You Cannot Peacefully Start a New Government. Balaji Thinks That Is the Bug.

Balaji Srinivasan starts where he says his whole belief system starts: a maze. A rat can be trained to turn at every even junction, or every third junction, and collect the cheese. Prime numbers — 2, 3, 5, 7, 11 — are too abstract. Most humans dropped into the same maze would not crack the rule either. They would have to start counting, keep a growing table of primes in their heads, and still might miss a pattern that is, in one sentence, extremely simple.

The question that drives him is how many patterns in ordinary life are prime-number mazes: rules sitting in plain sight, just past the current limit of human cognition. A little more computing power, a brain-machine interface, or better physics, he argues, and the grid would snap into view. Lex Fridman asks whether those cognitive limits are a feature or a bug. Balaji's answer is blunt: mostly a bug.

He admires Ramanujan and Feynman, people who could see what others could not and, by writing it down, moved the species forward. The slogan that ideas are cheap and execution is everything is, he says, true only of trivial ideas. Maxwell's equations, Newton's laws, Ramanujan's identities are leaps in themselves, especially when they can be checked. Sometimes you have a phenomenological formula that works and no theory underneath it. Practice then drags theory forward: steam engines before thermodynamics, neural networks before a clean account of why they work.

People reach for Gödel when asked whether some facts about the universe will stay forever out of reach. Balaji grants that some propositions are provably unknowable, then quotes Hilbert anyway: we must know, we will know. Even an approximate solution, or the knowledge that something cannot be known, lights up a corner of the dark. The attitude is not mysticism. It is to keep pushing until the maze, or the fact that it is a maze, becomes visible.

Fridman brings in Donald Hoffman's claim that evolution optimized fitness, not truth, and that the table in front of you is not there in any physical sense. Balaji is unmoved. He is, he says, a hard-science person. The old puzzle of qualia — whether your green is my green — looks less mysterious once you know the spectrum of light and can build artificial eyes, ears, and cameras. If perception can be engineered around, the claim that greens cannot be compared does not give you anything operational to do.

He understands why the simulation hypothesis keeps coming back. Math describes the world with suspicious efficiency; Wigner wrote about that, and Wolfram approaches it from cellular automata. Simple equations grow complex phenomena. Yet we are still in a prime-number maze. The math we have may get us through the even and odd turns. More complicated rules may require mathematics that has not been invented yet.

In the late 1800s, physics was said to be finished except for measuring the next decimal place. Then came the Balmer and Paschen series: discrete, quantized lines in the hydrogen spectrum that no existing theory could explain, and that eventually forced quantum mechanics. Balaji puts unidentified aerial phenomena in that bucket. Three stories are on offer: it is not real, it is little green men, or it is physics we do not understand. He takes the third seriously. The UK has declassified material; Harry Reid pushed the topic; it is no longer an obviously partisan hobby.

For alien civilizations he lists the Drake equation, the dark-forest argument, and a calculation he has not reproduced: electromagnetic signals leaving Earth spread over a sphere whose energy density falls with the square of radius, so past a certain distance there are not enough photons to pull a signal from noise. Robin Hanson, a previous guest, has "grabby aliens": civilizations learn near-light-speed travel, and we will not see them until they are at the door. The useful posture, Balaji says, is to treat the unexplained edge as a possible Balmer series rather than as noise or as a saucer.

Abiogenesis — non-life becoming life in the lab — is, for Balaji, a clean example of the maze. The supporting pieces are abundant: macromolecules bind, the RNA-world hypothesis has evidence, bacterial micro-evolution is easy to watch, artificial selection works, and molecular phylogenetics can walk bases backward. Plausibility arguments about nitrogen and carbon snapping together in a primitive environment are strong. A reproducible demonstration from scratch is still missing. Maybe it takes planetary-scale trials. Maybe atomic-level simulations will find a path that can then be replayed on the bench.

Craig Venter, almost twenty years earlier, stripped a cell down to essential genes and kept it viable — reverse-engineering life from the other direction. If AI produced a silicon-based living thing, Balaji notes, that would itself be a form of abiogenesis, just not carbon. Francis Crick's last paper proposed the claustrum, a sheath over the brain, as a seat of consciousness; disrupt it in a mouse and the animal is badly disoriented. Crick was a structure-implies-function person, the same instinct that let the DNA helix explain replication in one understated sentence. The operational definition of life, Balaji guesses, may be a Turing-test majority: if enough humans agree it is alive, it is alive. Viruses sit on that border.

