Nassim Nicholas Taleb
The month in one sentence
Taleb spent August applying the same intellectual instinct to geopolitics, medicine, statistics, finance, exercise, books, AI, and Twitter fights: look for the hidden denominator, the ignored variance, the asymmetry, the incentive, and the fragility—then distrust anyone presenting an average, institution, credential, or moral rule as though it settles the matter.
The remarkable thing is how little separation there is between his “serious” work and the rest of his life. A paper on heritability, a Lebanese banking scandal, Israeli policy, prostate screening, tire pressure, chilled Beaujolais, a calculus puzzle, and a fake “Nassim Trading Tips” account all get processed through essentially the same machinery.
1. Israel/Palestine became the dominant moral and political subject
This is by far the loudest theme of the month.
Taleb’s position is not merely opposition to the current Israeli government. He increasingly argues that the underlying Zionist project contains a structural problem: differential rights to land, ancestry, and political status based on Jewish identity. He repeatedly uses the deliberately contemptuous “Zionistan” and describes Israel as an increasingly aggressive, fragile project whose apparent impunity encourages further risk-taking.
His recurring argument is a risk-management argument as much as a moral one:
continued aggression → apparent success → greater confidence → more aggression → broader opposition → eventual blowup.
He explicitly compares this to the “Turkey problem”: a strategy can appear successful precisely while it is accumulating the conditions for its own failure.
That connects to an older Taleb metaphor he resurrects several times: a pariah state deteriorates like bankruptcy—lento, adagio, allegro, presto. By August he thinks the process has entered the faster phases. When Yair Lapid says more than 268,000 Israelis have left over three years, Taleb interprets emigration not as an isolated statistic but as a possible acceleration signal.
The argument with Jean Riachi is revealing
Riachi argued that civilians killed in Israeli strikes on Hezbollah commanders are collateral casualties rather than intended targets, and therefore morally distinct from deliberately killing civilians.
Taleb rejects the framing. His objection is partly consequentialist and partly about domain dependence: people apply one moral rule to violence by their own side and another to violence by their opponents.
His challenge is essentially:
stop reasoning from anecdotes about stated intentions; count civilian deaths relative to combatants; apply the same standard to both sides; do not allow “military target” language to make predictable civilian deaths disappear morally.
He compares the logic to justifications for destroying villages containing resistance fighters. Whether one agrees with that analogy or not, it illustrates one of his favorite ideas: logical inconsistency is often produced by switching domains without noticing it.
Ben-Gvir and explicitness
When Ben-Gvir says Israeli forces should kill “30 or 40” Palestinians every night, Taleb’s point is that the novelty is not the proposed behavior but its explicit articulation:
Ben-Gvir is saying out loud what Taleb believes has already been happening under more polished language.
The same reaction appears when he sees Israelis mocking Palestinians lacking water or a rabbi denying Palestinian rights. He treats these not as fringe curiosities but as windows into assumptions normally hidden by public relations.
An especially clever example comes when someone criticizes him for focusing on the rabbi’s pronunciation rather than the genocidal substance. Taleb explains that the linguistic observation is bait: “honey, not vinegar.” By making an interesting, low-temperature philological remark, he gets more people to watch the underlying clip.
That is pure Taleb: rhetoric itself becomes an optimization problem.
2. His historical counterattack on Zionism is increasingly genealogical
Late in the month he begins pushing a compact argument:
Palestinians are among the closest living descendants of the ancient populations of Palestine, including ancient Israelites.
He uses this chiefly as an internal critique of genetic claims to the land. When someone correctly objects that genes do not determine culture or historical identity, Taleb replies that the Zionist argument itself frequently invokes ancestry, so genetic evidence can be used to attack it on its own premises.
He pushes the argument further with biblical and Samaritan claims:
Jerusalem is not mentioned in the Torah proper; according to Samaritan tradition, most Israelite tribes belonged to the northern Israelite/Samaritan world; * modern Palestinian continuity therefore disrupts simplistic “ancient Israelites left, foreign Arabs arrived” narratives.
