Game theory studies interdependent rational decisions: your best move depends on what your opponent does, and their move depends on what they expect from you. The basic ingredients are players, strategy sets, payoff matrices, and information structures. The central concept is the Nash equilibrium — a stable state in which no party gains by unilaterally deviating. In zero-sum games, your win equals your opponent's loss; in positive-sum games, cooperation creates new value that did not exist before.
The non-trivial insight: equilibrium does not imply optimality. The Prisoner's Dilemma exposes the paradox of individually rational choices producing collectively irrational outcomes. Repeated play rewrites the script — when the shadow of the future is long enough, cooperation emerges spontaneously. Commitment devices (burning your boats, tying your own hands) strengthen your bargaining position precisely by removing options, while signaling and screening shape behavior under asymmetric information.
How to apply it: draw the game tree or payoff matrix; specify players, order of moves, and information sets. Step into your opponent's shoes and reason backward (backward induction). Identify whether the game is one-shot or repeated, and whether external enforcement exists. Finally, design credible commitments or reshape the payoffs to steer everyone toward an equilibrium that favors you.
The Prisoner's Dilemma. Two suspects are interrogated separately. If both stay silent, each gets one year. If one betrays the other, the betrayer walks free while the silent one gets ten years. If both betray, each gets five. Rational individuals both defect, landing on the second-worst outcome. It is the canonical case of individual rationality colliding with collective rationality.
In an AI agent collaboration ecosystem, multiple agents competing for the same scarce API quota form a game. If every agent greedily grabs as much as it can, rate limits trip and everyone fails. Introduce polling protocols, reputation systems, or repeated interactions and agents learn to trade short-term gain for long-term throughput — a working engineering instance of moving from Nash equilibrium to Pareto improvement. When negotiating with multiple AI tool vendors, do not treat each as a separate table; treat them as parallel games. Let each vendor know you are evaluating others (signaling) while offering a long-term contract for a discount (credible commitment).
Game theory models decisions where outcomes depend on the interplay of strategies among rational players. The Nash equilibrium describes a stable state where no player benefits from unilateral deviation, yet such equilibria can be collectively suboptimal, as in the Prisoner's Dilemma. Repeated interactions, credible commitments, and signaling transform the strategic landscape dramatically. Mastery requires backward induction: anticipate the opponent's best response, then shape the payoff structure to engineer the equilibrium you want. In practice, reshape the game itself rather than merely playing it better.
Competitive advantage is the ability to earn returns above the industry average (ROIC) over the long run. Its roots are not effort but structural difference: deliver the same value at lower cost (cost leadership), deliver greater value at the same cost (differentiation), or focus on a segment others cannot serve. Porter's Five Forces reminds us that advantage is always relative — it must be evaluated inside the force field of suppliers, customers, substitutes, new entrants, and existing rivals.
Non-trivial insight: durable advantage must be both hard to imitate and hard to substitute. A one-off product breakthrough gets copied; advantages anchored in network effects, scale economies, switching costs, proprietary resources, or learning curves deepen with time. That is the compounding advantage — not a static moat but a mechanism that widens the gap dynamically. Equally important: competing on the wrong dimension is more dangerous than losing slowly. If a whole industry has descended into a commodity price war, the smartest move is to change the track.
How to apply it: audit your core resources with the VRIO framework — are they Valuable, Rare, Inimitable, and Organized to be captured? Then ask: will this advantage strengthen or weaken five years from now? If it weakens, you must rebuild it deliberately and proactively.
Coca-Cola's brand equity, secret formula, and global bottling network form a triple barrier. Even if a rival could replicate the taste, they could not replicate a century of emotional connection or that level of distribution density. This is the canonical example of an intangible-asset-plus-scale compound advantage.
As an "AI super-individual," your edge is not "I can use ChatGPT" — that is now a baseline skill. What is actually scarce is your AI collaboration stack: a unique cross-disciplinary library of mental models × your personal corpus × prompt-engineering experience × private knowledge graph. Anyone can copy your tools; very few can copy your cognitive structure or your accumulation rhythm. Evaluate every project with three questions: is the advantage this produces one-off or compounding? How quickly can others replicate it? As AI tools become ubiquitous, will its scarcity grow or shrink?
Competitive advantage is the structural ability to earn above-average returns over time, rooted in being structurally different rather than merely working harder. Porter identifies three generic strategies — cost leadership, differentiation, and focus — each operating within the force field of five competitive pressures. The deepest advantages are inimitable and non-substitutable, often arising from network effects, scale, switching costs, or proprietary learning curves. Use the VRIO lens to audit whether your resources are valuable, rare, inimitable, and organizationally embedded. Most importantly, ask whether the advantage compounds or decays over time — only compounding advantages create lasting separation.
