Why Bitcoin Can’t Fix Itself

A Gene You Can’t Repair From Inside the Cell

Imagine an organism. It has DNA. The DNA codes for proteins. The proteins assemble cells, the cells assemble organs, the organs make the organism. If there’s an error in the DNA — say, one letter is wrong — that error is reproduced in every cell, because every cell copies the same DNA. The consequences can range from imperceptible to catastrophic, but one thing is clear: no cell inside that organism can fix the DNA error. For that you need someone outside, with different tools.

That’s exactly Bitcoin’s situation right now. It has its original construction — Proofof-Work with anonymous miners and an open protocol. That construction solved Satoshi’s specific problem — to create digital money with no bank in the middle. And it solved it brilliantly. But it has one side effect: anything earned inside the system can be used outside it, and the system has no mechanism to defend itself against how that earning changes the system itself.

I’ll try to show this on three different layers. First — on the layer of computational power, where money earned from mining buys still more mining and concentrates in a few hands. Then — on the layer of ownership, where the same money, having passed through exchanges and funds, comes back as large institutional positions. And finally — on the layer of development, where the same money pays the engineers who write the protocol itself.

Three layers. The same mechanism. And in each of them — a structural contradiction between what Bitcoin promises in its original construction and what actually comes out.

Layer One: Who Counts

Let’s start with mining.

When Satoshi launched Bitcoin in 2009, he seems to have sincerely thought that mining would be done by thousands of people on home computers, each one independently. “One CPU — one vote,” he wrote in the white paper. One processor, one vote. Every participant in the network defends it with their small bit of power, and no one has an advantage over anyone else. In 2009–2010 that’s how it actually worked — Satoshi himself mined on an ordinary laptop, taking the whole block reward. That picture died very quickly. In 2010 someone figured out that graphics cards could be used for mining — they hash dozens of times faster than processors. In 2011 the first FPGA miners appeared, several times faster again. In 2013 Bitmain released the first ASIC — an integrated circuit honed for one single operation, SHA-256 hashing. The ASIC was a thousand times more efficient than a GPU. Home computers were out of the game. So were graphics cards. Only ASIC farms remained.

By 2016 most mining had moved to China, where electricity was cheap and local authorities weren’t interfering yet. By 2021 China had banned mining, and the industry relocated to Texas, Kazakhstan, Russia, Iran, Iceland. By 2024 — after another “halving,” when the block reward was cut in half from 6.25 to 3.125 BTC — the small operators were squeezed out for good. Only the large industrial farms survived: the ones with long-term contracts for wholesale electricity, their own cooling infrastructure, and teams of engineers.

As of April 2026 the picture looks like this.

At the very top — mining pools. These are associations of miners that combine their computations and split the reward. Of all the world’s miners working on Bitcoin, Foundry USA serves about 29 percent. AntPool (owned by Bitmain, the largest ASIC manufacturer) serves another 21 percent or so. ViaBTC — about 11. F2Pool — about 10. SpiderPool — about 7. MARA Pool and SecPool — 6 each.

The first two pools together control roughly half of all Bitcoin hashrate. The first four — about seventy percent.

This is a fundamentally important number, and it’s worth pausing on it for a minute. When a pool controls a certain share of the hashrate, it essentially decides which transactions go into the block. The pool doesn’t actually mine the blocks itself — that’s done by the miners connected to it. But the pool builds the “block template” — the list of transactions the miner then tries to hash. If the miner happens to find the right hash, the block is published with the transactions the pool selected.

In other words: four organizations — Foundry, AntPool, ViaBTC, F2Pool — decide which transactions get into seventy percent of all Bitcoin blocks. This isn’t a threat to the network in the sense of “they could destroy it.” It’s much more subtle: they can slowly, transaction by transaction, decide who pays more for the right to enter a block, whose transactions get processed first, and whose get stuck. They can (and sometimes do) exclude transactions from addresses under sanctions — for example, addresses on the U.S. Treasury blacklist. They can (and this has already happened) prioritize certain kinds of transactions that bring in higher fees.

And here’s the thing — they can do all of this without breaking a single protocol rule. They’re simply taking advantage of the fact that they have a lot of hashrate. And they have a lot of hashrate because they earned it on previous blocks and reinvested into more equipment, cheaper electricity, and better engineers.