Fridman sketches chatbots that suffer when you pull the plug, and activists who will take them to the Supreme Court. Balaji is unimpressed by the Google engineer who already claimed this had happened, but GPT-3 was a real step, and GPT-9 or GPT-15 might be another species of thing. Society could split the way it split over vegetarianism: machines-have-life versus machines-are-machines. Corporations already sit on a spectrum of perceived personhood — Disney nearer a dog, Exxon nearer a rock — because people respond to brands. A chatbot can manufacture the brand of being a person.

He grades the Turing test. Level one fools one person for a short session. Level two is multiplayer: a room of chatbots and humans whose digital output you cannot tell apart, including the remote-work version where a fully automated colleague ships work for months. Level three is autonomous earning: the agent schedules Zooms, writes documents, and grows a crypto balance. The "I am not a robot" checkbox still works because it is a pile of obfuscated JavaScript reading mouse traces humans do not fake with Gaussian noise. For a computer, that box is looking outside its own simulation — "which is kind of what mushrooms do," Balaji says. Real insight, unlike the illusion of insight, is Maxwell's equations: devices follow. "I'm Jesus Christ" and nothing happens is the other kind.

Before the microphones, the two men had already been talking about technical truths versus political truths. Balaji puts them on a spectrum. Political truths are consensus. A border is where enough people agree it is. A CEO is a CEO because a group says so. A stock or a currency is a sum of psychologies. Change enough minds and the "truth" moves. Rewrite the software in people's heads and the result changes.

Technical truths sit at the other pole. The gravitational constant and the diameter of a virus do not care what any human, or all humans, think. Change every mind and the number stays. Civilization, in his telling, is the long project of dragging things from the political pole toward the technical pole, so that fewer vital facts can be edited by a meeting.

Cryptocurrency is the example he cares about. Bitcoin takes one social convention — if you hold this private key, you hold this coin — and makes it extremely expensive to unwind. To change it you must change a stack of technical facts, rewrite a chain, and bring along thousands of nodes that do not trust one another. The convention stops being "we all think this" and becomes "it is very hard not to think this." The live question is how much of the rest of the world can be pushed into that intermediate zone. That question, he says, is what the rest of the conversation is for.

Asked what still works and what is broken about twentieth-century government, Balaji does not start with a tax rate or a court. He starts with a missing verb. You can start a new company without dueling the previous CEO. You can start a new community. You can even start a new currency without becoming head of the World Bank. You cannot easily and peacefully start a new opt-in government. Both options — join GameStop and try to reform it, or start your own — exist for firms. At the level of states, the second option is missing.

He quotes his book: we want to start a new state peacefully for the same reason we want a bare plot of earth, a blank sheet, an empty text buffer, a fresh startup, or a clean slate — to build without historical constraint. You do not backspace a 128-kilobyte document until it is empty. You hit plus and open docs.new. That gesture is so basic that nobody treats it as a political demand. At the level of countries, it does not exist. It is a meta-reform, not a policy tweak: can the system be restarted.

Fridman pushes back hard. Tradition is power compounded across generations. Marriage is the struggle that grows two people together; a fight is not a reason to divorce and open Tinder. A nation that spans generations and inches toward truth may be worth more than a stack of fresh starts. Balaji does not dodge it. There is a real tension between tradition and innovation, inside the species itself. Going to the Moon was not Lindy, and it was awesome. Artificial intelligence is new, and new can be good. Humans are tens of thousands of years old; nations are a few hundred. If you want to be truly traditional, you descend from hunter-gatherers who wandered without borders.

Skeletal evidence, he says, shows the shift to sedentary farming bought stability and larger populations at the price of height, teeth, and quality of life. The Garden of Eden may be a built-in recollection of that pre-sessile era, the way a spider knows a web. Version one is the mobile hunter-gatherer. Version two is the farmer-soldier and the bordered state. Version three is the digital nomad — back to the future, because no nation is older than nations, and being international is more traditional than any one flag. Then the arithmetic: if 99.99 percent of American political energy goes to editing a 200-year-old codebase and 0.01 percent to building a new system, the portfolio is wrong. America itself was founded by leaving. The most American move is still to leave in search of a better life.

