Silvio Micali: Money Is a Shared Belief, and a Blockchain Makes It Common Knowledge

Silvio Micali, a computer scientist at MIT, Turing Award winner, and founder of Algorand, opens with a definition so plain it almost sounds too small. A blockchain, he tells Lex Fridman, is a common database — a distributed ledger. Anyone can write an entry on a page; anyone can read it. The extraordinary part is the guarantee that everyone holds the same copy. Whatever you see on page seven is what everyone else sees on page seven.

He calls that common knowledge a first for humanity. Communication can now fling a photo around the planet in an instant, but you still cannot be sure that everyone received the same image. A ledger that nobody can erase, tear, swap, or alter is different: it is an immutable common record. If it is not decentralized, he asks, how can it be common knowledge at all? If only one person or a few people hold the book, you have to ask them what is on page seven, and you have to trust that they tell everyone the same story.

He offers an auction. You have built a building and want to tokenize it and sell it in parcels worldwide. On a shared ledger, everyone sees the bids, and everyone knows that everyone else sees the same bids, so a fair price can actually be reached. In a centralized system, an announcer can congratulate you on winning at $12,570 and you have no independent way to know. Cryptocurrency, in the same frame, is simply currency written on that ledger: Micali writes that he is giving a hundred units to his sister, and because the entry is public she can later spend fifty-seven of them. The merchant does not have to wonder whether the check is covered. What you see is what the merchant sees.

Fridman asks the elementary question: what is money, and why did humans invent it? Micali answers that money is a social construct, a shared belief system that lets people transact even when they want different things. Two sheep and one cow and a demand for blankets is the textbook barter mess; money simplifies it. But even barter, he insists, already rests on faith. If a mason builds you a wall and you pay him a thousand sheep, he cannot eat them all. He has to feed them. If nobody else will take the animals in exchange, they die and the value goes to zero. Livestock is already a bet that someone else will accept them.

Gold does not escape the same logic. It has industrial uses and it does not oxidize, but the price at which it trades is not that industrial value. An ounce of gold is another way to express the belief that somebody else will want it. Fiat is the same bet, only now it is a bet that a country will not inflate the currency into nothing. Micali is not intellectually bothered when money is not tied to a physical good, because the physical goods were never the real story.

Scarcity, he grants, is very useful. If daisies were declared money and you could cultivate them at will, you would be a fool to accept a bouquet as payment. You need some inability to create units from nothing. Bitcoin ties that scarcity to proof of work, so minting is physically hard. Scarcity is not an intrinsic necessity of money, but it gives the mind a handle: you know you are receiving this much of a fixed pie, and therefore you know what you have been paid.

Micali splits a blockchain’s goals into scalability, security, and decentralization, and he rejects the industry trilemma that says you can have only two of the three. Scalability is how fast you can write, share, and inspect entries on the common ledger. One transaction per hour and the world does not go around. Credit cards already average about 1,600 transactions per second and peak over 20,000 or 40,000. Printing a private pile of transactions on your own printer does not count, because nobody else has time to see or inspect them. Value lives in what is shared and visible.

Security against tampering, he says, is the easy part. Cryptographers can use tools invented fifty years ago — prehistory in cryptographic time, cavemen around a fire — to make pages unerasable and unalterable. The fundamental problem is almost social, almost political: who the hell chooses or publishes the next page? Whoever assembles the pending transactions and appends them would have more power than any absolute monarch in history. That person would decide which transactions the entire world should see, and which never see the light of day. No one in history has held that power. The quintessential blockchain problem is to make that “somebody” not be a somebody.

Bitcoin’s answer is a hard cryptographic puzzle. The first to solve it may add a page on behalf of everyone else. Sometimes you solve it first, sometimes I do, and the effort you put in is supposed to correlate with how much trust you deserve. To keep two solutions from arriving at once — a double page, with no one sure which is real — Bitcoin aims for one solution every ten minutes, no matter how many people on Earth are trying. That forces the riddle to become extremely hard. A chance of winning inside ten minutes requires thousands of specialized machines that throw off heat like maniacs and need air conditioning galore. Costs climb until fewer and fewer people can compete. Depending on the day, Micali says, two or three mining pools are the ones capable of controlling the chain. It started decentralized. Expense recentralized it.

Delegated proof of stake, in his telling, is even plainer: pick 21 people who look honest and entrust them to add pages for the rest of us. Twenty-one is better than one, he says, but it is still very little. When people rebelled against centralized power — the French Revolution is his example — there was a monarch and there were nobles. Were there 21 nobles? No. There were thousands of them, and millions of disempowered citizens. One is centralized. Twenty-one is also centralized.

