Skin in the Game

"Don't tell me what you think, tell me what you have in your portfolio." — Nassim Taleb

Skin in the game isn't merely "incentive alignment" or "fairness" — it's a filter: only when decision-makers bear the downside of their decisions can a system weed out bad ideas, bad strategies, and bad actors. The moment a decision-maker enjoys the upside but escapes the downside (gains privatized, losses socialized), the system silently accumulates hidden fragility — until one day the tail risk detonates all at once.

Non-trivial: (1) it's essentially an information-transfer mechanism. Risk information propagates through "who bears the consequences"; cut that link (the agency problem) and the agent's incentives decouple from the principal's tail risk. (2) The key is the downside, not the aggregate — the difference between talkers and doers is that doers have bet themselves. Only those who pay for being wrong actually calibrate. (3) Structurally isomorphic to evolution: natural selection works precisely via skin in the game — wrong bets get removed from the gene pool. Markets, evolution, and science's self-correction all run on the same machine. (4) Its higher form is "soul in the game": bearing downside for others, not just yourself.

Practical test: before weighing any advice, ask — what does the advisor lose if they're wrong? If the answer is "nothing," sharply down-weight that advice.

Classic example

Nearly 4,000 years ago, the Code of Hammurabi: if a house a builder built collapses and kills the owner, the builder is put to death. Risk symmetry written straight into law — stronger than any quality standard, because it binds the builder's life to the house's. Contrast 2008: financial executives designed products that dragged the whole system under, then walked away with bonuses while taxpayers absorbed the downside. Where there's no skin in the game, fragility accumulates silently.

BigCat scenario

(1) Technical decisions: an architect who only draws diagrams and never goes on-call differs vastly in quality from one who carries the pager and gets woken at night by their own design. Being on-call is the architect's skin in the game — it makes designs antifragile automatically. (2) AI collaboration: when you deploy an agent that acts autonomously but bears no consequences, all tail risk lands on you (the principal). For high-downside decisions, keep yourself in the loop — don't let an agent with no skin in the game place bets for you. (3) Parenting: letting a child bear small natural consequences (forgot homework → faces the teacher) teaches more deeply than shielding her. Over-protection removes her skin in the game and breeds fragility.


English Prompt
I'm facing a decision / weighing advice from several sources: [describe decision and who's advising]. Analyze through the Skin-in-the-Game lens: 1. For each advisor/party, what do they actually lose if wrong? Who enjoys upside without downside? 2. Where do the downside consequences ultimately land (find the hidden "socialized losses")? 3. Propose one concrete redesign that re-binds decision rights to downside exposure, putting me and key parties genuinely in the game.

Via Negativa

"In the pursuit of the Dao, one subtracts daily." — Laozi / subtraction is more robust than addition

Via Negativa: improve by removing rather than adding. Our knowledge of "what's wrong, what's harmful" is far more robust than our knowledge of "what's right." Falsification beats verification; subtraction beats addition — because removing something known to be harmful has bounded side effects, whereas adding something "seemingly beneficial" injects countless downstream couplings you can't compute.

Non-trivial: (1) the asymmetry of knowledge: negative knowledge (what doesn't work) survives time far better than positive knowledge (a theory of "why it works"). Popper's falsifiability is essentially a via-negativa epistemology. (2) Antifragility: what survives time's filter is what's left after removal; addition is fragile, subtraction antifragile (echoing the Lindy Effect). (3) Isomorphic to Eastern wisdom: Buddhist apophasis (neti neti, "not this, not this") and Daoist "subtract daily" both approach truth by subtraction. (4) Humans have a strong addition bias — research repeatedly shows that, facing a problem, people overwhelmingly choose to "add elements" and barely consider "taking one away," even when removal is superior. It's a cognitive default that must be deliberately countered.

Practice: to improve any system, first ask "what can I delete?" rather than "what should I add?" Deletion is the default move; addition must carry the burden of proof.

Classic example

The history of medicine is a textbook of via negativa. Through the long ages before antibiotics, doctors' "active interventions" (bloodletting, purging, violent remedies) killed more patients than they saved. The largest leaps in human lifespan came not from "adding" some miracle drug but from removing harmful factors — handwashing, clean water, isolating pathogens, stopping harmful routine practices. "First, do no harm" (primum non nocere) is precisely this principle's medical oath.

BigCat scenario

(1) Productivity: the highest-leverage move is usually removal (cut meetings, cancel commitments, uninstall tools), not stacking on yet another new system. Subtracting from your calendar pays an order of magnitude more than adding to it. (2) AI workflows: when an agent's output degrades, the reflex is "add a few more instructions" — wrong. First delete the redundant context and conflicting prompts; the more bloated the context, the more confused the model. (3) Thinking: to see something clearly, first remove the old beliefs you're "certain of but wrong about," then talk about importing new frameworks — clearing out the false matters more than stuffing in the true. (4) Parenting: removing obstacles (over-packed schedules, screens before sleep) often develops a child more than enrolling her in one more class.


English Prompt
I want to improve [system/process/life area]; current state: [describe]. Invert my default thinking using Via Negativa: 1. List 3–5 things most worth removing (harmful, redundant, fragility-adding), ranked by removal-benefit × low removal-cost. 2. Point out my "addition bias" — what I instinctively want to add that's actually the wrong direction. 3. Propose a subtraction-first improvement plan, explaining why each removal's side effects are bounded.