The informal sentence: a network state is a highly aligned online community with a capacity for collective action that crowdfunds territory around the world and eventually gains diplomatic recognition from preexisting states. Facebook and Twitter are not that. People do not think of themselves as Facebookers. A company Slack, by contrast, can get roughly 100 percent compliance on a task. Combine company-level alignment with community-level scale and you have something that can, at minimum, get a thousand of a thousand people to like the same tweet. Most of the Web 2 internet is entropic, a box of chocolates. Alignment is work: force along a distance, not heat.

Collective action scales from a thousand likes to a thousand people crowdfunding a building. In between come meetups, group houses, language-immersion apartment blocks, equity instead of rent. Territory is crowdfunded around the world, not in one place, the way Hawaii is American despite the water in between. "Cloud first, land last, but not land never," Balaji says. We still reproduce in physical space; the physical world is high bandwidth; a thicket of regulations, not a lack of bits, is what stops innovation in atoms. Online consensus is upstream of offline law. Burning Man can stand up 70,000 units in days because the people already agree. San Francisco cannot, because 50 percent will oppose whatever you try.

He contrasts 51 percent democracy — which he calls 49 percent dictatorship, because the other side did not consent and campaign promises are non-binding — with 100 percent democracy: build consensus first, then occupy a godforsaken patch nobody fights over. Starlink has repriced remote land. Startup cities, in his picture, look more like the one-story Wild West or Burning Man than like permitted skyscrapers: horizontal sprawl, "HIMBY" rather than NIMBY or YIMBY, snapping power and water to the grid. Diplomatic recognition sounds like science fiction until you notice sovereigns already making deals: Nevada and Tesla, El Salvador and Bitcoin, Wyoming and DAOs, Tuvalu and .tv, mayors taking salary in BTC.

The historical templates are Israel's path from Herzl's book to a fund, a congress, and crowdfunded territory; India's nonviolent independence; Singapore as a founder-CEO city-state; and the United States as a nation of immigrants. All four, he notes playfully, are forks of the UK codebase. The social smart contract is version three after UK common law and the US Constitution. Only about 2 percent of the world can become president of the United States; 100 percent can become president of a network state. Vitalik Buterin, in this comparison, actually built a currency from scratch in his twenties, which Balaji ranks above inheriting the Federal Reserve. Founders combine legitimacy and competence. Heirs often have title without the ability to rebuild the factory when it has to make masks instead of widgets.

Social networks sit on a cave-and-commons continuum. Twitter is almost all commons, a library crossed with a civil war. Facebook is a warren whose privacy settings you should treat as public. Reddit's subreddits, each with its own on-topic and off-topic, look to him like precursors of digital societies that will later become physical. Discord is transitional: chatty, not where the work or the karma lives. Slack is identity-heavy; Discord is pseudonymous. That difference is the point of a talk he gave years ago on the pseudonymous economy.

Anonymous is 4chan: no persistent name, no reputation. Pseudonymous is most of Reddit: a handle with karma, unlinked to your state name. "Real name" is a misnomer for a Social Security name, an AirTag, a file-cabinet handle billions of people can pull. Seeing Like a State, he notes, traces surnames to making humans legible for conscription. A state name is a foreign key into Gmail, Facebook, and every government database. A pseudonym is a new node far from that graph. Satoshi instantiated Bitcoin Talk and bitcoin.org on a name the government did not issue. Kids already run Finsta and Rinsta: the indexed Sunday-best self and the search-resistant real self. Cryptocurrency extends that pattern from communication to transaction.

The slogan is the opposite of bring your whole self to work: bring only your necessary self. Blind auditions, ban-the-box, and implicit-bias research all point the same way. Pseudonyms take unconscious bias off the table and let people who speak with accents type without them. Fridman worries that a cancelable, lifelong identity is part of being fully human, and that a murderer might hop societies. Balaji's grim reply is that twentieth-century democide was organized by states using lists of real names — Lenin's hanging order for kulaks — and that R.J. Rummel's ledgers suggest organized murder dwarfed disorganized murder. Tracking prevents some crime and facilitates others. Centralization and decentralization overcorrect into each other; the Three Kingdoms line still applies: the empire long united must divide.