Algorand, the company of cryptographers, engineers, and mathematicians Micali founded in 2017, takes a different cut. Imagine on the order of ten billion tokens, each with equal power, distributed around the world. By a cryptographic process that is mathematics rather than magic, a thousand of them are selected at random. The owners of those tokens form a committee, agree on the next page, and sign it. The role rotates quickly. If the majority of tokens sit in honest hands, the chance that a random thousand contains 501 bad ones is, he says, very, very improbable.

Every system, he reminds Fridman, assumes that a majority of participants is honest. The only difference is the majority of whom. Proof of work assumes a majority of mining power. Delegated proof of stake assumes a majority of those 21 people. In both cases the whole economy is asked to be secure if a small piece of it is honest. Algorand’s claim is that the whole economy is secure if the majority of the economy is honest. Subverting it would require a majority of token holders to conspire against an economy they themselves mostly own — a self-destructive majority.

Who performs the random selection? If a person does it, you are back to trusting a center. Algorand’s unorthodox answer is that the tokens choose themselves. Each user, on a laptop, runs a private lottery — pulling the lever of a slot machine once, not until they win. A winning ticket means the network should listen; without one, nobody pays attention. The lottery is cryptographic: even an entire nation with enormous computing power cannot improve, even minimally, the probability that one of its tokens wins. Running it takes about a microsecond, whether you hold one token or a billion, so it neither cooks the planet nor your computer.

The security argument is the one Micali is proudest of. Suppose you are so evil and so powerful that you can corrupt anyone on Earth instantaneously. You would want to corrupt the committee. The trouble is you do not know whom to corrupt — a woman in Shanghai, a sovereign guy in Paris? The winners are random, including to themselves, until they come forward with a ticket and an up-or-down opinion of the block. By then the messages have propagated virally. You can buy all thousand of them, but you cannot put a virally propagated message back in the bottle any more than a government could unsay WikiLeaks. Distributed, scalable, and secure, he argues, can be had together.

If Algorand is so clean on paper, Fridman asks, why is it not the dominant technology? Micali gives Bitcoin its due first. What it got right was seeing that the time was ripe for a cryptocurrency — being right is not enough; you have to be right at the right time. It also made subverting or canceling a transaction very hard, if not impossible. What it did not get right is that money wants to be transacted, not only stored under a mattress. Bitcoin, in his view, is a great brand as a store of value and a poor medium for the interactions people actually live on. Building infrastructure on top of those flaws, rather than starting from scratch, is difficult. When he decided to throw his heart in the arena, he chose to start from scratch.

Ethereum, he says, had another great idea: money and payments are only the first stepping stone, and the next level is smart contracts. People need a way to transact securely without being shopped around by a trusted third party. Mediators are hard to find if you live in Thailand and the other party lives in New Zealand, and even when you find them they want to be paid. He cites a figure that six percent of world GDP goes into financial friction — essentially third parties. Ethereum saw the need. Again, he argues, the scalability is not there: smart contracts on that system are slow and expensive, and not enough for the appetite he thinks the world has.

On NFTs he is interested on both the technical and the social side. A song or a piece of art has unique representations; the important part is now being able to trade those rights quickly, fast, and securely, knowing who owns what. That is a new way for content creators to be remunerated. Fridman pushes further: you can start investing in a YouTube creator, trading their work, even making a market out of people and ideas. As physical labor recedes, those digital economies gain power — but only if scalability, security, and decentralization still hold. Decentralization is the tricky one, Micali agrees, because people in power get nervous. Congratulations, he says: you have the top job, and now everybody wants it.

Can power actually be decentralized, given that power corrupts and absolute power is wonderful to hold? Micali splits the history. Grabbing power is often easier for a single person. Old technology also justified centralization: when even one-way communication was hard, “you do as I say” was the simpler alternative. Common knowledge changes that. He also believes distributed systems are more stable and durable, not only against corruption but against everything that goes astray. Give it long enough, and a stranger version of Murphy’s law applies: whatever goes wrong, goes wrong. Diffused power is more stable.

Living beings, he notes, are already distributed. Nobody issues an order that it is time for Silvio to eat. You have millions of cells and billions of bacteria in the gut, a soup that somehow keeps us alive. Yet when we design architectures we still draw pyramids, put someone on top, and let power flow down. Machiavelli, “an Italian competitor,” looked at the small republics of Florence and Venice beside the centralized Ottoman Empire and observed that a centralized government is hard to overthrow — but if you get it, keeping the population is easy. Distributed power is more resilient and lasts longer. The human spirit, Micali thinks, wants that resilience. Machiavelli simply did not have the computer or the internet.