Barbell Strategy

Both extremes, nothing in the middle — cap the downside, open the upside

The Barbell Strategy: load resources at two extremes — extreme safety + extreme aggression — deliberately avoiding the "seemingly prudent middle." For example, 90% in the safest place + 10% on high-convexity small bets. It's called a barbell because all the weight sits at both ends and the middle is empty.

Non-trivial: (1) the core is convexity: the safe leg seals the downside (you lose at most a little), the aggressive leg keeps unlimited upside — you want asymmetric payoff: bounded loss, unbounded gain. (2) "Medium risk" is a trap because it relies on knowing precisely how large the risk is — and tail risk is exactly what's unknowable. A basket of "medium-risk" assets will correlate and crash together in a crisis; the so-called diversification is an illusion. (3) It transfers across domains: career (stable income + wild side bets), reading (timeless classics + frontier ventures, skip the mediocre middle), health. (4) Same family as optionality — you're buying lots of positive options at small cost (echoing "superposition & optionality" from physics thinking).

Practice: don't ask "is this high or low risk?" — ask "is my downside capped and my upside still open?" The truly dangerous choices are the ones that feel moderate but are secretly tail-exposed.

Barbell: load both ends, avoid the middle ~90% max safety ~10% convex bets ✕ middle "medium risk" = hidden tail exposure downside capped upside unbounded
Barbell: a downside-capping safe leg + an upside-opening convex leg, with the middle left empty
Classic example

Taleb's own allocation: roughly 85–90% in the safest assets (e.g., short-term Treasuries), the remaining 10–15% on highly speculative, strongly convex bets. In the 2008 crisis, the safe leg preserved capital while the convex leg actually paid off big in the crash. Contrast the "balanced portfolios" holding a basket of "medium-risk" assets — when the crisis hit, they all correlated and dove together. The moderate portfolio is precisely the most fragile.

BigCat scenario

(1) Career design as an "AI super-individual": keep a stable income base (the safe leg) while making asymmetric small bets on emerging AI capabilities (the convex leg) — don't stake everything on one "looks okay" single path. (2) Time allocation: 80% on reliable core competence, 20% on wild experiments that could 10x. (3) Reading: read time-tested foundations (math, classics) + the bleeding edge, skip the middle ground you'll forget the moment you finish. (4) Parenting: a solid secure base (attachment, stable routines) + permission for bold risk-taking and exploration in play; don't over-optimize the "safe middle."


English Prompt
I'm allocating in [domain: career/investing/time/learning]; current state and options: [describe]. Restructure with the Barbell Strategy: 1. Am I sitting in the dangerous "medium-risk middle"? Flag any option I think is prudent but is actually tail-exposed. 2. Design a barbell: which part should be extremely safe (cap downside), which should be high-convexity bets (open upside), and in what proportions? 3. List 2–3 low-cost, high-upside asymmetric bets, stating whether each one's maximum loss is bounded.

Naive Interventionism

Iatrogenics: harm caused by the healer — routinely underestimated

Naive interventionism: the urge to act even when doing nothing is better — while systematically ignoring "the harm the intervention itself causes" (iatrogenics, literally "harm brought by the healer"). It complements Via Negativa: via negativa tells you to actively remove what's harmful; naive interventionism warns that the "intervention" you blindly add is often itself the harmful thing.

Non-trivial: (1) action bias: visible intervention is rewarded, visible inaction punished — even when inaction is correct. And the harm of intervening is usually delayed and invisible, so it's badly underweighted. (2) In complex systems, interventions have nonlinear, lagged side effects; the more robust and self-healing the system, the more likely an external intervention backfires. (3) "Procrastination as a filter": sometimes a delay lets the system self-correct; naive intervention dismantles that filter. (4) The real skill is discernment: in fragile, genuinely critical situations, intervene decisively and hard (true emergencies); in robust, self-healing systems, restrain. This maps to Daoist wuwei and medicine's "watchful waiting," and echoes "Chesterton's Fence" — if you don't understand why it's there, don't remove it yet.

Practice: before acting, ask "what happens if I do nothing?" Put "non-intervention" on the table as a formal option, rather than defaulting to "I must do something."

Classic example

Forest fire suppression: extinguish every small fire (a seemingly responsible intervention), and dead leaves and brush keep piling up — fuel accumulates until it erupts into an uncontrollable megafire. The intervention hand-built the very tail catastrophe it tried to prevent. Ecologists later understood: periodic small fires were the system's self-regulation; suppressing them traded small fluctuations for one giant disaster. The same pattern appears in medical over-treatment — frequent intervention in self-limiting conditions brings iatrogenic harm exceeding the benefit.

BigCat scenario

(1) Engineering: constantly "optimizing" a well-running production system, itching to tweak parameters, often introduces more failures than it prevents. The on-call engineers who "resist touching a system that's working" tend to have the highest availability — don't intervene in a system that's working normally. (2) Management: micromanaging a high-performing team (intervention) destroys the self-organization that made it effective. (3) AI: repeatedly re-tuning prompts and over-engineering a pipeline that already works — restraint usually beats fiddling. (4) Parenting: hovering over a child and solving her every small struggle (iatrogenic) robs her of the chance to build resilience; "watchful waiting" lets the natural laws of development do the work.


English Prompt
I'm about to intervene in [system/team/child/project]: [describe what I want to do]. Stress-test me with Naive Interventionism: 1. What happens if I do nothing? Can this system self-heal, or is it a genuine emergency? 2. What delayed, invisible iatrogenic side effects might my intervention cause? Could it create the very problem it aims to prevent? 3. Judge whether this is a "fragile situation demanding decisive action" or a "robust system demanding restraint," and propose the minimal necessary intervention (or "stand pat").