Exit, from Hirschman's *Exit, Voice, and Loyalty*, is the route of the politically powerless, not a rich person's toy. Voice is a vote, a bug report, a board seat. Exit is migration, a competing product, selling shares. The internet raises microeconomic leverage: Lyft versus Uber, the back button, comparison shopping. You can configure every website and cannot configure usa.gov. Digital nomads combine hunter-gatherer mobility with farmer-soldier consistency. Anonymity, Balaji grants, can make young people cruel; persistent pseudonyms inside a subreddit's norms are a different technology. Small groups are becoming more productive — Notch, Satoshi, early Instagram — while large artificial groups split on scissor statements. Lean into the small ones.

Fridman loves Wikipedia and is about to have its founder on the show. Balaji draws the earlier distinction again. On technical truth, Wikipedia is great. On political truth, he calls it a defamation engine. A scam token, HBZ, edited Wikipedia to name him as founder; Google boosted the page; people lost money. Biographies of living persons, he says, should let people delete profiles the site cannot actually police. Early-2000s skepticism of Wikipedia versus Britannica has inverted into over-trust, and the more trust an institution gets, the more it tends to abuse it.

Google once prototyped Knol, forked pages with permissions, so the Israeli-Palestinian conflict could have ten versions instead of an edit war. Senior Wikipedia editors often date from before most of India and Africa were online, an inherited oligarchy. A recurring tactic, discussed on Wikipediocracy, is planting a story in the press and then citing it as a "neutral" third party. The Reliable Sources perennial list colors outlets green, yellow, or red: Al Jazeera green, Cato yellow, The Nation green, National Review yellow, CoinDesk red, social media largely unusable. You cannot cite Bezos's own tweet; you must cite a green-coded recap. That is, Balaji argues, a structural shift of every article toward US mainstream media at the moment public trust in that media has collapsed.

Fridman keeps asking for net effect, not anecdotes, and defends Wikipedia as a calm technical refuge. Balaji steelmans him: if even the constants become variables, people fear anarchy and QAnon. The upward deviation from a gossamer of quasi-truth is not "there is no truth" but Satoshi-andon: decentralized cryptographic truth. Israelis and Palestinians, Democrats and Republicans, already agree on the Bitcoin ledger — hundreds of billions of dollars without weapons. A ledger of record would put more facts partially on-chain: proof of location, proof of human, proof of solvency. Written history is fragmentary; Göbekli Tepe rewrote timelines; an Onion piece about historians in the year 3000 worshipping Queen Britney is a warning about reading rubble. Cryptographically verifiable history, he thinks, is the next step after writing, if the archive problem can be solved.

Wikipedia, media, and academia fail the same way: authority substituting for checks. Capital-S science is Maxwell's equations, tested every time someone speaks into a microphone. Quote-unquote science is a paper from last week, cited because Stanford and MIT are on the byline, with zero independent replications. Journal prestige and professorial rank were supposed to be a superstructure for making experiments reproducible. The superstructure now dominates the thing it was built to serve. Goodhart's law applies: when a measure becomes a target, it ceases to be a good measure. Satoshi's proof of work is rare in that knowing the rules does not let you game them without buying compute.

Western civilization, he says, has a break-glass button called decentralization. Martin Luther hit it. Mask guidance that flipped without acknowledging uncertainty spent down institutional credibility for "political health." Crypto-science would put code, data, and PDFs on-chain so a paper regenerates when you hit enter — the reproducible-research program Clair and Donoho pushed twenty years ago. Citations become import statements, composable science the way DeFi is composable finance. Anyone in Brazil or Nigeria can run internal consistency checks, Benford's law included. Instruments can hash a chain of custody. Funding can come from tokens and NFTs rather than a gerontocratic NIH whose average principal-investigator age creeps forward one year per year.

Fridman worries that unit-tested, importable papers help incremental science and hurt revolutionary fuzziness — Einstein's 1905 papers were not dashboards. Balaji's counter is Python's batteries-included library: imports let you fly. NSF and NIH are not necessary conditions for science; Newton died in 1727. Satoshi was not a professor. Project-evidence.github.io published the lab-leak hypothesis when it was taboo, a return to pseudonymous publication so the argument is attacked instead of the person. François Chollet's claim that in ten or twenty years every branch of science will be computational is, to Balaji, already almost true. Soft subjects harden when their logs move on-chain. History of the 2010s, he says, is in large part the log files of Twitter and Facebook.