That is why he rejects crypto projects that boast that the rules are fixed in stone: code is law, law is code, the code has been verified and will never change. To him that is a recipe for disaster, not immediately but soon. Imagine an ocean liner from Lisbon to New York that will not alter course for iceberg or tempest. You need a tiller. Algorand was designed so the code evolves with needs, with each adjustment requiring a vote — currently orchestrated at 90 percent of the stake — before the network picks up the next version. A blockchain is a perfect platform for that, because votes can be logged so everyone sees the same referendum. A system that wants to live long has to adapt. Nothing fixed lasts very long. Life is intelligent adaptation.

Leadership sits in tension with that. Fridman mentions Vitalik Buterin as a face of Ethereum, Satoshi as the faceless face of Bitcoin, Linus Torvalds of Linux: humans seek leaders who stabilize a vision. Micali agrees that a leader matters in the initial phase, when something is generated out of nothing, and that intimate emotions paradoxically become universal. The true test, though, is to disappear after you have led the community. His quintessential leader is George Washington: two terms, then retirement as a private citizen. Leadership ought to ignite and disappear. If you do not disappear, the system is going to die with you. Demanding another four years, then eight, then sixteen to “fix all your problems” is, in his view, failed leadership.

Asked what else a blockchain might revolutionize besides money, Micali starts with financial freedom and then moves to speech: write something on a chain and nobody can take it out. Transparency is not money, but it is an ingredient of finance. He is enthusiastic about decentralized finance, yet he insists that the first guarantee of financial growth is still the legal system. Courts look like boring lawyers until you notice that without them there is no certainty, no equality, no way to resolve disputes — and that is what thrives commerce. A ledger nobody can change automates a chunk of that trust. Technology need not be antagonistic to legal scholarship; it can coexist and then do the interesting things the economy alone cannot.

Fridman raises the tension with privacy. Micali, a cryptographer who has devoted a big chunk of his life to privacy even when it seemed almost impossible, says yes, it is possible on a blockchain too. He also believes in timing. People first have to understand the system they live in: an unalterable, transparent ledger, plus a vanilla sort of privacy from the fact that public keys are not name tags. You can split funds across keys until nobody knows which ones are Silvio. He will not rush stronger privacy tools until the public understands them rather than taking scientists on faith. First absorb the power of the chain; then build privacy on top of it.

He does not think Algorand, or anyone, should win a winner-takes-all contest. A few blockchains will each have a brand and be great at something — scalability, a set of uses — and they will need interoperability so assets can move to whichever tool best serves the need of the moment. Bitcoin maximalists bet that the best technology should dominate store of value and exchange. Micali quotes an Italian saying: appetite grows while eating. Bitcoin does store of value well, but a world that only stores value under the mattress would be a sad one. He longs for a world that wants to transact and interact. Something invented decades ago will not fulfill the needs of his kids, let alone theirs. Static loses. Intelligent life is adaptability.

Does Micali have any clue who Satoshi Nakamoto is? First, he says, it is not him, and he can prove it: if he had been Satoshi he would not have founded Algorand, which takes a totally different principle. The right answer, though, is not him or she or they. Satoshi Nakamoto is Bitcoin. Creator and creation identify themselves. You can list Michelangelo’s Sistine Chapel, St. Peter’s dome, Moses, and the Pietà, but asking who he was besides the work is the wrong question. Bitcoin is a piece of work with defects, like anything human, that captivated millions and subverted the status quo. Whoever that person or those people are, they live in the work.

Biology wants a face attached to an idea, Fridman adds, when the idea is what we love. Steve Jobs is Apple more than the man is Jobs; the sense of aesthetics he brought to daily life is what remains. Elegance, Micali notes, is supposedly something very few people care about, and yet millions do, because we are attacked by beauty. Fridman quotes Dostoevsky: beauty will save the world. On whether cryptocurrency will become the main method of storing and exchanging value in this century, Micali says the notion has to evolve — programmable money, smart contracts, peer-to-peer interaction among people who have never met — but yes, unless some existential crisis intervenes. Asteroids, nuclear war, and artificial intelligence make the joke list.