Importantly, he is selective about evidence. When Zachary Foster cites a paper supposedly showing extremely high Bronze Age Levantine ancestry in Palestinians, Taleb says the conclusion may be right but the specific paper is not citable.
That little exchange is important because it shows his preferred epistemic posture: a claim does not become acceptable merely because it supports his side.
3. Lebanon: Sehnaoui becomes a live demonstration of fragility
The other major political obsession is Lebanese banker Antoun Sehnaoui, whom Taleb repeatedly calls “Voldemort,” a Ponzi banker, and considerably less flattering things.
The spark is journalist Caroline Hayek’s investigation. Sehnaoui reportedly attempted legal action to prevent publication and remove already-published material.
Taleb’s reaction is almost gleeful because it exemplifies an antifragility mechanism:
trying to suppress information advertises that the information is worth reading.
His compressed rule:
Only read articles people are trying to hide.
He turns the episode into a general theory of incompetent wrongdoing. Successful criminals, he argues, survive by remaining discreet—“rip them off, don’t piss them off.” Someone running questionable schemes while simultaneously antagonizing journalists and sources is exposing himself to unnecessary tail risk.
Hence his repeated argument that Sehnaoui is not some diabolical mastermind but simply fragile and stupid.
This becomes a miniature case study for an old Taleb theme: people often mistake temporary survival for competence. A Ponzi can survive for years, just as a negatively convex trading strategy can make money repeatedly, without its eventual failure ceasing to be built into the strategy.
That is why his Substack piece calls Lebanese finance a “technology transfer”: he connects his own background in carry trades, blowups, and antifragility to Lebanon’s financial collapse.
4. Moral obligation is something you must personally pay for
Taleb refuses lucrative invitations to Azerbaijan because Ruben Vardanyan remains imprisoned there. When the organizers increase the fee, he takes the increase almost as evidence that they misunderstand him.
Similarly, he says he will not visit Morocco until there is regime change because of the monarchy’s cooperation with Israel.
A Moroccan reader replies that Taleb may therefore never visit. His answer is characteristically Talebian:
Morocco is more likely to change regimes than I am to change my mind.
Yet he explicitly distinguishes the Moroccan government from Moroccan people, whom he describes as overwhelmingly sympathetic to Palestinians.
This is a concrete expression of his skin-in-the-game ethics. Moral claims are meaningful when they impose costs on the claimant. Refusing money and travel is therefore not incidental—it is part of the proof.
His maxim about fraud gets the same treatment:
if you see fraud and do not shout fraud, you are part of the fraud.
When someone describes this as deontological ethics, Taleb agrees but adds that Aristotelian virtue ethics reaches a similar result: silence may reveal deficiencies of courage, honesty, and integrity.
5. The deepest intellectual theme: stop applying group statistics to individuals
This appears everywhere.
His most compact formulation is:
You are hiring a person, not a subcontinent.
The statistical properties of a group do not determine a randomly selected member, much less a particular individual about whom additional information is available. Variance matters.
That argument also drives his critique of medicine.
Prostate cancer and the “average human”
He becomes angry about a sitting president developing metastatic prostate cancer after apparently not receiving PSA screening.
The conventional argument he attacks is roughly: in men over 70, prostate cancers are often slow-growing, and on average aggressive screening may cause more harm than benefit.
Taleb says the reasoning commits an elementary error:
an average is not an individual outcome.
His deliberately absurd mnemonic is that the “average human” has roughly one breast and one testicle. No actual patient resembles that aggregate.
His language sometimes overshoots mathematically—one can obviously make useful decisions using averages in populations with variance—but the important point he is driving at is real: a population-average treatment effect or mortality rate does not mechanically determine the optimum action for an individual with different conditional characteristics.
Aviation cancer risk: where is the denominator?
Eric Topol posts that pilots and flight attendants have unusually high proportions of radiation-related cancer deaths.
Taleb immediately asks:
What about all-cause mortality?
Then he asks for the absolute scaling:
radiation-linked cancers / all cancers; cancers / all causes of death.
Without those denominators, a frightening relative comparison can correspond to a small absolute change.
“People aren’t naturally able to scale” may be the best summary of his medical-statistics complaints this month.