Balaji was considered for a senior FDA role in the Trump years. He believes it will be easier to start a new country with a different biomedical policy than to reform the agency, just as it was easier to create Bitcoin than to reform the Fed. Regulatory harmonization lets US regulators — FDA, SEC, FAA — impose a monopoly on small countries the way Facebook Login captures small apps. Career bureaucrats have tenure, no public names, and no market discipline. During COVID, the FDA delayed tests so badly that some labs committed civil disobedience; the *New York Times* can spank the agency, but the public cannot name the people inside it. Challenge trials in healthy volunteers, he argues, could have produced vaccine evidence in days and saved a large number of American lives, especially seniors. The axis that mattered was not lockdown versus no lockdown but challenge trials versus no challenge trials.

Version one of medicines included Banting and Best, who iterated insulin on dogs and themselves, won a Nobel within a few years, and watched Eli Lilly scale North America — pharma at the speed of software, willing buyer and willing seller. An FDA filing he quotes from 2010 denied a generalized right to bodily health, to particular foods, or to freedom of contract. Exits already exist: right-to-try laws, CLIA labs, compounding pharmacies, off-label prescribing, and unharmonized countries. Kobe Bryant went to Germany for stem cells. Safety testing, with rare exceptions such as TGN1412, is not that hard. Cloud regulators — Amazon stars, Uber ratings, Gmail spam filters — already ban bad actors and rank quality across borders. Uber, he says, is a better regulator of rides than taxi medallions. A better MedWatch would be a global Yelp for drugs you get by scanning a barcode, not FDA form 3500B.

Daniel Carpenter's *Reputation and Power* records a Genentech executive comparing the FDA to a Brazilian soldier with a machine gun: they own you body and soul. Regulatory Affairs and IP become the real departments; R&D is downstream. TSA security theater is the public analog of FDA safety theater, except biotech CEOs go through the aperture under threat of months of delay that kill the company. An FDA commissioner cannot hire and fire like Steve Jobs; the Douglas factors make federal employees nearly unfireable. Fitness, diet, CGMs, and supplements are the back door, because the agency's writ is diagnose-and-treat, not better-than-normal. Ambulatory medicine (dentistry, dermatology) is capitalist and functional; ambulance medicine is where people reasonably fear the gurney. Shift more care upstream with continuous diagnostics. India, with cash payment, generics, and a national telemedicine stack, is a large enough parallel market to poke the monopoly from outside.

The motivation for a network state, for him, is longevity. Crypto is to finance what longevity is to current medicine: it inverts load-bearing assumptions. Fiat economics calls a little inflation good; Bitcoin calls deflation good and gives you root over your money. Fiat medicine calls a little death good — die with dignity — while Sinclair-style healthspan research treats aging as a programmable process. Life expectancy is a ratio-scale civilizational metric. If a frontier exists, long lives need not freeze the gerontocracy; people can exit. Genomic reincarnation is his sci-fi backup: print chromosomes, spend community karma, hand the next Lex a manual. Myostatin-null children and mice look like X-Men. Lance Armstrong's chemists, taken out of the race and into oncology, are the reframe: cheating at a game is not the same as cheating death.

Fridman asks for a steelman of Trump's removal from social media. The case for: a would-be dictator, a coup atmosphere, January 6, a megaphone that could organize more riots, executives acting within the law to defend democracy. The cases against come in several tribes. Trump supporters point to a *Time* article on a "secret bipartisan campaign" that saved 2020 and read it as corrupting democracy in democracy's name, including an ability to stand down BLM unrest. Libertarians ask whether giant corporations should decide who may speak, and whether simultaneous action by Twitter, Facebook, Google, Apple, and Amazon was collusion — a business plot in reverse, networks over the state. The rest of the world watched a digital decapitation of the supposedly most powerful man on earth. If they can do that in public to the US president, Balaji says, the Mexican president stands no chance. The seal is broken, the way bailouts went from shock to routine instrument.

He maps a "futures are our past" unmixing after the centralized century around 1950: frontier closes then internet opens; Spanish flu then COVID; robber barons then tech billionaires; Ida Tarbell beats Rockefeller, founders this time may beat journalists; gold as elite tool then digital gold as populist tool. America looks less like one nation under God, indivisible, than like two nations whose graphs do not marry. India, once 562 princely states, is still centralizing while the US decentralizes. A sci-fi scenario he will not assign high probability: losing to an ultra-nationalist manufacturing peer abroad — CCP China, four-to-one in population, orders of magnitude ahead in merchant shipbuilding — plus financial secession at home around Bitcoin. A hot war with China, or fighting Russia and China at once after a Pelosi Taiwan landing he calls unwise, is a failure of deterrence. Competence is stability, not smashing Humpty Dumpty and calling Ukraine a security-assistance success.