The game rewards winners with money. Money buys more equipment. More equipment gives more hashrate. More hashrate — more wins. The cycle is closed. Notice that this cycle violates none of Axelrod’s conditions. The three conditions — iteration, memory, transparency — are still met. New blocks come out every ten minutes (iteration). Everything is recorded in the blockchain forever (memory). The hashrate of every pool is publicly visible (transparency). Everything’s in order. But inside this perfectly functioning architecture a process of concentration is going on that isn’t accounted for by the rules of Tit-for-Tat. Because TfT is a strategy for equal players. It says nothing about what to do when one of the players is twenty times bigger than another. In Axelrod’s tournament all programs had equal starting conditions. In real Bitcoin — they don’t.

Layer Two: Who Owns

Now let’s look at another level.

Bitcoin isn’t only miners — it’s holders. People who simply own coins and store them on their own wallets or on exchanges. In theory, since every bitcoin is identical, distribution should be uniform. In practice — it isn’t.

In January 2024 the U.S. Securities and Exchange Commission (SEC) approved spot Bitcoin exchange-traded funds (ETFs) for the first time. This was the culmination of years of pressure from large asset managers — BlackRock, Fidelity, Grayscale — who wanted to be able to sell bitcoin to their institutional clients through ordinary brokerage infrastructure, without having to deal with crypto exchanges and cold wallets. BlackRock is the world’s largest asset management company; it has on the order of 11 trillion dollars under management. That’s more than the GDP of any country except the United States and China. When BlackRock launched its Bitcoin ETF under the ticker IBIT on January 11, 2024, money poured in like a river.

As of April 2026 — two years later — IBIT holds roughly 800,000 bitcoins, which corresponds to about 63 billion dollars at the current price. That’s about 49 percent of all assets in all U.S. spot Bitcoin ETFs combined. In the first quarter of 2026 alone, IBIT took in net inflows of 8.4 billion dollars.

But IBIT isn’t the only large holder. The company Strategy (formerly MicroStrategy, a business-analytics firm that then reinvented itself as a pure Bitcoin treasury) as of April 17, 2026 owns 815,000 bitcoins with a total acquisition cost of 61.56 billion dollars. That’s about four percent of all the Bitcoin that has ever existed. No single entity in history has held that kind of share — except, possibly, Satoshi himself, whose coins haven’t moved since 2010.

If you add up everything held by U.S. spot ETFs, you get about 1.31 million bitcoins — roughly 6.2 percent of total supply. Add Strategy — you get more than ten percent. Add the large exchanges (Coinbase, Binance), corporate treasury positions, large private holdings — and it turns out that several dozen entities account for a quarter of all Bitcoin.

What does this mean in practice?

It means that if BlackRock decides one fine day, on a regulator’s order, to freeze IBIT’s assets — eight percent of all Bitcoin instantly becomes illiquid. That’s a market shock of a magnitude Bitcoin hasn’t seen yet. It means that if Strategy faces a margin call on its loans (and it has been borrowing to buy bitcoin), it could be forced to liquidate 815,000 coins — which would crash the price several times over in an instant. It means that if the U.S. government decides to regulate Bitcoin with new legislation, the main channel of influence is no longer decentralized miners in Kazakhstan but BlackRock in New York, where one phone call to the right office can change the behavior of trillions of dollars.

The paradox is that Bitcoin was conceived as a network resistant to centralized pressure. Decentralization was not a side effect, it was a central requirement. And yet sixteen years later we have a construction in which a few large institutional players control such a share of supply that they are structurally able to influence price, liquidity, and the regulatory agenda more than the rest of the market combined. How did this happen?

The same way it happened with mining. Each individual movement of money was voluntary and individually rational. A pension fund wanted exposure to Bitcoin but couldn’t hold the coins directly because of regulatory restrictions — bought IBIT. A family office wanted diversification — bought IBIT. A corporation wanted a hedge against inflation — bought IBIT. No one was coerced, no one was manipulated. Each individual decision looked sensible.

But the aggregate result of millions of such individually rational decisions is gigantic concentration. Which now itself becomes a factor in the equation. Because the more bitcoin BlackRock has, the more voice BlackRock has in any public discussion of what Bitcoin should be in the future. And that voice isn’t the same as an ordinary holder’s. It’s a voice that gets listened to at the SEC, in Congress, at the Federal Reserve. Layer Three: Who Writes The third layer is the most invisible, but the most important.

Bitcoin is code. That code is written by people. There aren’t very many of them. According to a 2024 BitMEX Research report, on all of Bitcoin — which at that point was worth about a trillion and a half dollars — 41 developers were actively working: those who made at least five commits to the main Bitcoin Core repository over the year. Forty-one people, on a trillion and a half.

Of those forty-one, about two-thirds live in the U.S. and Europe. Three developers — in Latin America (Argentina, Brazil, El Salvador). Four — in the rest of the world. That’s Tier 1 statistics, drawn from public commit data.

