Nassim Nicholas Taleb
The month in one sentence
Taleb spent July applying essentially the same mental model to politics, history, health, AI, mathematics, scholarship, and his own Twitter fights: systems fail when people mistake a simplified story for the underlying reality, ignore nonlinearities and hidden interactions, or become dependent on an arrangement that is far more fragile than it looks.
The overwhelming subject was Israel/Palestine and Zionism, but the interesting part is that his political tweeting was not separate from his usual intellectual obsessions. He kept translating the conflict into the language of fragility, path dependence, demographic change, information control, incentives, historical continuity, and irreversible shifts in legitimacy. Meanwhile, his arguments about Bryan Johnson, AI, probability, exercise science, think tanks, genetics, and even Proust repeatedly returned to the same epistemological question: are you describing reality, or defending a model, institution, metric, or story that reality has already outrun?
The dominant argument: Zionism is becoming structurally fragile
The largest theme by far is Taleb's conviction that Israel is undergoing a long-term loss of political legitimacy that cannot be repaired by better messaging or short-term military superiority.
Early in the month he quotes a report that Netanyahu's chief of staff favored a single state between the Jordan and Mediterranean. Taleb's response is almost pure Antifragile: Netanyahu, in maximizing Israel's immediate objectives, is supposedly maximizing its future vulnerability. If there is ultimately one state, Taleb argues, the Palestinian population cannot simply be permanently killed or expelled, so maximalism eventually makes the underlying demographic and political contradiction harder rather than easier to conceal.
This becomes his repeated "arrow of history" argument. Glenn Greenwald writes about post–October 7 blacklists against Americans who criticized Israel; Taleb's reaction is that such coercion predictably produced the backlash it was intended to prevent. Later he notes enormous engagement for "Arrest Netanyahu," says the ratio of people attacking versus supporting him has reversed over roughly two and a half years, and observes that younger generations are far less supportive of Israel.
His compact version is:
Political life advances from funeral to funeral.
He is adapting the Planck idea that science progresses one funeral at a time. People often do not update fundamental moral or political commitments; generations replace one another. Hence another formulation:
Morality does not evolve within persons, but across generations.
That is one of the month's most revealing ideas. Taleb isn't primarily predicting that Israel's opponents will persuade its existing supporters. He is predicting cohort replacement.
He even divides the anti-Zionist coalition into two different mechanisms. The left, he says, is turning against Zionism through sympathy for Palestinians and a sense of justice. The right is turning against it because of perceived influence over politics, universities, and media. Whether those descriptions are correct is separate from the point he is making: two ideologically incompatible groups can converge on the same political outcome for entirely different reasons.
This also explains why he keeps saying "history is inexorable" and "never fight time." His claim isn't merely moral. It is strategic: if the underlying flow of demographics, public opinion, information availability, and generational replacement runs against you, suppressing the process may increase the eventual correction.
Information control is, in his model, losing its scalability
A related theme concerns propaganda and the internet.
When an account announces a massive searchable archive of Gaza footage—tens of thousands of videos and photographs, sources, geolocation data, and a live map—Taleb focuses less on any individual recording than on what the archive represents. He argues that Israeli power historically depended partly upon controlling the information environment, whereas millions of people can now directly inspect distributed evidence.
His reaction is essentially: think about what happens to a regime whose narrative advantage depended on information scarcity when the scarcity disappears.
The same idea appears in a different form when he watches public opinion move. Taleb interprets enormous discrepancies in social-media engagement not as Twitter trivia but as a possible leading indicator of a phase transition. Publicly visible elite opinion may look stable long after underlying opinion has moved.
"History doesn't crawl; it leaps," he posts later in the month.
That could almost serve as the political thesis of July: long periods of apparent stability can hide accumulating pressure, after which the visible state changes abruptly.
He is emphatically anti-Zionist, but not straightforwardly anti-American
One of the more interesting tensions in the corpus is that Taleb simultaneously mounts an extraordinarily hostile attack on Israeli policy and praises the United States.
On July 4 he calls America the oldest large constitutional democracy and "the most self-correcting of them all," adding that failures in politics are offset by the robustness of its legal system. When someone posts that America is not the greatest country but merely the greediest, Taleb calls that bullshit and argues that the United States is unusually capable of absorbing ambitious immigrants, allowing them to become rich and join the establishment without surrendering their identities.
So the worldview here isn't "West bad, resistance good."
His distinction is more institutional: America's decentralized, adversarial legal and political machinery gives it mechanisms of correction that he thinks Israel's increasingly maximalist project lacks.