Fridman then asks for the most beautiful idea in cryptography, computer science, or mathematics, setting Micali’s own work aside. After a few contenders, he lands on the one-way function: given x you compute f(x) easily; given f(x) it is very hard to go back to x. Breaking a glass is easy, reconstructing it hard. Frying an egg is easy, unfrying it hard. Killing a living being, unfortunately, is easy; the other way around is very hard. You have the recipe in front of your eyes and still cannot reverse it. That, he says, is one of the most elegant and momentous notions there are — computational because of the difficulty, mathematical because it is a function, and the foundation of all cryptography.

Mythology already knew the pattern. The most powerful gods come in pairs of opposites, love and death; the opposites do not erase one another, they create something more powerful. A one-way function is easy for the good guys and hard for the bad guys. By the magic of reductions it morphs into pseudorandom generation, where bits must be produced quickly and the next bit must be unpredictable, and into digital signatures, where signing is easy and forging is hard. Easy and hard coexist in one object. That is the elegance.

People think a pseudorandom generator creates random numbers. Micali does not believe you get something from nothing. What you can do is expand secret randomness. Give him 300 truly random bits and he can give you 300 million, 300 trillion, 300 quadrillion bits. He tells you the recipe but hides the seed. Looking at all the bits produced so far, you should not be able to bet on the next one better than 50-50. He calls these machines expanders of secret randomness. Programmers who call a function `random` are usually doing that expansion, often with pre-modern cryptography; they would be better served taking a 300-bit real seed and expanding it properly.

Does God play dice? Einstein said no. Modern complexity-based cryptography, Micali argues, lets you sidestep the metaphysics. If the universe has 300 random bits, you can expand that seed forever so that every experiment and inference you could run fails to distinguish the sequence from the truly random. If you have no method, in the lifetime of the universe, to tell them apart, are they really different things? Intellectual honesty, he says, is to admit that this super-duper pseudorandomness is as good as random. True randomness, if it exists, is probably expensive to get — and if he hands you one of his sequences you will never tell them apart.

Free will gets the same treatment. His gut reaction is that it is important we believe free will exists, and that by a weird logic, if we believe it, it does. We so want it for our social apparatus that we almost conjure it into being. Genetics keeps shrinking the space of unexplained preference — why we like this music, that poetry — but whether free will exists in a philosophical sense is like randomness: if pseudorandom is as good as random vis-à-vis the lifetime of the universe, when does the distinction matter?

Quantum computing threatens some cryptographic assumptions. Micali is a little agnostic about whether quantum mechanics will actually accelerate computation, but he boils the fear down: does it kill the one-way function? “Easy” and “hard” were defined against classical machines. If quantum computers arrive, easy must mean easy for them and hard must mean hard for them. Then you can use quantum machines for the easy direction, and the candidates for quantum one-way functions multiply. Particular one-way functions may fall; quantum one-way functions may continue. He almost hopes they exist. If everything becomes easy, what kind of life is that? Life is interesting because hard problems pop up in abundance. Fridman connects the thought to cellular automata: simple rules generate complexity that is surprisingly hard to reverse, and that difficulty is part of the mystery that makes life fun.

Interactive proofs, Micali says, are a modern realization of something we already knew in the classroom. We invent schools and spend the treasury sending children to a blackboard rather than mailing them a book from the minister of education, because interactively you can convey truth much more efficiently. Psychologically it is better to hear an explanation than to belabor a paper. The mathematical version is this: if a theorem is true, there is a game associated with it in which the prover has a winning strategy. Play the game. If the theorem is true the prover wins all the time; if it is false, at most half the time. Win after win after win, and you either accept the theorem or accept that someone just flipped a hundred heads. A formal proof becomes a miniature game you can play quickly and judge by statistics — one of the big conquests of modern complexity theory, and a new insight into what truth and proof even mean.

Classically, verification is brutally inefficient. You have to check every line of a very long proof, because an error can hide in the line you skipped. Interactive proofs let you verify in minutes an assertion that would have taken a lifetime to belabor. Once you define a proof system as something that can convince you of a true statement and cannot convince you of a false one, the realm of what is provable enlarges. Elegance is optional; distinguishing true from false is not.

On top of that came zero-knowledge proofs: you can verify that an assertion is correct without having any idea why. Knowledge and verification, which people had treated as the same concept, come apart. You no longer need to possess the proof — and therefore understand why the claim holds — in order to check it. At first they labored to find one zero-knowledge proof, then a second, then a third. A few years later they proved a theorem that every theorem, no matter what about, can be explained in a zero-knowledge way. For thousands of years the identity of knowledge and verification had been taken for granted, for no reason at all, until the technology existed. Micali is “very big on technology” because it makes us more human and lets us understand more than before.