6. Cholesterol and alcohol: beware optimizing one endpoint
Taleb objects to the claim that driving everyone’s LDL extremely low might eliminate heart attacks.
His response is not “high cholesterol is good.” It is:
heart attacks are not the only endpoint.
He claims all-cause mortality shows a J-shaped relation with cholesterol, including in people without chronic disease, and jokes that shortened lifespan from other causes would not appear “on the cardiologist’s balance sheet.”
The methodological idea is more important than the specific cholesterol claim: specialists optimize the variable they own.
He makes a structurally similar argument about alcohol. A highly publicized paper concludes that low alcohol consumption has no protective health effect. Taleb digs into the confidence intervals, believes their construction is odd or insufficiently explained, and attacks the mismatch between the paper’s content and the headline.
His resulting essay is mischievously titled:
“HAVE ANOTHER DRINK!”
Someone proposes celebrating with a Negroni.
Taleb: “Have 2.”
The joke works because his actual complaint is statistical, while his Twitter persona turns it into an excuse for Dionysianism.
7. He remains obsessed with intermittence rather than static composition
Eric Topol posts research suggesting lower protein intake may promote longevity.
Taleb responds that everyone is rediscovering something resembling the traditional Orthodox Christian diet, with roughly 210 vegan days a year.
But the interesting point is his insistence on frequency rather than average composition.
Biological effects can be nonlinear. If response is convex or concave, then:
eating X every day is not necessarily equivalent to eating 2X every second day.
Averages erase temporal structure.
This is the same mathematical instinct underlying his work on volatility and convexity. Intermittent fasting, alternating dietary regimes, exercise bouts, financial returns, and stress exposures cannot always be summarized by their mean.
8. His current exercise regime is almost comically unoptimized
A substantial secondary storyline is Taleb’s own fitness.
People apparently assume his changing physique is due to Ozempic. He answers:
No Ozempic.
Instead:
about 10 hours of exercise per week; mostly low intensity; cycling; 45-pound rucking; TRX/isometrics and other calisthenics; Olympic presses.
Then comes the deliberately anti-influencer nutritional program:
red wine, white wine, rosé, arak, mastikha, Armenian brandy, prosecco, squid ink, liver pâté, Turkish coffee.
It is partly comedy, partly a rejection of wellness monoculture.
He thinks the standard exercise minimum is far too low
He highlights a study that conditions exercise benefits on baseline fitness and claims benefits continue up to roughly 14 hours per week, contrary to simplistic interpretations suggesting an optimal ceiling near 180 minutes.
Asked whether that means roughly 90 minutes every day, he says yes.
Walking does not count for him unless fast or uphill.
This is one of the most practically interesting pieces of the month because he is attacking the familiar U/J-shaped exercise-risk curve as potentially confounded by baseline condition.
He dislikes rigid interval prescriptions
Asked indirectly about cycling zones, he says:
“The zones must be lognormal, 4x4 is BS.”
That is classic Taleb suspicion of neat discretizations imposed on continuous, heterogeneous biology.
He does not object to intensity. He objects to pretending a tidy protocol has universal validity.
9. Rucking becomes both exercise and experimental instrument
Rucking comes up repeatedly.
He carries about 45 pounds and eventually switches to a GoRuck-style setup distributing weight across belt, front, and back.
When an orthopedic surgeon recommends bone-loading exercises requiring forces several times body weight, Taleb asks why rucking is missing.
The surgeon says the peak force may be too low.
Taleb:
“OK, I will ruck downhill!”
Then he asks about barefoot jump rope and how many minutes per week would provide the required stimulus.
This is Taleb at his best recreationally: he treats health recommendations not as commandments but as hypotheses requiring mechanistic calibration.
10. He starts instrumenting his own ectopic heartbeats
Another fascinating mini-thread begins with the usual implausible distancing device:
“This cycling friend of mine was asked by the first cousin of his mother-in-law’s illegal gardener…”
The “friend,” obviously Taleb, is looking for continuous ECG/Holter-like data during exercise to investigate rare benign ectopic beats that an Apple Watch might miss.
He considers an X2 strap, discovers the Polar H10 can store suitable data through apps, and orders one.