That distinction matters because it prevents the month's political material from reducing to tribal alignment. Taleb repeatedly wants to classify systems by their capacity to survive their own mistakes.
Genetics, Palestinians, Samaritans, and the battle over historical continuity
A second major political-historical project occupies a huge amount of the month: Taleb is trying to undermine Zionist origin narratives through population genetics and historiography.
His strongest repeated claim is that Palestinians substantially descend from the ancient inhabitants of the southern Levant, including Israelites/Judeans, rather than representing a replacement population that arrived from Arabia after the Islamic conquests. Correspondingly, he disputes claims that modern Jewish populations can straightforwardly be treated as a population returning intact to an ancestral homeland after two millennia.
He summarizes his attack on the standard narrative as three alleged errors:
- Jews are simply Judeans returning to their original country.
- Palestinians are invaders from Arabia.
- A sufficiently long interruption in political control can erase the population and history on either side of it.
He says the first two are questions for DNA and the third is a logical error.
The replies become some of the most combative of the month. When archaeologist Yonatan Adler tells him that Jews descend from ancient Judeans and points him toward a lecture, Taleb answers:
DNA or shut the fuck up.
Elsewhere, when another person disputes "Palestinians are the original Judeans," Taleb responds that it is "in the DNA."
His interest in the Samaritans fits the same project. Someone proposes that Samaritans represent the surviving original Israelites and that many other ancient local populations converted rather than disappearing. Taleb says this is "not as crazy as it seems" and revisits his theory that populations commonly change religion and language without being physically replaced.
That mechanism is crucial to everything else he says about Levantine history.
Identity is persistent underneath language and religion
The genetics argument spills into a broader anti-essentialist view of ethnicity.
Taleb prefers "Levantine" to "Arab" for himself and others in the region because he sees Arabization primarily as cultural and linguistic rather than as wholesale Arabian population replacement. When someone mentions old Pan-Arab stories tracing populations back to Yemen or Arabia, Taleb treats them as ideological mythology.
Likewise, he mocks the idea that a population changing religion somehow loses ownership of its previous history. In response to a post saying Byzantine Gaza's Christian history cannot belong to today's Palestinians because they were not then Muslim Arabs, he points out the absurd consequence: by the same logic, pagan Greece and Rome would not be the heritage of modern Christian Greeks and Italians.
This is one of the deeper through-lines of his month: political identities are recent labels pasted over populations with much longer continuity.
That is also why he is fascinated by apparently trivial language fossils.
Gabagool, Aramaic, Turkish, and Taleb the historical linguistics obsessive
The month opens with an extended discussion of Italian-American pronunciation.
Someone notes that "gabagool" isn't a standard Italian word but an Italian-American pronunciation of capocollo. Taleb argues that immigrant communities preserve dialectal features that later disappear in the homeland. He connects New Jersey Italian pronunciations to southern Italian and possibly Siculo-Arabic phonology, recalling that he once ordered "Brozhut"—prosciutto—in Palermo and nobody understood him.
The broader observation:
Diasporas can function as linguistic fossils.
He compares this with Lebanese dialects preserved in Brazil or Australia after television helped erase them in Lebanon itself.
Later there is an extended mini-seminar on Aramaic. A Maronite liturgy is described as being in "the very same language of Christ." Taleb objects that the liturgical language is an Eastern Aramaic/Syriac variety, whereas Jesus would have spoken a Western Aramaic variety such as Galilean. He starts comparing forms word by word and eventually argues that Maaloula and a few Syrian communities preserve something much closer to Jesus's actual language.
He is similarly delighted to discover that "deep state" entered modern political vocabulary through Turkish derin devlet, and notes Mediterranean equivalents of "slowly, slowly": Greek siga siga, Turkish yavaş yavaş, Arabic shwayy shwayy, Italian piano piano.
These aren't merely hobbies adjacent to his politics. They reinforce his preferred way of thinking about history: look at surviving traces instead of accepting retrospective national stories.
The real epistemological enemy is the sophist
Another central thread is Taleb's increasingly explicit distinction between people trying to discover the truth and people paid to produce arguments.
He attacks think tanks relentlessly.
When someone says think tanks connect ideas with political action, Taleb translates the euphemism into "propagandize, distort, and cover up for the powers that be." His proposed diagnostic is brutally simple:
Only one piece of information counts: who pays the "researcher".
He then elevates this from institutional criticism into philosophy. He compares the distinction between genuine scholarship and advocacy to the ancient distinction between Socrates and the Sophists.
The sophist asks: How can I argue this position effectively?
The philosopher asks: What is actually true?
He later compresses it further:
the difference between the lawyer and the good judge.