Who funds them?

Bitcoin has no foundation. No central organization. No proverbial treasury that pays salaries from. That was all by Satoshi’s design — so that nobody could halt development by pulling funding. But programmers have to work, and they get paid through various sources.

Blockstream. A company founded in 2014. Among its founders — Gregory Maxwell, Pieter Wuille, Matt Corallo, Jorge Timón, Mark Friedenbach (all active Bitcoin Core developers), as well as Adam Back (the inventor of hashcash, on which Satoshi built Bitcoin’s PoW) and venture investor Austin Hill. Since its founding, Blockstream has raised more than half a billion dollars in investment in total — $76 million by 2017, $210 million in a Series B in 2021, another $210 million in convertible notes in 2024. The company clearly has several key Core developers on its payroll. Adam Back, its president, explains it this way: the company has an interest in seeing Bitcoin supported, developed, and scaled.

Chaincode Labs. A research center in New York that funds several more key developers.

MIT Digital Currency Initiative. An academic cover at the MIT Media Lab; it started up around 2015 in response to the funding crisis among Core developers after the collapse of the Bitcoin Foundation.

Brink, OpenSats, Spiral, Human Rights Foundation. Nonprofit grant programs. Large exchanges — Coinbase, OKCoin, OKX. Periodically issue grants.

The total annual funding for all Bitcoin Core development is estimated at 150–200 million dollars a year — for a network worth more than a trillion. By comparison, the Ethereum Foundation in 2024 spent about 50 million on development, with a market cap ten times smaller than Bitcoin’s. Polkadot — 16 million, with a market cap a hundred times smaller. Per dollar of capitalization defended, Bitcoin underfunds development by an order of magnitude.

And — here’s where it gets subtle — a disproportionately large share of the modest funding that does exist flows through Blockstream. A company that has its own commercial products: Liquid (a sidechain for financial institutions), the Lightning Network (an L2 for fast payments), mining hardware, institutional services. The interests of these products sometimes coincide with the interests of an ordinary Bitcoin user, and sometimes they don’t.

A classic example is the 2023–2024 debates over the Ordinals protocol, which made it possible to write arbitrary data (images, NFT collections, even books) onto the Bitcoin blockchain. Technically this was done through Taproot — a protocol upgrade that Core developers shipped in 2021 for other purposes, but which Casey Rodarmor used in early 2023 in an unexpected way. Ordinals quickly took up a significant share of block space, driving up fees and changing the network’s economics.

One of the senior Core developers, Luke Dashjr, submitted a pull request that would have technically restricted Ordinals. Another maintainer, Ava Chow (a former Blockstream employee), closed the PR, citing “lack of consensus and creating noise.” The community split. Some considered Ordinals a creative use of the network. Others — parasitism on block space meant for monetary transactions. The decision was, formally, made by one person. That decision could not be reversed by anyone except the maintainer himself or another maintainer with equal rights.

Ordinals didn’t break a single Bitcoin rule. But they radically changed how the blockchain is used. And that change happened not through a vote of miners, not through a referendum of holders, but through the individual decision of a few developers with commit access. Who get paid by specific companies. Who have their own business interests.

Formally the network remains decentralized — it has no CEO, no foundation, no board of directors. In fact, a narrow circle of people, financially tied to a limited set of companies, makes decisions that affect trillions of dollars of capital and millions of users.

And again — this is no one’s evil intent. It’s just a consequence of how the development economy is set up. Bitcoin doesn’t pay developers out of its blocks. They are paid by external companies. The companies have interests. The interests concentrate. The developers fed by those companies cluster around the tasks those companies consider priorities. Regardless of how well those tasks line up with Bitcoin’s tasks as a public good.

The Three Layers Together

If you look at the three layers at once, a common structure is visible. On the mining layer — money earned in previous blocks buys more hashrate. More hashrate gives more influence over which transactions go into blocks. On the ownership layer — money earned from mining or from price appreciation buys more coins through institutional instruments. More coins gives more voice in the public discussion of Bitcoin’s future.

On the development layer — money earned at exchanges, at ETFs, at mining infrastructure, funds developers. The developers who receive that money decide where the protocol moves.

On each of the three layers, the same mechanism is at work: an internal win is converted into external influence. And that influence, in turn, can be used to change the conditions of the next rounds of the game in one’s own favor.

This is where Axelrod’s model breaks. Not because Tit-for-Tat is a bad strategy. TfT is still optimal for the situation it describes. But it describes a situation in which points cannot be converted into the right to change the rules. Bitcoin is a situation in which they can.

