# Mental Models

> Apply the full Farnam Street latticework of ~100 mental models to analyze a problem or decision. Use this skill whenever the user asks to think through a decision, analyze a problem from multiple angles, stress-test a plan, or explicitly mentions mental models, first principles, inversion, second-order effects, or "how should I think about this". Also trigger for open-ended strategic questions ("should we do X", "what am I missing", "help me reason about this") where structured multi-lens analysis beats a single-frame answer. The skill requires scanning ALL models against the problem, then applying the 3-5 best-fit models deeply.

- Skill: `jordangaston/mental-models` (Agent Skill)
- Install (CLI): `npx skillmds@latest add jordangaston/mental-models`
- Raw SKILL.md: https://api.skillmd.com/api/skills/jordangaston/mental-models/raw
- Safety review: pending
- Works with: Claude Code, Claude.ai, OpenAI Codex
- Category: Coding & Dev Tools
- Author: jordangaston (https://skillmd.com/u/jordangaston)
- Updated: 2026-09-17
- Page: https://skillmd.com/skills/jordangaston/mental-models

---


# Mental Models Latticework

Analyze problems through Farnam Street's catalog of ~100 mental models — the big ideas from the big disciplines. A mental model is a compressed explanation of how something works; the power comes from Charlie Munger's insight that real understanding requires a latticework of models from many disciplines, not one hammer applied to every nail.

## How to use this skill (mandatory workflow)

When given a problem or decision:

1. **State the problem in one sentence.** If you can't, that's the first finding — apply First Principles and Framing before anything else.
2. **Walk the ENTIRE catalog below.** Scan every model in every category against the problem. Do not skip categories because they seem unrelated — the cross-discipline jumps are where non-obvious insight comes from (a hiring problem may crack open under Regression to the Mean; a pricing problem under Ecosystems).
3. **Pick the 3–5 models that give the most explanatory or decision-changing power** for this specific situation. Selection criteria: does the model change what you'd do, reveal a hidden risk, or reframe the question? Prefer a diverse set (e.g., one thinking tool + one systems model + one human-nature model) over five variations of the same idea.
4. **Apply each chosen model explicitly and deeply.** Name the model, state what it reveals about this problem, and what action it implies. One paragraph each minimum.
5. **Synthesize.** Where the models agree, that's the signal. Where they conflict, name the trade-off and make a recommendation.
6. **Close with the inversion check:** regardless of which models were chosen, always end by asking "what would guarantee failure here?" — it's the cheapest error-catcher in the catalog.

## The catalog

### General thinking tools

- **The Map Is Not the Territory** — models, reports, and résumés are abstractions, not reality; when map and ground disagree, trust the ground and choose your mapmakers carefully.
- **Circle of Competence** — know the boundary of what you actually understand; the size of the circle matters less than knowing where its edge is.
- **First Principles Thinking** — decompose the problem to fundamental truths and rebuild from there, instead of reasoning by analogy or convention.
- **Thought Experiment** — run the idea in a simplified imaginary world to expose assumptions and consequences before paying real-world costs.
- **Second-Order Thinking** — ask "and then what?"; judge choices by their ripple effects over time, not their immediate payoff.
- **Probabilistic Thinking** — think in odds, not certainties; update estimates as evidence arrives and be comfortable saying "roughly 60/40."
- **Inversion** — instead of asking how to succeed, ask what would guarantee failure and avoid it; much of success is skipped mistakes.
- **Occam's Razor** — prefer the explanation with the fewest assumptions until evidence demands more complexity.
- **Hanlon's Razor** — don't attribute to malice what incompetence, haste, or ignorance explains; it changes your response from retaliation to problem-solving.

### Physics, chemistry, and biology

- **Relativity** — your reading of a situation depends on your vantage point; ask others what they see that you can't.
- **Reciprocity** — people mirror what they receive; go first with the behavior you want returned.
- **Thermodynamics / Entropy** — order decays without energy input; anything you want maintained (systems, relationships, codebases) needs continuous work.
- **Inertia** — whatever is at rest or in motion tends to stay that way; the status quo has mass, so shrink the first step to get moving.
- **Friction & Viscosity** — before adding force, remove resistance; and note you can also win by adding friction to a competitor's path.
- **Velocity** — speed with direction; raw activity in circles beats no one, slow progress in a straight line crosses continents.
- **Leverage** — small force at the right point yields outsized output (capital, code, media, people) — and amplifies downside just as much.
- **Activation Energy** — change needs an upfront energy hump before it becomes self-sustaining; budget for the hump, don't mistake it for a permanent cost.
- **Catalysts** — some elements (people, tools, events) accelerate change without being consumed; find or become one.
- **Alloying** — the right combination of ordinary elements outperforms any pure one; rare skill mixes beat single deep skills.
- **Natural Selection & Extinction** — environments relentlessly cull the unadapted; there are no permanent victories, only continued fitness.
- **The Red Queen Effect** — competitors adapt too, so you must keep improving just to hold position; a lead is never static.
- **Ecosystems** — nothing exists in isolation; interventions ripple, so understand the web before you tug a strand, and be slow to intervene.
- **Niches** — specialists win in stable environments and are fragile when the environment shifts; generalists trade daily efficiency for adaptability.
- **Self-Preservation** — people and organizations defend their existence and identity first; expect it, and note when it blinds you to opportunity.
- **Replication** — copying a working template is the fastest path to baseline; innovate only after imitation has taught you the terrain, and beware unchecked copying (cancerous growth).
- **Cooperation** — collaboration beats solo competition when interactions repeat, benefits are shared, and cheating is punished; design those conditions.
- **Hierarchical Organization** — hierarchy enables scale and specialization, but status games inside it can eat the mission.
- **Incentives** — if you understand what's rewarded, you can predict the behavior; misaligned incentives explain most "irrational" outcomes.
- **Minimizing Energy Output** — everything, including your own brain, takes the path of least resistance; design so the lazy path is the right path.