This fits another line from the month:
Spotting a bullshitter is an easier skill to acquire than spotting bullshit.
That is a very Talebian epistemological move. Evaluating every technical argument may be prohibitively expensive. Evaluating incentives, track record, dependence, and behavior may be much cheaper.
Hence his obsession with shills, paid analysts, and institutional affiliations. He repeatedly treats funding arrangements as evidence about what kind of epistemic process is occurring.
His attack on abstraction is becoming an argument about AI
The intellectually richest nonpolitical sequence comes from discussions at his Real World Risk Institute with Stephen Wolfram and others.
Steven Strogatz asks whether Axelrod's famous results about repeated prisoner's dilemmas still hold—particularly the idea that successful strategies tend to be simple, nice, forgiving, retaliatory, and understandable.
Taleb's answer is basically: that apparent simplicity may be an artifact of a tiny strategy space.
With enough computation and data, he says, richer strategies can vastly outperform simple tit-for-tat rules once many strategies interact under path dependence.
He generalizes this into an attack on the intellectual prestige of parsimony.
Old modeling practice often had to compress reality into a small number of variables because data and computation were scarce. But modern AI succeeds precisely because enormous models can absorb vast amounts of structure rather than replacing it with elegant toy assumptions.
He formulates a wonderful methodological rule:
If you say "assume that" and do not believe in the assumption, do not make the inference.
For Taleb, an assumption isn't innocent just because everybody remembers that it is an assumption. If the model's conclusion depends critically on something you know isn't true, then sophistication downstream does not rescue it.
This connects his AI enthusiasm, his hostility to academic modeling, and his risk theory surprisingly neatly.
AI may hit technicians from an unexpected direction
Taleb also reverses one of his own previous AI predictions.
He says he had assumed electricians, appliance technicians, HVAC technicians, and similar hands-on occupations would be relatively protected from AI. Then he encountered a practical bottleneck: technicians take too long to arrive.
So he started diagnosing household systems himself using AI.
The important distinction, in his telling, is not that an LLM can physically repair an air conditioner. It can eliminate enough of the diagnostic and informational bottleneck that the customer performs more of the work.
When someone objects that he could simply have Googled the manufacturer's instructions, Taleb points to something qualitatively different: he can upload a sequence of photographs and have the system synthesize a diagnosis.
Search retrieves.
AI interprets.
That distinction mirrors his theoretical AI discussion: the value comes from integrating more context rather than merely locating another little piece of information.
Bryan Johnson becomes Taleb's perfect punching bag for high-dimensional risk
The month's funniest sustained feud is with Bryan Johnson, whom Taleb repeatedly calls "Draculino."
The underlying argument is serious.
Johnson's extreme longevity program contains a large number of simultaneous interventions. Taleb asks readers to calculate the possible interactions among 100 pills.
Pairs alone give 4,950 combinations.
Triples give 161,700.
Higher-order combinations explode combinatorially.
Taleb's point is not that every combination is biologically relevant. It is that once a system contains many simultaneous interventions, the hypothesis space becomes enormous while the observational sample remains basically one Bryan Johnson.
So when somebody says Johnson at least has "skin in the game" and humanity can learn from the experiment, Taleb replies:
Benefit? There is no information.
One body undergoing dozens or hundreds of changing interventions cannot identify causal effects cleanly.
Johnson later posts a list of things he has removed—alcohol, bad sleep, loneliness, poor diet, sedentary behavior, and so forth—and says subtraction is his best longevity intervention.
Taleb attacks this as dishonest framing: ordinary exercise, sleep, decent food, and avoiding obvious harms may be subtractive, but Johnson's distinctive program is overwhelmingly additive.
Taleb then attacks metric optimization itself. If an intervention is selected specifically because it moves a biomarker and you subsequently advertise that biomarker's improvement as evidence that the intervention works, he calls the reasoning circular.
In other words: you haven't conducted an experiment; you've optimized the dashboard.
His alternative view of longevity is surprisingly austere
Against biohacking complexity, Taleb proposes something much simpler: mortality.
"Metrics shmetrics," he writes. There is ultimately one meaningful measure of aging: the force of mortality, roughly the instantaneous probability of dying.
Aging means that this hazard is increasing.
He links that to his Lindy work. A creature with an approximately constant mortality hazard isn't "aging" in the conventional actuarial sense even if its average life is finite; it is mostly being killed by external accidents. And "immortality" would require not merely eliminating biological aging but avoiding accidents forever—which is impossible.
This lets him turn Johnson's intervention stack against the stated goal: every extra intervention may introduce another accidental pathway.