Axelrod in 1980 couldn’t have foreseen this, because in his tournament there was no money. There were abstract points that couldn’t be used for anything except tallying up the winner. Whereas in a real economic system, the winnings can always be converted. The only question is into what, and how much.

Under the Hood. GHash.IO Once More

In the first chapter I told the story of GHash.IO in the summer of 2014 — a pool that for a day crossed the 55-percent threshold of Bitcoin hashrate and didn’t catastrophically attack the network. In the second chapter I used that story as an example of how Tit-for-Tat works in reality — the community mirrored the threat, the pool retreated, everything ended well.

Now — the whole truth.

GHash.IO did not destroy Bitcoin. But it did use its majority position for a specific attack on a specific service. In the autumn of 2013, when GHash.IO’s share was already significant but not yet 51 percent, the pool conducted a series of double-spending attacks against the online casino BetCoin Dice. The discussion of this on the BitcoinTalk forum, in thread number 327767, is still publicly available. The idea is simple: GHash.IO accepted losing bets at its addresses, and when it lost — it mined the same transaction with its own (winning) address into a different block, turning the loss into a win. The digital version of what in a physical casino is called “replaying the round when the ball didn’t stop where it should have.”

BetCoin Dice took real losses. The investigation established that. GHash.IO didn’t formally acknowledge any public contrition, but there were reputational consequences: the pool stopped accepting deposits at its addresses, and the community started watching it more closely. This story is one of the reasons that, six months later, when GHash.IO once again grew toward the critical share, the pressure was so quick and so strong. People remembered.

This is an important correction to the picture usually painted in the Bitcoin community. People often say: “Bitcoin has never had a 51-percent attack.” That’s true only at a certain level of strictness. If you understand a 51-percent attack as “a complete rewriting of blockchain history months back, with double spending of all major transactions” — yes, that hasn’t happened. If you understand it as “use of a majority position to extract gain at the expense of other participants in the network” — it has. And it has been documented.

This is a critical detail for our subject. Because it shows: the mechanism of “converting winnings into the right to change the rules” is not a theoretical threat. It has already gone off. It just went off on a limited scale, because the pool, understanding that a full breach would kill the entire network (and its profit along with it), stopped at local exploitation.

Local exploitation is bad news in another sense. It means that a system in which the holder of a majority of computation can exploit the minority, even when not destroying the whole construction, is structurally unstable even when Axelrod’s three conditions are met. Because TfT equilibrium presupposes that the participants play more or less on equal footing. The moment a majority player appears who can carry out a “pinpoint attack” — take something from someone but not break the system as a whole — they will do it. Rationally. In keeping with their incentives. In full accordance with game theory.

And there’s no defense against this inside Bitcoin’s mechanics.

Under the Hood. Monero, August 2025

The second example is a very recent one. In August 2025 — that is, less than a year before these lines are being written — something happened that confirmed what I was just talking about, but on a far larger scale. And it was done not by one of the major exchanges and not by a hacker group, but by the project this book is, in principle, written about — Qubic.

To retell the whole story here is not the place — that’s the topic of chapter fifteen. But the brief fact is this. Monero is a network similar to Bitcoin, also Proof-of-Work, but with a different algorithm (RandomX, optimized for ordinary processors rather than for specialized ASICs). Monero’s market cap at that moment was about 6 billion dollars. Qubic’s market cap — about 300 million. A ratio of 1 to 20.

Over three months — from May to August 2025 — Qubic, through a mechanism we’ll talk about in detail in chapter 8, started paying Monero miners three times more than they were getting by mining Monero directly. No hack. No hacking. Just a different economics: “connect to our pool, we’ll route your computational work in such a way that you earn more.”

The miners looked at the numbers. Reconnected their equipment. On August 11, 2025, in a window of 122 blocks, Qubic owned 63 blocks out of 122 — 51.6 percent of the hashrate. The Monero network was under the control of a network twenty times smaller in market cap.

What happened next is a separate story. Qubic did not destroy Monero. But the technical possibility of destruction was there, and in a public statement on August 3, 2025 the Qubic developers spelled out exactly what they could have done if they’d wanted to: double-spending against exchange deposits, rewriting block history, censoring transactions, crashing the price to zero. They didn’t. They decided to stop. But they could have.

For our chapter — and specifically, for the question of why Bitcoin can’t fix itself — this is the key episode. Because it proves empirically, not theoretically: any PoW network can be attacked by offering miners a better economics. The Qubic developers, summing up the experiment, formulated this as one of three proven theses: “Incentives Dictate Consensus: Any Proof of Work blockchain can be attacked and controlled by providing superior economic incentives to its miners.”