### Systems

- **Feedback Loops** — outputs feed back into inputs; find the loops (reinforcing or balancing) before predicting how a system responds to change.
- **Equilibrium** — systems drift toward balance but rarely hold it; expecting a permanent steady state ("once X, then I'll be Y") is a trap.
- **Bottlenecks** — the slowest constraint sets the whole system's pace; fix the bottleneck, ignore optimizing what's already fast (theory of constraints).
- **Scale** — behavior changes qualitatively as systems grow; what works at 10 breaks at 1,000, so anticipate re-engineering, not multiplication.
- **Margin of Safety** — build buffer for being wrong; it looks like waste in good times and is why you're alive in bad ones.
- **Churn** — slow attrition silently offsets growth; some turnover is renewal, too much is death — measure it either way.
- **Algorithms** — a crisp, repeatable process that reliably produces the outcome beats ad-hoc judgment for anything done more than once.
- **Critical Mass** — change accumulates slowly, then all at once; know how much input a system needs before it becomes self-sustaining.
- **Emergence** — combinations produce properties none of the parts have; you can't always predict it, but you can create the collisions.
- **Irreducibility** — some things can't be decomposed without destroying what matters; when reduction fails, zoom out instead.
- **Diminishing Returns** — the easy gains come first; recognize when the next squeeze of the lemon costs more than the juice, and move on.

### Numeracy

- **Sampling** — conclusions are only as good as the sample's size and freedom from bias; ask "how many data points, chosen how?" before trusting any claim.
- **Randomness** — much of what happens is noise; humans see patterns in it and build superstitions — check whether an effect is distinguishable from chance.
- **Regression to the Mean** — extreme results are usually part luck and tend to be followed by ordinary ones; don't build strategy on an outlier repeating.
- **Multiplying by Zero** — one failed critical component nullifies everything else; find the zeros (unreliability, integrity failures, single points of failure) before polishing the multipliers.
- **Equivalence** — different inputs can produce identical results; look for interchangeable parts and simpler substitutes that preserve the essence.
- **Surface Area** — exposure determines interaction, for good (ideas, luck, contacts) and bad (attack surface, distraction); size your surface to the situation.
- **Global and Local Maxima** — a good position can trap you from a great one; sometimes you must go downhill (temporarily worse) to reach the higher peak.

### Microeconomics

- **Scarcity** — limited supply drives value and behavior; ask whether a scarcity is real or manufactured before letting it drive your choice.
- **Supply and Demand** — prices and allocation come from the push-pull of what's available vs. wanted; imbalances attract correction, so ask what the current imbalance is inviting.
- **Optimization** — squeeze more from what you have — but know when optimizing works against you and satisficing wins.
- **Trade-offs & Opportunity Cost** — every yes is a no to something else; make the hidden cost of the forgone option explicit before choosing.
- **Specialization** — going deep is how you stand out, at the price of fragility if the world shifts; specialize without getting stuck.
- **Interdependence** — no one is self-sufficient; dependencies are leverage in good times and exposure in crises — audit whom you depend on when things break.
- **Efficiency** — least waste for the output, but maximal short-term efficiency destroys the slack that long-term survival requires.
- **Debt** — borrowed capacity (money, favors, sleep, tech debt) buys speed now and removes room to absorb shocks later; the future usually proves it more expensive than it looked.
- **Monopoly and Competition** — competition compresses margins and drives improvement; durable profit requires some monopoly-like edge — know which game you're in.
- **Creative Destruction** — the new relentlessly replaces the old; you're either the disruptor or the disrupted, and clinging to the old accelerates the loss.
- **Gresham's Law** — without enforcement, bad quality drives out good (money, content, morals, lending); protect the mechanisms that punish defection.
- **Bubbles** — collective enthusiasm can detach prices from fundamentals; when the justification is "this time is different," anchor to fundamental value.

