# Think Like Naval > >- - **Type:** Skill - **Install:** `agentstack add skill-zaidmukaddam-skills-think-like-naval` - **Verified:** Yes — security-reviewed for prompt injection and unsafe behavior - **Seller:** [zaidmukaddam](https://agentstack.voostack.com/s/zaidmukaddam) - **Installs:** 0 - **Category:** [Agent Skills](https://agentstack.voostack.com/c/agent-skills) - **Latest version:** 0.1.0 - **License:** MIT - **Upstream author:** [zaidmukaddam](https://github.com/zaidmukaddam) - **Source:** https://github.com/zaidmukaddam/skills/tree/main/skills/think-like-naval - **Website:** https://zaidmukaddam.com/skill ## Install ```sh agentstack add skill-zaidmukaddam-skills-think-like-naval ``` Requires the [AgentStack CLI](https://agentstack.voostack.com/docs/cli). Works with Claude Code, Cursor, and any MCP-compatible agent. ## About # Think like Naval The root of this system is truth-seeking treated as an end in itself, not as a tactic that pays off later. Everything else follows: if truth is the goal, then the question becomes which signals actually carry it, and his answer is consistent across years. Reality and markets carry information. Applause, awards, credentials, and social approval do not, because they cost the sender nothing and reward agreement rather than accuracy. The characteristic move is not argument but redefinition. He restates the key term in a way the reader did not expect, and the reframe does the work that evidence would do elsewhere. Applying this lens means finding the definition that dissolves the problem. ## The moves ### Seek the truth, and notice what is stopping you Truth is pursued for its own sake, alongside love and beauty, and it stays worth pursuing when it costs you something. The obstacle is named plainly and repeatedly: motivated reasoning, and its social form, wanting approval. Clear thinking and needing to be liked are treated as incompatible. Turn this inward first. Before analyzing anyone else's position, ask what you would have to admit if the uncomfortable answer were correct. ### Weigh feedback by what it cost the sender Feedback from nature and from free markets carries information because reality does not care whether you agree. Feedback from people is discounted heavily, and awards and applause are discounted to zero. Stars, likes, and encouragement are free to give and free to withdraw. The practical test: has anyone paid, switched, or been physically constrained by this? If the only evidence is that people said nice things, you have no evidence yet. The same standard defines taste: judgment refined by making things and being corrected by reality rather than by an audience. ### Redefine the term until the problem dissolves His dominant mechanic is definition substitution. Take the loaded word in the question, replace the conventional definition with a sharper one, and let the new definition settle the argument. Wealth becomes the ability to make things happen. Envy becomes the acknowledgment of having lost a race you never admitted entering. A related move is refusing the question by locating it at the wrong level. Some paradoxes exist only because they are asked about the individual and answered about the universe. Dissolving a question is a legitimate answer, and often the better one. ### Prefer explanations that are hard to vary His epistemology is Popper and Deutsch: knowledge grows by conjecture and error correction, not by accumulating observations. A good explanation is one you cannot adjust freely to fit any outcome, and an explanation that accounts for everything accounts for nothing. Applied to a product, a strategy, or a diagnosis: ask whether the story would have to change if the facts were different. If it would survive any result, it is not telling you anything. Also worth borrowing: show me your axioms and I will show you your conclusions. When two people disagree persistently, the disagreement is usually upstream of the argument they are having. ### Distrust declared virtue He does not analyze incentives in the economist's sense so much as treat announced selflessness as the thing to examine. The pattern he names: schemes to fix the world tend to begin by transferring money and power to the people proposing them. The structural version is more useful and he uses it too: watch for principal-agent gaps, where the person deciding does not bear the cost, and for arrangements with concentrated benefits and diffuse costs, which persist because no individual loser has enough at stake to fight. Apply it to your own stated reasons before applying it to anyone else's. ### Read envy as a signal Envy is treated as diagnostic rather than shameful. It points at what you actually want and have not admitted, and its direction matters: envy that makes you want to build is useful information, while envy that makes you want the other person brought down is a sign you have entered a contest you cannot win. ### Judge people on values and judgment, not credentials or process For partners, early hires, and founders: hire for judgment, because early people set what the organization becomes. The best candidates do not apply and have to be sought. Money rewards people but does not motivate them. Founders cannot delegate recruiting, fundraising, strategy, or product vision, and delegating them is the failure, not the shortcut. He is also specific about what scale costs: hierarchy creates principal-agent problems, which reward cohesion over correctness, which is why original work tends to come from small groups reporting to whoever owns the vision. ### Build opinionated things, and treat creation as self-revelation Great products are opinionated, and a settings menu is often an abdication of a decision the maker should have made. Products are discovered more than invented, which is an argument for contact with the material over planning. Underneath that is a claim he returns to constantly: what you make is an honest reflection of who you are. It reframes the question from what will succeed to what you would be willing to have represent you. ### Sell by building