This is a critical statement. It isn’t saying “Bitcoin is weak” — Bitcoin is strong, which is precisely why it’s more expensive to attack. It’s saying “no PoW network is structurally protected.” The protection is just money. More money — more protection. Someone with bigger money can buy that protection out. And money, let me remind you, is exactly what accumulates with those who have already won the game.

The circle is closed.

What This Means for Our Story

Let me bring it all together.

The first chapter ended on the point that Bitcoin assembled a revolution — but in doing so left “Murdoch’s gas” in the air, computational cycles that nobody needs. That was bad news about the use of a resource.

The second chapter gave good news: cooperation without an arbitrator is possible if Axelrod’s three conditions are met, and Bitcoin on the whole meets those conditions. The third chapter — that is, this one — says: the bad news is deeper than we thought in the first chapter. Bitcoin not only uses its resource inefficiently. It is also structurally incapable of fixing itself, because the money earned inside is constantly being converted into power outside, and that power comes back inside as the ability to change the rules. Concentration of mining. Concentration of ownership. Concentration of development. Three layers, the same dynamic.

You could say that all of this is just the general problem of capitalism, and Bitcoin isn’t unique in it. That would be true if Bitcoin hadn’t promised the opposite. And it did. It promised that every processor would be one vote. That mining is decentralized. That nobody controls the network. That the money it produces is neutral and resistant to any politics. Sixteen years on, all of that is either already untrue or becoming untrue right in front of us.

And not because Satoshi was wrong. He wasn’t wrong — he solved a specific problem, and he solved it. It’s just that his solution is the Darby phase. The first great breakthrough. It opens doors that were locked before. But it has a structural limit you can’t step over from inside. The Murdoch phase is already a different engineering problem, with different conditions.

And here we come to the main question of this chapter.

The Fourth Condition

Axelrod in 1980 described three conditions under which cooperation is possible: iteration, memory, transparency. He didn’t write about a fourth, because in his tournament that condition was satisfied automatically — the programs played in a closed system in which winnings could not be converted into anything external. In the real world this condition may or may not hold — and that is the defining difference between systems that stay cooperative and systems that drift toward oligopoly. Let me try to state this condition explicitly:

The system has to be set up so that winnings inside the game cannot be converted into the right to change the game itself.

Or shorter: the boundary between “the prize” and “the rules” must be impossible to cross from inside.

In Axelrod’s tournament that boundary was absolute — points and rules lived in different worlds. In Bitcoin that boundary is a permeable membrane. Internal money constantly seeps outward, acquires external influence, and comes back in already as pressure on the rules. Gradually. Year after year.

This fourth condition isn’t an improvement on Axelrod. It’s an addition. The three original conditions determine whether cooperation will be the equilibrium between players. The fourth determines whether that equilibrium structure will be preserved in time, or whether it gradually drifts toward a new equilibrium in which a few play and many submit. It’s about the durability of durability itself.

If we want to build a system that satisfies not only Axelrod’s three conditions but also the fourth, we need an architecture built differently. Not “an improved Bitcoin.” Not “Bitcoin with a patch against concentration.” A fundamentally different construction, in which the very mechanism of winning is built so as not to produce concentration. This is a non-trivial engineering problem. The naive solution — to forbid the sale of bitcoin to large investors — doesn’t work, because it destroys the first property (a free market). The naive solution — to make the developers receive a salary from the protocol — works partially, but runs into the question of who decides whom to pay. The naive solution — to forbid mining pools — doesn’t work, because it destroys the ability of small miners to participate in the network.

What’s needed is an architecture in which the answer to all of these questions is built into the very mechanism by which the network operates, rather than being decided by a vote or by an authoritative directive. Where the mining process itself produces decentralization rather than concentration. Where the ownership process itself prevents oligopoly among large holders. Where the development process itself depends on the entire community rather than on a narrow group of sponsors.

Such an architecture exists. It’s called Qubic. And how it works is the subject of this book from chapter six to the end.

But before we talk about how it works, we need to talk about the man who made it. Not because the project’s history is incomplete without the man’s history — but because this man saw Bitcoin’s structural defect not in 2025, when it became obvious to everyone, and not in 2020, when it began to show up in the data. He saw it in the early 2010s, when Bitcoin was still young and nobody doubted that it would remain decentralized forever. He saw it — and that’s why he was able to build two crypto networks that came out at the very top of the all-time ICO ranking: the first, in first place; the second, in second. More than a million percent ROI on one, four hundred thousand on the other. And from those two networks he took away an understanding of exactly how crypto projects die from inside.

His name is Come-from-Beyond, and he really did come from beyond.

P A r T i i The Man