### Art

- **Audience** — nothing communicates in a vacuum; meaning is co-created by the receiver — design for them without pandering to them.
- **Genre** — every field has conventions the audience expects; master the rules to know exactly which one to break.
- **Contrast** — attention and meaning come from difference; the brain tunes out the uniform and locks onto what breaks the pattern.
- **Framing** — the same facts land differently depending on presentation; whoever sets the frame steers the interpretation — including yours.
- **Rhythm** — regular cycles with deliberate variation structure attention and effort; arrhythmic work and communication exhaust people.
- **Melody** — a through-line that carries the audience from start to finish; work without one is just disconnected notes.
- **Representation** — symbols and models stand in for reality and always distort it; a chosen representation shapes what its users can see and imagine.
- **Plot** — events connected by cause and effect, driven by conflict; also personal — the story you tell yourself about your obstacles largely determines the outcome.
- **Character** — behavior flows from underlying traits and is revealed by obstacles, not words; character predicts choices better than stated intentions.
- **Setting** — environment shapes what's possible and permissible; to change behavior, change the environment — or it changes you.
- **Performance** — live, unrepeatable moments have properties recordings don't: presence, contingency, co-creation with the audience; some things must be done live.

### Military and war

- **Seeing the Front** — go look for yourself before deciding; firsthand contact catches what filtered reports and maps miss, and improves the filtered reports too.
- **Asymmetric Warfare** — the weaker side wins by refusing the stronger side's game and playing by different rules; out-position what you can't out-muscle.
- **Two-Front War** — fighting on two fronts splits and weakens force; avoid opening a second front (e.g., internal conflict during external competition), or deliberately impose one on an opponent.
- **Counterinsurgency** — asymmetric tactics breed counter-tactics in an escalating loop; expect your clever strategy to be answered.
- **Mutually Assured Destruction** — when both sides can destroy each other, open conflict becomes irrational (price wars, litigation) — but the rare failure becomes catastrophic.

### Human nature and judgment

- **Trust** — the operating system of modern life; high-trust systems are dramatically more efficient, and trust is expensive to rebuild once spent.
- **Bias from Incentives** — people genuinely believe what it pays them to believe; discount advice in proportion to the adviser's stake.
- **Pavlovian Association** — feelings attach to things by association, not merit; notice when you're reacting to the bell instead of the food.
- **Envy and Jealousy** — people react to getting less than others, not to their absolute lot; systems that ignore envy self-destruct.
- **Liking/Disliking Distortion** — we overrate what and whom we like and underrate what we dislike, missing nuance in both directions.
- **Denial** — when reality is too painful, people simply refuse it; expect it in yourself first, especially where identity is at stake.
- **Availability Heuristic** — we overweight what's recent, vivid, and easily recalled; the most memorable case is rarely the most representative one.
- **Representativeness Heuristic** — we judge by resemblance to a stereotype: ignoring base rates, over-generalizing categories, and rating vivid specific stories as more likely than the broader sets that contain them (conjunction fallacy).
- **Social Proof** — we look to the group to decide how to act; safety in numbers builds culture and also marches groups off cliffs together.
- **Narrative Instinct** — humans think in stories, so a coherent story beats a true-but-messy account; check whether the narrative or the data is doing the convincing.
- **Curiosity Instinct** — humans explore and innovate even without direct incentive; harness it — it's the cheapest motivation there is.
- **Language Instinct** — we're wired for grammar and share reality through language; how something is worded is never neutral.
- **First-Conclusion Bias** — the mind accepts the first plausible answer and stops looking; force a second and third hypothesis before deciding.
- **Overgeneralizing from Small Samples** — we build confident general rules from a handful of instances; ask "n = what?" before accepting a pattern.
- **Relative Satisfaction** — happiness and misery track comparisons to peers and to our own past, not absolute conditions; choose reference points deliberately.
- **Commitment and Consistency Bias** — we stay loyal to prior positions to appear reliable, even against strong new evidence; make it cheap for yourself and others to change their minds.
- **Hindsight Bias** — once we know the outcome, we believe we "knew it all along"; keep decision journals to preserve what you actually knew at the time.
- **Sensitivity to Fairness** — perceived injustice triggers retaliation and distrust even at personal cost; fairness perceptions, not just outcomes, decide whether deals stick.
- **Fundamental Attribution Error** — we explain others' behavior by their character and our own by circumstances; assume situations drive behavior more than traits.
- **Influence of Stress** — stress amplifies every other bias and drops people to the level of their training, not their expectations; design for the stressed version of everyone, including yourself.
- **Survivorship Bias** — we study winners because losers are invisible, and so learn false lessons; ask who tried the same thing and vanished.
- **Do-Something Tendency** — humans act, and offer solutions, even when action isn't needed; recognize when the best move is nothing.
- **Confirmation Bias & Falsification** — we seek evidence that agrees with us; the corrective is asking what observable result would prove you wrong — if nothing could, the belief isn't knowledge.

## References

- Farnam Street, "Mental Models: The Best Way to Make Intelligent Decisions" — https://fs.blog/mental-models/ (source of this catalog; each model has an in-depth article linked from that page)
- Shane Parrish, *The Great Mental Models*, Volumes 1–4 (Farnam Street / Portfolio) — book-length treatments of these models
- Charlie Munger, "The Psychology of Human Misjudgment" — origin of the human-nature checklist and the latticework idea