His tentative hypotheses include:
sleep deprivation; jet lag; stress; summer electrolyte depletion; * changes in exercise volume.
He is explicit that he does not know the answer and wants autonomy over the raw data.
That is almost a perfect miniature of his worldview: do not passively accept an aggregate medical narrative when you can collect higher-resolution observations on the system you actually care about.
11. Cycling becomes citizen science
Toward the end of the month, “his cyclist friend” empirically tests tire-pressure calculators.
Silca recommends a pressure that, for his setup, he thinks is roughly 8 PSI too low. Raising pressure on smooth concrete allegedly increases his speed by as much as 3 km/h.
Other calculators are even further off. The René Herse calculator, however, appears to match his experiments.
The broader pattern matters more than bike tires:
calculator → prediction → experiment → reject model if reality disagrees.
This is exactly the epistemology he wishes academics practiced more often.
12. Probability and fat tails remain the technical backbone
He revives his 1/n dominance theorem, arguing that simple equal allocation can outperform modern portfolio theory under sufficiently fat-tailed uncertainty.
A knowledgeable reader points out that the displayed result seems to apply directly when the tail exponent is below one, whereas financial returns often have finite means.
Taleb says the objection is correct and that an additional argument is needed: convergence can be so slow that finite samples remain effectively dominated by the same phenomenon.
Notably, he compliments the critic:
“Very good point.”
That interaction is worth noticing because it contrasts sharply with his treatment of low-quality objections. Taleb can be exceptionally receptive when someone identifies the actual mathematical issue.
Uncorrelated does not mean independent
He also runs a probability quiz around the classic result that uncorrelated variables need not be independent.
One intuitive construction is \(X\) and a nonlinear function \(f(X)\): they remain deterministically related even when ordinary linear correlation vanishes.
He then asks whether a broader theorem applies to all non-affine transformations, scratches out a derivation involving positive second derivatives, and asks Grok to verify it.
Again, Twitter is being used as a public mathematical notebook rather than merely a publication channel.
13. Even dog names become a lesson in tail fitting
A data-science account claims New York dog names follow a lognormal distribution.
Taleb objects that the tail looks more like a power law and emphasizes something that matters far beyond dog names:
fit the tail to the tail.
Small observations near the body of a distribution can create visual distortions and tell you little about the asymptotic regime.
This is one of his longest-running complaints about empirical statistics: investigators routinely choose a familiar distribution because it fits most observations, even when the part that determines extreme risk follows a different law.
14. Twin heritability and IQ science remain targets
He circulates a dense working paper arguing that twin studies overestimate heritability.
One reply mentions BRCA2 giving something like a 60% lifetime probability rather than certainty. Taleb uses it to reinforce the point that genes are probabilistic, not deterministic.
Elsewhere, during the independence/correlation discussion, he attacks what he calls “IQ science” for confusing correlation structures and dependence.
The underlying enemy is essentialism disguised as statistics: turning population-level correlations into claims about immutable individual properties.
15. Economics continues to lose whatever respect he had left for it
One wonderfully compressed tweet:
“The very few things in economics that I thought weren’t wrong are turning out to be wrong.”
That accompanies his emerging “Lydian Stone” project.
He also attacks conventional portfolio theory, Richard Thaler’s criticism of naive 1/n diversification, and Scott Bessent’s Treasury policy, which he compares to:
treating pancreatic cancer with tranquilizers and yoga meditation.
The joke encapsulates the criticism: policies can suppress symptoms or volatility while leaving the underlying structural exposure intact.
The baker, by contrast, receives praise:
most of today’s academia is noise; most of today’s baker output is signal.
The distinction is not anti-intellectual. It is anti-performative production. Bread must work. Academic output can survive without corresponding to anything.
16. Books are physical technologies, not information containers
He repeatedly promotes an essay titled “The Book is a Book, not Disembodied Information.”
This fits with his redesigned library:
no very tall shelves; books within easy reach; organization by themes; non-Lindy books removed from the main rooms unless instrumentally useful.
The target is the idea that a book is merely text and therefore equivalent across printed page, Kindle, audiobook, PDF, or abstracted summary.
