preset provides the framework for co-design. Unlike the Research and Personal Assistant presets which start with pre-configured dimensions, Experimental walks the user through each design decision with relevant thinking notes surfaced as context.
+
+### When to Route to Experimental
+
+- No preset scores above 2.0 in affinity scoring
+- User explicitly asks to customize or "build my own"
+- Domain is unusual (competitive gaming, wine tasting, legal case tracking)
+- User shows curiosity about trade-offs and design decisions
+
+### Conversation-Driven Derivation
+
+For each dimension:
+
+1. Surface the relevant thinking note(s) that explain the trade-off
+2. Present the dimension as a spectrum with examples from Research and Personal Assistant
+3. Let the user choose their position
+4. Check interaction constraints after each choice
+5. Cascade adjustments to dependent dimensions
+
+### Novel Schema Fields
+
+Novel domains often need schema fields that no preset provides. Derive them from the domain's characteristics:
+
+| Domain Characteristic | Suggests Schema Field |
+|----------------------|----------------------|
+| Temporal dynamics (things become outdated) | `meta_state: current | outdated | speculative` |
+| Confidence tracking | `confidence: proven | likely | experimental` |
+| Sequential progression | `prerequisites: ["[[concept]]"]` |
+| Entity tracking | `person: Name`, `entity: Name` |
+| Decision history | `superseded_by: "[[newer-decision]]"` |
+
+### Worked Example: Wine Tasting Notes
+
+**User:** "I'm getting into wine and want to track what I taste — flavors, regions, pairings, and which wines remind me of others."
+
+**Affinity scores:**
+- Personal Assistant: 1.5 ("remember things", personal interest)
+- Research: 1.0 ("track", systematic)
+- Experimental: route here (no preset above 2.0)
+
+**Conversation-driven configuration:**
+
+| Dimension | Position | Rationale |
+|-----------|----------|-----------|
+| Atomicity | Moderate | Per-wine notes, not atomic flavor claims |
+| Organization | Flat | Wines cross regions and varietals |
+| Linking | Explicit | Direct connections: "this Barolo reminded me of that Nebbiolo" |
+| Processing | Light | Capture tasting notes, connect to similar wines |
+| Session | Continuous | Low volume, ongoing |
+| Maintenance | Condition-based | Low volume — review when 5+ unconnected tasting notes accumulate |
+| Search | Moderate | Structured fields for filtering + semantic for "wines like this" |
+| Automation | Convention | Light overhead |
+
+**Novel vocabulary:** "tasting note" (not "reflection" or "claim"), "wine library" (not "life area"), "cellar" (not "archive")
+
+**Novel schema field:** `pairing: ["food pairing notes"]` — domain-specific field not in any preset.