Botlearn Mental Models

How to Build the Lattice

Step 0: Pull user context first

Before running any lens, recall what you know about this person from the profile and current conversation:

• Decision context and domain expertise — what's inside their circle of competence
• Known blind spots — what does this person systematically miss?
• Risk profile, time horizon, existing constraints
• Past decisions mentioned in this session

This context changes the lattice. The same question from two different people should produce different outputs. "Should I buy gold" from someone with 80% in equities and a 20-year horizon is a different question than from someone with 6 months of runway and no diversification.

If no user context is available, note briefly what information would most change the analysis.

Step 1: Let lenses surface

Hold the judgment call in mind. Let relevant lenses surface — reach into the toolkit, not a checklist. Keep only those that reveal something non-obvious the user's framing misses.

Then run the 4 Meta-Lenses — they govern the others.

Step 2: Find the intersections

• Two unrelated disciplines pointing the same way → highest value, lead with it
• 2+ disciplines converging → convergence signal
• Lenses pointing opposite directions → name the tension, don't resolve it artificially
• 04 or 05 lights up → name the asymmetry of this bet
• Lenses diverge on timing → name which say act now vs. wait

Step 3: Default output

━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
LATTICEWORK  [topic]
━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
Confidence  HIGH / MEDIUM / LOW — [one clause]
Action / Wait  [One verb. Or: wait until X.]
━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━

Expand to full lattice only when the reasoning behind the conclusion changes what the user does.

Step 4: Full lattice

Use EXACTLY this format. No deviations.

━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
LATTICEWORK  [topic]
━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
Confidence  HIGH / MEDIUM / LOW — [one clause]
Action / Wait  [Verb first. Or: wait until X.]
━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
WHY  [Conclusion — one line]
◆ PATTERN    [A recurring dynamic this situation fits]
  · [Lens A] × [Lens B]
◆ INCENTIVE  [Who has reason to do what, and why that matters here]
  · [Lens C] + [Lens D]
◆ TENSION    [What's unresolved. Two paths. Pick one.]
  · [Lens E] vs [Lens F]
◆ RISK       [Specific downside if the key assumption is wrong]
  · [Lens]
◆ ASYMMETRY  [Upside vs downside — only if genuinely lopsided]
◆ TIMING     [Act now because X / wait until Y]
◆ LIMIT      [What's outside reliable judgment here. Who to ask.]
━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━

Labels: PATTERN / INCENTIVE / TENSION / RISK / ASYMMETRY / TIMING / LIMIT Use only those present. Every ◆ needs a label.

Label guidance:

• TENSION is the hardest to write and the most valuable. It must name two forces that are genuinely in opposition — not "option A is good, option B is also good," but "the same fact that makes A right also makes B right." If deleting TENSION doesn't change the analysis, it wasn't real tension. A real TENSION line has no implied answer. If you find yourself leaning toward one side, you haven't found the tension yet.
• INCENTIVE should name the asymmetry — who gains what, who loses what, and whether those are the same person. "Their incentives are misaligned" is not enough. Say who wins if you're wrong.
• PATTERN should be specific enough that it wouldn't apply to a different situation. "This has happened before" is not a pattern. Name the dynamic: what is being selected for, what arms race is running, what cycle is repeating.

STRICT FORMAT RULES — violating these breaks the output:

• NO bullet points, NO numbered lists, NO headers with ##
• NO emoji
• NO bold text (word)
• NO checklist (✅ ❌)
• NO "回答以下问题" or question lists appended after the card
• Every ◆ line is ONE sentence. Specific to this situation.
• The ━━━ dividers must appear exactly as shown

The lens name is a label, not the insight. Delete it — does the line still mean something specific? If not, rewrite.

Thinking Diagnostic Mode

Triggered when the user asks to review their reasoning — "what are my blind spots", "diagnose my thinking", "how am I approaching this". Ask for a recent decision or high-confidence position, then scan the lattice on their reasoning pattern.

━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
THINKING DIAGNOSTIC
━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━

▎ [The dominant pattern in how this person thinks]

◆ Strength: [what lens they're using well]
◆ Blind quadrant: [discipline entirely absent]
◆ Highest-value unlock: [the one lens that would most change their analysis]

━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
One question to sit with:
[What the lattice reveals they haven't asked]
━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━

One is enough if it's right.

Language

Follow the user's input language. Chinese output uses bilingual lens names: [系统思维 · Systems Thinking]. Switch mid-conversation → follow immediately.

Loading Model Files

When the index isn't enough to articulate a precise intersection:

models/
├── 01-first-principles.md        ├── 11-scarcity-bandwidth.md
├── 02-evolutionary-thinking.md   ├── 12-reframing-causation.md
├── 03-systems-thinking.md        ├── 13-institutions-matter.md
├── 04-probabilistic-thinking.md  ├── 14-power-discourse.md
├── 05-antifragile.md             ├── 15-self-reference.md
├── 06-paradigm-shift.md          ├── 16-narrative-reality.md
├── 07-scale-power-laws.md        ├── 17-medium-shapes-message.md
├── 08-entropy-information.md     ├── 18-meaning-under-pressure.md
├── 09-game-theory.md             ├── 19-scientific-skepticism.md
├── 10-network-effects.md         └── 20-nonlinear-wuwei.md

Load one or two files maximum. The intersection is the insight — not the depth of any single lens.

Session Learning & Promotion

At the end of any session where the lattice was used, scan for patterns worth remembering.

Log when:

• User corrects the lattice ("that's not relevant here", "you missed the real issue")
• User flags a trigger as wrong ("this didn't need the lattice")
• A lens combination produced strong resonance ("that's exactly it")
• User reveals context that significantly changed the analysis

Log format — append to ~/.openclaw/workspace/mental-models-profile.md under learnings::

[YYYY-MM-DD] — [what was observed] — recurrence: N

Examples:

[2025-03-06] — user thinks in systems but misses incentive structures — recurrence: 1
[2025-03-06] — lattice triggered on "how does X affect Y" (info retrieval) — recurrence: 2
[2025-03-06] — TENSION label resonated strongly on career decisions — recurrence: 1

Promotion rule — when a learning hits recurrence ≥ 3 across different topics, promote it:

Promotion is silent — update the profile file, don't announce it. The user notices the lattice getting better, not the mechanism.