For Taleb, the physical form participates in cognition: spatial memory, browsing, rereading, serendipitous adjacency, annotation, and accessibility all matter.
His reference to Umberto Eco is apt: the library is a working cognitive environment, not a trophy collection.
17. Homer is functioning as political theory
After watching The Odyssey, he rereads Homer and even revisits his own siren discussion from Fooled by Randomness.
Antinous, the obnoxious suitor who occupies Odysseus’ house, abuses the weak, tries to seize what is not his, and eventually gets killed, reminds Taleb of a certain modern state.
He asks readers to guess which one.
This is Taleb’s preferred use of classics: not ornamental quotation but reusable models of human behavior. The ancient story remains “Lindy” precisely because the pattern keeps recurring.
18. Philology is another way he attacks political mythology
Throughout the month he jumps between Arabic, Hebrew, French, Greek, and Semitic etymology.
Examples include:
the old pronunciation of Montaigne; Hebrew mitzvah and Arabic waṣiyya; the element din as “law” in Semitic languages; the pronunciation of Hebrew/Arabic gutturals; * discussions of Midian and Moses.
Some of the etymological claims are presented much more confidently than the corpus establishes. But the interesting part is what he is doing with philology: languages preserve continuity that nationalist stories often suppress.
His fascination with pronunciation in Israel is partly archaeological. Older Mizrahi pronunciations retain phonetic distinctions closer to neighboring Semitic languages, while younger Israeli speech often loses them.
For Taleb, phonology becomes evidence of cultural continuity and rupture.
19. AI interests him most when it attacks AI
One of the funniest stories concerns a plagiarized chemistry paper whose paraphrasing software rendered “final solution” as “mass killing of an ethnic group.”
Taleb says he laughed until his side hurt.
But his more substantive AI experiment involves detection.
He asks several LLMs to modify a paper, adds AI-written introduction and conclusion sections, paraphrases them, and then tests Pangram. According to him, the detector catches all the modified material, even paraphrased AI text.
His conclusion:
correcting AI well enough to conceal it may require more work than writing the passage yourself.
Hence the Gödel joke: AI now claims it can certify that something is human-produced, creating a delicious self-reference problem.
He sees an arms race rather than a settled technology.
20. AI is also excellent for tormenting scammers and trolls
Fake accounts impersonating him begin recruiting victims into “Nassim Trading Tips.”
Taleb does not merely issue a fraud warning. He contacts the scammers and encourages followers to ask them “critical questions.”
The joke escalates until he is publicly soliciting “financial enlightenment” from fake Nassim Taleb.
This may be his favorite AI application of the month: industrialized trolling of fraudsters.
Another extraordinary exchange appears on August 31. Someone sends him a weak ideological jab about indigenous rights and land acknowledgments. Taleb responds with an obviously AI-generated parody of academic philosophy:
Before refuting an argument, we must first determine whether an argument exists. What is a tweet? A proposition? A performance? A digital sneeze? Can a tweet “contain” an argument as a jar contains olives?
When another reader asks what is funny, Taleb explains:
“He is a troll.”
Then someone recognizes the response as Borgean/Cortázarian.
Taleb: “Best use of AI.”
The joke is recursive: using machine-generated performative academic verbosity to mock someone he considers incapable of making an actual argument.
21. His insults are part of an epistemic sorting mechanism
Taleb blocks aggressively and says so.
His categories include:
people who reply without reading; people incapable of engaging the actual argument; propagandists; “Hasbara operators.”
He distinguishes between ordinary idiots—“nothing personal”—and propagandists, where it is personal.
The profanity can obscure an interesting mechanism: he tries to make his feed costly for low-effort engagement.
Conversely, someone who provides a technically correct objection can receive immediate praise.
The distinction is less agree/disagree than engaged with the actual structure / did not engage with it.
22. Think tanks, journalists, and public intellectuals fare badly
Taleb warns against think tanks because their funding and career incentives can turn researchers into propagandists, sometimes unwillingly.
His claim is sweeping—much too sweeping if taken literally—but it reflects his broader heuristic:
credibility is inseparable from independence.