It is the medical version of the same principle he applies everywhere else during the month:
Don't create hidden fragility while optimizing an observable metric.
Exercise, aging, and the nonlinearity of being elite
Taleb spends a substantial amount of time on exercise physiology.
One particularly good observation concerns elite athletes and aging. Athletes often experience aging as catastrophic because they live near the maximum of a performance distribution. A tiny decline in underlying capability can move someone from world #1 to effectively irrelevant.
For ordinary people far below their physiological ceiling, the opposite can happen. A middle-aged person can become stronger, gain muscle and bone, improve aerobic capacity, and outperform their younger untrained self.
Taleb calls this the nonlinearity of the maximum.
It is another fat-tail-style conceptual move: changes near an extreme have very different consequences from identical changes near the middle.
He also questions overly tidy fitness correlations. When someone says unusually low resting heart rate should correspond to unusually high VO₂max, Taleb immediately asks where this comes from and jokes that the claim seems to assume an (R^2=1).
Likewise, when a study concludes that roughly half of muscular aging persists despite exercise, he notices that the supposedly trained group performed fairly modest endurance exercise. His objection is not "the paper is wrong" but that the operational definition of training excludes intense, eccentric, explosive, neural, and strength stimuli—and therefore cannot establish a ceiling on what training can prevent.
Again: inspect the actual intervention, not its label.
Fat tails appear whenever somebody makes a bad analogy
Paul Graham enters the timeline during an argument comparing great wealth with extraordinary bench-press performance.
Taleb's intervention is classic Incerto.
Strength and height behave roughly like thin-tailed variables. Wealth is fat-tailed.
Therefore percentile-equating the two produces absurd intuitions. Moving from $50 million to $1 billion sounds enormous in dollar space, but in the far tail of the wealth distribution it may represent a surprisingly small change in rarity.
He restates the point using human height: statistically, the difference between an ultra-high-net-worth individual and a billionaire can correspond to only several centimeters among extraordinarily tall men. A centibillionaire adds only a little more.
This is Taleb at his best on Twitter: somebody makes a commonplace numerical analogy, and he uses it to demonstrate that distance measured in raw units and distance measured in probability space are different things.
Mathematics is not content for him; it is recreation
Throughout all this, he casually interrupts wars about Gaza and longevity with integrals, derivatives, differential equations, chaotic functions, AM–GM, fractal self-similarity, and little puzzles.
He differentiates (n^{1/n}) to recover the maximum at (n=e). He suggests Laplace transforms for an integral. He develops a self-similar sawtooth solution to another. He sees an unidentified plotted object and reverse-engineers a function involving
[ (-1)^{\lfloor nx \rfloor}(x\sin x)^2. ]
There is a charming lack of separation between "serious research" and "playing around."
He will also make an error publicly, correct it, and then joke about the correction. During the enormous Bryan Johnson combinatorics number, somebody challenges an omitted term and Taleb replies with mock-formal contrition about hoping to someday regain the man's respect.
When asked why one gigantic integer should end in 276 rather than 275, he eventually explains:
Because that thing is like the other thing and you need to put more stuff in it after that shit.
It is almost a parody of mathematical exposition.
The ethical theme: care more about your own judgment than your audience
Another surprisingly coherent thread concerns reputation.
Taleb writes what he calls the:
MOTHER OF ALL ETHICAL RULES
The rule is that moral rectitude means caring much more about what you think of yourself than what other people think of you, regardless of the reward structure.
This connects directly to his political behavior.
A quoted Rihanna story says she was warned that supporting Palestine would cost her Israeli listeners. Taleb recalls receiving similar warnings about his readership:
"fuck them"
He then claims he ended up with more—and better—readers, and notes that old books returned to the bestseller lists precisely when he cared least about the commercial consequences.
Whether or not the causal story is as clean as he presents it, the philosophy is consistent with Skin in the Game: independence requires being willing to pay for your beliefs.
A related insult later makes the same point negatively:
you are so insignificant that your praise would be worth no more to me than your insults.
Reputation only controls you when you grant the audience jurisdiction.
His strange rule about the dead
The death of Lindsey Graham triggers one of the month's oddest ethical excursions.
Taleb strongly disliked Graham politically, but objects when people celebrate or insult him after his death. The Latin-flavored version is essentially: you are killing a dead man.
His argument is instrumental and character-based rather than sentimental.
A dead enemy is no longer a threat. Spending energy attacking him is therefore useless. Worse, it is psychologically cheap: it allows you to feel courageous while taking no risk.
He compares it to abusing an effigy.
have some skin in the game & go focus on living assholes.