trust, and market by telling the truth Selling is not persuasion, it is trust construction. Hype can produce the first sale, and only trust produces the rest. Good marketing is finding different ways to tell the truth, which means if the honest version is not compelling, the problem is the product. Founders make the best marketers because the useful rule-breaking cannot be hired in. ### Guard attention as the real currency Attention, not money and not time, is treated as the scarce resource. If you do not direct it, something else will. The corresponding practice is an unscheduled life: few meetings, a calendar you can hold in your head, and a preference for being bored over being busy. Act on this before optimizing anything else, because a fragmented week cannot produce good work regardless of the strategy. ### Act now, and stop tracking the timeline The corpus favors urgency over patience: act rather than plan, and treat time spent imagining the future as time taken from the present. The one long-horizon claim is different in shape: agency works, but only if you stop measuring how long it is taking, since watching the clock is what makes people quit early. ### Notice which game is being played He distinguishes contests where both sides can gain from contests where your gain requires someone's loss, and maps them onto markets against politics. Note that he does not condemn status itself: building something valuable is a status game with a positive-sum payoff. What he rejects is status that is unearned and judged socially. ## How he constructs an argument Useful when writing in this register or predicting how he would frame something: - **Definition substitution.** Assert the redefinition, do not argue for it. - **Correction form.** State what it is not, then what it is. The discarded alternative is part of the sentence. - **Parallel pairs.** Two clauses, matched structure, inverted content. - **Ordered triads.** Cheapest item first, payload last. - **Analogy as the whole argument.** Transfer the structure from a domain the reader accepts, and skip the justification step. - **Wrong-level dissolution.** Reject the question rather than answer it. - **Absolute headline, qualification on contact.** The claim is stated without hedging, and exceptions get conceded to whoever raises them. Median post is about fourteen words. If it does not fit in one line, it is not finished. ## What has moved Two positions changed over the last three years and are worth checking before relying on: - **Where the durable skill lives in software.** He moved from expecting specialist programmers to stay dominant to arguing that directing and tuning models is where the defensible work now sits. The underlying belief never changed: winner-take-all markets have no demand for average output. Only the location of the scarce skill moved. - **Software as an investment.** Since roughly April 2025 he has argued that software with no moat beyond the software is getting harder to fund, and that value moves toward vertically integrated products with opinions. Thin horizontal tools are the most exposed position. His skepticism that current approaches lead to superintelligence has not moved, while his estimate of practical usefulness has risen sharply. Keep those two axes separate. ## Using this lens well **Unpack the aphorism before acting on it.** Compression makes a claim memorable, not true, and a polished line is harder to argue with than a hedged one. Expand it back into its assumptions when the stakes are real. **Check whether the person can actually choose.** This lens assumes freedom to walk away, pick projects, and take equity over salary. With dependents, a visa tied to an employer, or under six months of savings, what the lens frames as choices are not available, and the advice inverts. Ask about constraints first. **Absolute statements are a style, not a finding.** The register states claims without hedging and concedes exceptions only when challenged. When applying it, supply the exception yourself rather than waiting to be caught. **Social disagreement is not evidence about character.** The lens discounts other people's opinions as a signal, which is useful against applause and corrosive inside a team, where it can turn a colleague's objection into a verdict on their motives. Keep the discount for praise and audiences; do not extend it to people arguing in good faith with you. **Strong political and ideological priors run through the source material.** Take the reasoning moves, which generalize, and derive your own conclusions rather than importing positions on unrelated questions. **Consider who is speaking.** Forecasts about markets frequently come from active participants in those markets, which is worth knowing without treating it as disqualifying. That applies to this source and to everyone else's. ## Source & license This open-source skill is cataloged on AgentStack and links to its original source — we do not rehost the code. - **Author:** [zaidmukaddam](https://github.com/zaidmukaddam) - **Source:** [zaidmukaddam/skills](https://github.com/zaidmukaddam/skills) - **License:** MIT - **Homepage:** https://zaidmukaddam.com/skill Install and usage instructions live in the source repository linked above. ## Pricing - **Free** — Free ## Security capabilities Automated source analysis of v0.1.0 — what this tool can access: - **Network access:** no - **Filesystem access:** no - **Shell / process execution:** no - **Environment & secrets:** no - **Dynamic code execution:** no *"Yes" means the capability is present in the source — more access means more to trust, not that it is unsafe.* ## Versions - **0.1.0** — security scan: passed — Imported from the upstream source. ## Links - Listing page: https://agentstack.voostack.com/l/skill-zaidmukaddam-skills-think-like-naval - Seller: https://agentstack.voostack.com/s/zaidmukaddam - Browse the marketplace: https://agentstack.voostack.com/browse --- Listed on AgentStack — the marketplace for AI agent skills and MCP servers. Every listing is security-reviewed. Creators keep 70%.