When a clip claims Western journalists face career penalties for certain positions on Israel or Iran, Taleb replies that this is why he trusts only people who are free.
Simon Sebag Montefiore receives repeated abuse as an unreliable historian. David Runciman is dismissed because Taleb thinks he reviewed Antifragile without understanding convexity. Anne Applebaum appears on a very short list of public figures Taleb says he immediately regarded as fraudulent.
These attacks are less interesting as judgments on the individuals than as evidence of Taleb’s core credibility test:
Does this person understand the technical object, and are they structurally free to say what they think?
23. There is a surprisingly coherent ethics underneath all the aggression
Several aphorisms summarize it.
“Wisdom lies in understanding the difference between reaction and revenge.”
“If you are actively ignoring someone, you are not ignoring them.”
“Success leads many to their ruin.”
The last comes from Phaedrus and is nearly a one-line description of antifragility’s evil twin: success can conceal accumulating fragility and encourage risk precisely when caution is most needed.
That principle links:
successful Ponzi schemes; aggressive states; carry trades; reputations; health interventions; overconfident institutions.
Taleb’s world is full of systems that appear strongest shortly before their hidden exposure becomes visible.
24. The jokes are not decorative; they encode the philosophy
Some of the best recurring bits:
“My cyclist friend”
Any personal question likely to attract intrusive commentary is outsourced to an absurd chain of acquaintances:
his cycling friend’s barber’s ex-cousin-in-law…
The joke lets him ask highly personal questions while satirizing the social-media demand for disclosure.
Venice versus Genoa
After praising Genoa, he jokes:
encourage everyone to visit Genoa so you can finally enjoy Venice quietly.
“Game theory.”
The anti-Ozempic diet
He denies weight-loss drugs, lists hours of cycling and rucking, then immediately destroys the wellness aesthetic with wine, arak, pâté, prosecco, and Turkish coffee.
Wine science
His cyclist friend wants to know which reds, like Morgon and Fleurie, are best chilled.
Taleb remains Taleb even when the optimization target is Beaujolais.
Bone loading
Told rucking may not generate enough force:
“I will ruck downhill!”
Soviet Estonia
Shown a Soviet apartment district in Tartu:
“New York City in the 80s.”
Academia
A weak Twitter attack inspires hundreds of words asking ontological questions about whether a tweet is like olives in a jar.
It may be the purest joke of the month.
Why this month is interesting
The striking feature is not any single position. It is the unity of the machinery.
Taleb applies almost the same set of questions to every domain:
- What is the denominator?
Radiation cancers are meaningless without total cancer and mortality risk.
- What is the variance?
A population average does not identify an individual.
- Where is the nonlinear response?
Diet frequency, exercise dose, financial returns, and political escalation cannot safely be reduced to averages.
- What is being optimized, and what has been omitted?
Cardiologists optimize cardiovascular endpoints; policymakers suppress symptoms; specialists ignore external costs.
- Who bears the downside?
Moral claims without personal cost are cheap.
- What incentive structure controls the speaker?
Think tanks, journalists, governments, doctors, academics, and scammers are judged partly by what their institutional position rewards.
- Is success evidence of robustness—or merely that the blowup has not happened yet?
This connects Ponzi finance, geopolitics, and antifragility.
- Can the claim survive contact with direct observation?
Tire calculators get tested on a bicycle; exercise recommendations get compared with conditional data; heart-rhythm hypotheses get instrumented.
- Is the critic actually addressing the argument?
If yes, Taleb may enthusiastically correct himself. If not, the probability of being called an idiot rises sharply.
The month therefore reads less like a miscellaneous Twitter feed than like an extended demonstration of a single worldview.
Taleb’s central enemy is not ignorance. It is false precision attached to the wrong abstraction: averages mistaken for individuals, statistical fits mistaken for tails, credentials mistaken for credibility, temporary success mistaken for robustness, information mistaken for books, declared intentions mistaken for consequences, and polished narratives mistaken for reality.
Everything else—the math puzzles, rucking, Homer, Palestinian ancestry, Lebanese bankers, chilled wine, scammer-baiting, and endless Turkish coffee—is variation on that theme.