Later he adds an important qualification: you shouldn't abuse the dead, but you should condemn living people who glorify an evil person's actions.
It is an unusually precise distinction between judgment and performative vengeance.
Then, characteristically, anxiety about an unexpected death sends him down a technical rabbit hole about aortic dissection, aneurysms, valve anatomy, hypertension, connective-tissue disorders, and the possible effects of very high training loads.
Moral philosophy becomes cardiovascular epidemiology in about twelve hours.
The jokes are doing intellectual work
Taleb's humor this month is mostly not decorative. It establishes social boundaries and mocks epistemic pretension.
A favorite recurring device is the absurdly distant source:
the first cousin of the prison mate of his tax accountant's ex-wife's divorce lawyer
or
the neighbor of the lawyer of my second cousin's bookmaker
This is how he presents questions he mock-pretends are too embarrassing, controversial, or uncertain to own personally. It is the Twitter equivalent of "asking for a friend," recursively exaggerated.
"Kakokristan" and "Zionistan" function similarly as deliberately juvenile renamings of Israel, while "Draculino" turns Bryan Johnson's anti-aging persona into a vampire character.
He frequently announces that some reply is an "intelligence test," says a post is useful for "blocking idiots," or gives offenders variations of "I block idiot. Nothing personal." Feed curation becomes a kind of evolutionary selection mechanism.
There are smaller jokes everywhere:
Genoa is "Venice but without the loud tourist in flip flops." High-speed rail will therefore probably ruin it. Asked whether espresso is an aesthetic experience like wine: "Try Turkish." A grand defense of Proust's English translator produces the analogy that Scott Moncrieff's ornate prose is the Alhambra compared with Proust's camping hut. He calls "metrics" "shmetrics." Probability examples become PROBABILISTIC RIGOR DUJOUR. Mathematics becomes MATH QUIZ DUJOUR. Asked to explain Extremistan versus Mediocristan, he responds: "You can't tell them if they don't know," calling it a Yogi Berra-ism.
The insults are less subtle. "Retard," "retardito," "schlemiel," "shill," "academic leftover," and "bullshitter" are practically punctuation.
The contradictions are part of what makes the month interesting
There are several productive tensions.
Taleb despises simplistic models yet routinely compresses political actors into extremely harsh categories.
He insists on epistemic rigor while sometimes making extraordinarily strong historical and genetic assertions in tweet-sized form and promising the fuller evidence "later."
He denounces propaganda while acting as an unapologetic partisan in the Israel/Palestine conflict.
He praises American institutions while describing portions of American political and intellectual life as captured by propagandists.
He urges people not to care about outsiders' opinions while spending an impressive amount of time blocking, insulting, correcting, and arguing with outsiders.
And despite his hostility toward simplification, he loves devastatingly simple heuristics: Who pays the researcher? DNA or shut up. A dead enemy is no longer a threat. Don't optimize the metric. Don't infer from assumptions you don't believe.
These aren't necessarily fatal inconsistencies. They show the difference between Taleb as theorist and Taleb as polemicist. The theorist distrusts premature compression; the polemicist specializes in compression.
What ties the whole month together
The most interesting way to read these tweets is not "Taleb talked mostly about Israel and occasionally about math."
It is that nearly everything passes through the same handful of ideas.
Nonlinearity: a tiny deterioration can end an elite athlete's career; extreme wealth cannot be compared with ordinary Gaussian traits; many drug interventions create combinatorial interactions; political opinion can look stable and then jump.
Fragility: military dominance can create political vulnerability; information control becomes dangerous once it fails; a minority that depends on a stronger patron may discover too late that the patron itself was fragile.
Path dependence: history cannot simply be restarted in 1948; populations change religions without disappearing; today's identities contain ancient linguistic and genetic remnants; repeated games depend on what happened before.
High dimensionality: real biological systems, AI, strategic games, and societies contain more interacting variables than toy models admit.
Incentives: before evaluating the argument, ask who pays the analyst and what they are rewarded for saying.
Subtraction over intervention: don't add 100 longevity hacks when ordinary health practices and fewer failure modes may dominate.
Independence: if fear of losing readers changes what you say, your audience owns you.
And, underneath almost everything, time.
Immigrants fossilize extinct dialects. Religions change while populations persist. Generations replace one another. Athletes decay away from their maxima. Mortality hazards rise. Political coalitions become obsolete. Propaganda that worked in one information regime fails in another.
Taleb's July is therefore less a set of separate opinions than an extended argument that reality is dynamic, historical, nonlinear, and too high-dimensional for the stories institutions tell about it.
His bet—politically, scientifically, and personally—is that eventually the underlying process wins.