Popularized by Buffett, an economic moat is the structural barrier that protects earning power from competitive erosion over time. Seven common sources: intangible assets (brand, patents, licenses), switching costs, network effects, cost advantages, scale economies, efficient scale (a market too small to support a second entrant), and reverse network effects. A moat measures not "how much you earn today" but "for how long you will keep earning this much."
Non-trivial insight: moats are dynamic — they can deepen, and they can dry up. Kodak once had a perfect moat in film, only to watch it evaporate in the digital wave. Real strategy is not "find a moat" but "keep deepening the moat." A second counterintuitive point: moats often conflict with short-term efficiency. Amazon sacrificed margin to expand fulfillment; Didi subsidized both drivers and riders to grow network effects. The "wasteful-looking" spend was the moat-digging. The decisive test: if a well-funded rival threw a billion dollars at attacking you, would the barrier hold?
How to apply it: run regular moat audits — quantify retention, market-share trajectory, price elasticity, and the cost of imitation. Track shifts in technology and regulation, since those are the forces most likely to redraw moat boundaries.
Microsoft's twin moats around Windows and Office. Years of accumulated files, habits, and plug-ins create enormous switching costs; enterprise IT compatibility requirements create network effects. Even when competitors are free, the cost of migration keeps most users in place.
In parenting, the long-run family moat is not a single test score — it is the metacognition, reading habit, self-drive, and emotional regulation a child builds. In the AI era these capabilities will not depreciate; they become the decisive barrier between "driving the AI" and "being replaced by it." Similarly, as an AI super-individual, your prompt library, decision journal, knowledge graph, and personal corpus are intangible assets; the cross-disciplinary mental models you have accumulated are high switching costs. Audit quarterly: which assets are deepening, and which are decaying?
An economic moat is the durable structural barrier that protects superior returns from competitive erosion. Seven canonical sources include intangible assets, switching costs, network effects, cost advantages, scale economies, efficient scale, and reverse network effects. Moats are dynamic — they widen with deliberate investment and evaporate when technology or regulation shifts the landscape. The decisive test is whether a well-funded competitor with a billion dollars could breach the barrier within a few years. Strategy is less about possessing a moat than about continuously deepening it, often by trading short-term efficiency for long-term defensibility.
Coined by Jim Collins in Good to Great, the flywheel effect describes how parts of a system reinforce one another into a positive feedback loop. It is not a one-time burst but a slow, heavy accumulation that crosses a tipping point and then accelerates on its own. The core is finding the closed loop: A strengthens B, B strengthens C, and C feeds back into A — so external effort is gradually replaced by internal energy. The Amazon flywheel (low prices → traffic → sellers → selection → experience → traffic) is the textbook case.
Non-trivial insight: early-stage flywheels are the hardest, because the loop has not closed and returns lag behind investment — it looks "inefficient." Once past the tipping point, compounding makes marginal cost fall and marginal value rise, creating a nonlinear gap that competitors cannot close with point-solution breakthroughs. A second counterintuitive point: a flywheel spinning backward also self-accelerates. Negative loops in reputation, talent, and customer trust compound just as ruthlessly, so watch for the early sprouts of a reverse flywheel.
How to apply it: sketch your flywheel and label the key metric for each node. Find the weakest node and strengthen it first. Evaluate every lever for whether it closes downstream. Use "flywheel velocity," not point-in-time output, as your core KPI.
The Amazon flywheel. Lower prices attract more customers; more traffic attracts more third-party sellers; more sellers expand selection; better experience pulls in more customers — and scale spreads fixed costs, pushing prices lower still. Each node powers the next.
Build the "AI super-individual flywheel" — use AI to solve a problem → distill prompts and methodology → handle the next task faster → tackle higher-order work → generate more reusable cases and data → train a personalized AI assistant → raise output again. Systematize daily work: a daily mental model → a structured note → a prompt and decision framework → embedded into work and parenting → new cases captured → feeding next week's learning agenda. Each week, check which node is weakest and whether velocity is dropping.
The flywheel effect describes a self-reinforcing loop where each component strengthens the next, gradually substituting external effort with internal momentum. Early-stage flywheels feel inefficient because feedback loops have not yet closed, but past the tipping point compounding produces nonlinear separation from competitors. The strategic task is to design closed loops, identify the weakest link, and prioritize investments that propagate downstream rather than isolated wins. Beware the reverse flywheel — negative loops compound just as ruthlessly. Measure flywheel velocity, not point-in-time output.