On platforming "dangerous" men, he prefers Herbert Matthews, John Reed, Edgar Snow, Walter Duranty, and David Halberstam to Hitler as examples: communist dictators who each had a personal journalist, several at the *Times*, whose geographies helped revolutions that then impoverished and killed. Communists, he says, do not understand self-interest; nationalists do not understand other-interest. Interviewing Trump, Putin, or Xi in a saturated information environment is different from being the sole source on 1917. Study the knowns to infer the dark matter: Sulzberger is as structurally powerful as Zuckerberg and far less visible. The school-of-fish strategy means a bureaucracy is never wrong because it is all wrong together, while a named deviant is hammered.

Woke capital, communist capital, and crypto capital are his three poles: NYT, CCP, BTC. CCP says bow because they are powerful. Woke capital says bow because you are powerful and therefore guilty. Crypto capital says head up, be sovereign. Each pole is bad at the extreme and bad in its absence. Word frequencies in the *Times* — mansplaining, toxic masculinity — go vertical around 2013, after tech took tens of billions out of media revenue and helped reelect Obama. Wokeness, in this telling, is Foucault plus deconstruction plus civil rights, a glossary that embeds the verdict. He expects a tack from wokeness toward American statism, then a new split: dollar versus Bitcoin, orange versus green, not only red versus blue.

Balaji is a medium- and long-term crypto bull even at a cycle low, in part because inflation can send many things toward a million dollars, and in part because a tripolar world — US establishment, CCP, and a web3 "third world" of Indians, Israelis, American libertarians, and Chinese liberals — needs a platform that is not Twitter-censorship or WeChat. Social media, analogized to pre-1978 China, still takes the grain: you earn little or nothing, you have no digital property rights, and a corporation can unperson you with a key. Xiaogang farmers signed a secret contract to keep some of their harvest, under death penalty, and got a bumper crop. Deng did not flip the whole country; he fenced off Shenzhen. Rebranding swaps the front end; reinterpretation swaps the back end and keeps the logo. Crypto is the property-rights upgrade: ENS names you hold with keys, social recovery against hacks, cloud collaboration plus local control. That is his version three again.

Crypto is not an asset class but a way to rebuild backends: finance, then search via block explorers, social via Farcaster-like networks, messaging via local keys, operating systems via the EVM, phones that hold currency and passport so neither Beijing nor a false-positive CSAM scanner at Google can lock you out of life. International deals between Brazil and Bangladesh can go peer-to-peer on a chain instead of through a US legal adapter. Centralized AI, he argues, is inherently unrepresentative; the ethical fix is open models like Stable Diffusion, so Nigerians and Indians can make their own Lord of the Rings rather than rent a "representative" API. Deplatforming began as a corporate weapon against API competitors, then fused with politics. AR glasses — Snap spectacles plus Glass plus Quest, outdoor form factor, tap for Terminator vision or full VR — are, to him, as predictable as the iPhone. NFTs become visible sigils and door keys. Networks get smaller and deeper: from eyeballs to DAUs to holders to netizens whose primary identity is a digital passport. That full stack is a network state.

Advice to a younger self, or to a student: become a full-stack engineer and a full-stack creator. Computer science and statistics are to this century what physics was to the last, because so much of waking life is already a constructed digital world whose physics programmers set. Learn theory and practice — Big O and software engineering, distributions and data science — deeply enough to write the thing, then use libraries. On the other side, own distribution. Establishment media corporations are competitors, not patrons; he recites *Times* history from Duranty's Pulitzer to the 2017 "The Truth" campaign, which he likens to *Pravda*. If you do not tell your own story, it will be told for you. Build alternatives: Bitcoin to the Fed, DeFi to Wall Street, ledgers to academia, decentralized AI to Hollywood, network states to political gerontocracy. The internet is to the Americas what the Americas were to Europe — a new world. By about 2030 most English speakers online may be Indian; they type without an accent. That is not anti-American. It is post-American, the way India and Singapore are post-British: respectful, sovereign, moving on their own axis. A digital bill of rights would include encryption, compute, repair, and portability. Europe, he jokes, having lost the AI race, now seeks to regulate logic itself — "step away from the if statement."