Mechanism design, which Micali has spent recent years on, is the art of wanting a social behavior to arise while knowing that people do not care about your goals. They care about maximizing their own utility. Ideally you design a game so that while they play to that self-interest they achieve the social goal you wanted. It is a hard science and a hard art. Game theory has compelling tools — dominant-strategy equilibrium among them — but they analyze a restricted class of games whose rules, constraints, and utilities are all set. Most games do not fall in that class.

His own style is coarser. Sometimes a game has no move that is best no matter what the others do, and that is too much to ask. He would rather design so that, looking at the options in front of you, the stupid ones are obvious and can be set aside. If in any combination of the remaining non-stupid moves the players still produce the social outcome he wants, he is done. He does not give a damn about finding a unique equilibrium. As long as nobody does stupid things, good social outcomes arise. Human behavior can be conceptualized in many ways. It is a long game.

Algorand, in that light, is a game designed to be scalable, secure, and decentralized — with almost no incentives at the core. Incentives, he learned, are very hard because people are complex. The technical notion he likes is an epsilon-utility equilibrium: if you are honest you make a hundred bucks, and if you are as dishonest as you like you make a hundred bucks and one cent. Technically the extra cent is an improvement. Practically, nobody bothers. Having no incentives can be a feature, because it stops people from reasoning about how else to game the system. Bitcoin has to pay miners because they do a tremendous amount of work. If generating the next block is as simple as adding two and two, Micali asks how much you want to be paid — and answers that you can add two and two in your sleep. Make consensus cheap enough and you need not incentivize it.

The conversation softens. Micali is Sicilian; Fridman asks for a last meal. In the Covid era he has not been able to see his mother, a fantastic chef of traditional food that survived hundreds of years of culinary innovation. He wants her parmigiana, a laborious act of love that takes the entire morning. Italian cuisine, he says, believes in very few ingredients. Pesto is olive oil, basil, pine nuts, pepper, a clove of garlic — not too much or it takes over — and Parmesan or pecorino or a mix: six ingredients. French cooking is combinatorial; some Chinese cooking uses many more. Italy strives for simplicity, the right few things. Japan’s sushi, he agrees, has the same mastery of restraint, propagated through generations.

The book that impressed him tremendously is Dante’s *Divine Comedy*, a medieval poem in hell, purgatory, and paradise: a middle-aged man in personal crisis who purifies himself until he can meet God. The sophisticated language for that in the 1200s would have been Latin. Dante chose his own Florentine dialect, spoken outside his immediate circle by almost no one, and in doing so he made Italian. Intricate rhymes let readers guess what unknown words meant and invent the language by osmosis — a miracle of communication in a poor, unsophisticated tongue. What Micali took from it is that limitations are our strength. Dante could not quote prior Italian books because there were none. Rather than complain, embrace the limit; people with fewer constraints would not even think of the solutions you find. Limitation is a kind of superpower if you choose to see it that way.

Coming to America, he knew he would not express himself at the same level in English, and that precision of expression is how elegance, knowledge, and education show. He found himself at the bottom and had to fight back up. Fridman, whose mother tongue is Russian, says he became a deeply thoughtful human in English, while the pain of World War II and the music of the people live in Russian; he cannot do philosophy in Russian or that melancholy in English. Micali understands completely. He cites Cortés invading Mexico with a few hundred men against a hundred thousand, sinking his own ships so there was no return. Advice for the young is to be true to your own emotion: a life spent untying the emotional knots that intrigued you from the beginning is a life you will never tire of, whether you are a scientist or an artist. Love is the big, scary part of that, full of vulnerability and of compassion that makes us one with other people’s suffering. The downs always come with the ups, and the ups only come with the downs.

Does he fear death? He hopes so. Without death there is no meaningful life; death is the ultimate motivator to live a beautiful one. As a young man he told himself that if he had something to say and could not say it, he would not go on — a way to motivate himself that death already provides. His legacy, as he hopes it, is twofold: a biological one in two children and their children, and a social one in providing security and privacy. Those words are metaphorical. He wants people to interact more, take more risks, and reach more people, difficult and dangerous as that is. His scientific work is about guaranteeing privacy and the security of interaction, not only of a blockchain transaction. Privacy and security are freedom, and freedom, like emotion, is dangerous — and that is how we create the beautiful things around us. The meaning of life, he says, is three things: to seek, to seek, and to seek. There is no answer to “seek what.” The journey is more important than the destination. We may never get there. Hell, it was a great ride.