Nobody Was Watching. That Is Why It Worked.

A hammer is useful to the only person who owns one. So is a coat, a field, a book. Almost everything humans have ever valued keeps its value when the rest of the world looks away.

Money does not. Money is the one possession whose entire worth is other people's willingness to take it. Carl Menger put this in writing in 1892, in a paper that still reads better than most things published since: a good becomes money because it is more saleable than the alternatives, and saleability is not a quality of the object at all. It lives in everyone else.

Which means that money is not a thing with a network effect. Money is a network effect, wearing the costume of a thing.

The Vulnerability That Came First

Everyone describes Bitcoin's network effect as its defensive moat. That is the second half of the story and it skips the interesting half.

On 3 January 2009 the network had one participant. Nine days later it had two, when ten bitcoin moved from Satoshi Nakamoto to Hal Finney in block 170. For most of the first year, the value of holding it was indistinguishable from zero, because there was almost nobody to be saleable to.

Rohlfs called this the start-up problem in 1974. Below a certain size, a network has no stable state except non-existence. What he did not have to consider is the version of the problem that applies to a system with no owner: a decentralised network is at its least defended exactly when it is at its smallest, and it cannot call anyone for help, because having someone to call is the thing it was built to avoid.

This is the trap. If a network is to be decentralised, it has to be decentralised from the beginning. At the beginning it has no network effect. With no network effect it is cheap to attack, and there is no authority to intervene when someone does. Every project that launched after Bitcoin faced that trap knowing it existed, and solved it the only way it can be solved: by keeping a central point of control during the fragile years, whether that is a foundation, a validator set, a multisig, or a team that can halt the chain.

Bitcoin got through those years for an unglamorous reason. Nobody was watching. It grew up in an empty room. By the time the room filled, attacking it had stopped being cheap.

That is not a compliment to Satoshi's foresight. It is the point Brian Arthur made in 1989 and it deserves to be conceded plainly: where returns increase with adoption, the winner is decided by early accidents rather than by merit. Bitcoin's founding advantage was partly luck. It is also, for exactly that reason, not reproducible. You cannot arrange to be ignored on purpose in a world that is now watching.

Where We Actually Are

It is worth being unsentimental about the scale of this.

Roughly 74 percent of humanity is online. WhatsApp reaches about half of those people. Bitcoin, by River's estimate in early 2025, is held by under 4 percent of the world population, with the United States around 14 percent and Africa at 1.6 percent.

The network is seventeen years and eight months old. It has 56,789,267 addresses holding a balance, 25,689 reachable public nodes and 20,078,253 coins in existence. Measured against the internet or the smartphone at the same age, that is early. Measured against the confidence with which Bitcoin is discussed, it is very early.

I want to be careful here, because this is where the argument usually goes wrong. It is tempting to observe that the internet took about twenty five years from ARPANET to mainstream households, note that Bitcoin is seventeen, and conclude something about the next eight. That is not a forecast, it is a coincidence of arithmetic dressed as one. Adoption curves have a characteristic shape. They do not have a characteristic schedule.

Why the Store of Value Case Is the One That Matters

Bitcoin has been described as a payment system, a settlement layer, a hedge, a protest and a technology platform. The claim that has actually been winning is the smallest one, and the one we set out in Bitcoin Isn't the Money You Were Promised: it is a place to put savings that nobody can inflate.

The numbers say the competition is not other cryptocurrencies. In the group of assets people hold specifically as a store of value, Bitcoin's share sits between 97.2 and 99.4 percent depending on which coins you count. That comparison excludes Ethereum and Solana on the grounds that they are platforms rather than savings vehicles, and I should say plainly that this is a distinction which happens to flatter the argument. Count the whole market instead, Ethereum's 295 billion dollars included, and Bitcoin holds roughly 58 percent. Both numbers are real. Neither is the interesting one.

The interesting comparison is gold, and there the picture reverses: gold's above-ground stock is 222,600 tonnes worth about 29.0 trillion dollars, against Bitcoin's 1.55 trillion. Bitcoin is 5.36 percent of gold.

Why would that gap close rather than persist? Not because of anything Bitcoin does. Because of what the alternative does.

The United States money supply has multiplied by 7.33 since January 1990, growing at 5.61 percent a year. Federal debt has multiplied by 12.80 over the same period, reaching 39.07 trillion dollars, which works out at 7.34 percent a year. The debt is growing faster than the money that would service it. Neither of those numbers is a scandal in any single year. Compounded across a working life, they are the whole argument, and we have made it at length in Impatience Is a Monetary Policy.

States understand this perfectly well. They are already moving, just into the older asset: central banks bought 863.3 tonnes of gold in 2025, and while that is below the extraordinary years above 1,000 tonnes, the World Gold Council notes it "remained significantly above the 2010-2021 annual average (473t)". That is 1.83 times the pre-2022 norm. The institutions with the best view of the currency system are diversifying out of it.

The Sentence a Government Has Already Written Down

On 6 March 2025, the United States established the Strategic Bitcoin Reserve by executive order. Two sentences in that document are worth reading slowly, because a G7 government wrote them about an asset with no issuer.

"Bitcoin is the original cryptocurrency. The Bitcoin protocol permanently caps the total supply of bitcoin (BTC) at 21 million coins, and has never been hacked."

And:

"Government BTC deposited into the Strategic Bitcoin Reserve shall not be sold and shall be maintained as reserve assets of the United States."

Note what is not being said. The order does not authorise purchases. The reserve is capitalised with coins forfeited in criminal and civil proceedings, which means the United States became one of the largest holders on Earth by confiscating rather than buying. The companion Digital Asset Stockpile, which covers every other seized digital asset, explicitly allows the Treasury to sell. Only bitcoin is ring-fenced.

Twelve governments now hold bitcoin on the public trackers, together 650,008 coins, worth around 50.3 billion dollars. The United States holds 328,372 of them.

A state that writes "shall not be sold" into law has not endorsed an investment. It has classified something as a reserve. Those are different acts, and the second one is heavier.

Who Actually Holds This Network

Now the part that changed how I think about the whole subject.

You can add up who owns Bitcoin, and the exercise is instructive mostly for how badly it fails.

Holder	Bitcoin	Share of coins in existence
ETFs and exchanges	1,550,270	7.72 percent
Listed companies	1,270,028	6.33 percent
Governments	650,008	3.24 percent
DeFi and other	376,345	1.87 percent
Private companies	284,650	1.42 percent
Satoshi, untouched since 2009 (estimate)	about 1,100,000	5.48 percent
Everyone else, plus everything lost	about 14,846,952	73.95 percent

Every row above the last one has caveats, and they should be stated rather than buried. "ETFs and exchanges" mixes fund holdings with customer balances held in custody, so a large part of that 7.72 percent economically belongs to individuals. The Satoshi figure is inferred from a mining pattern, not observed. And the final row cannot be split into "people who hold their own keys" and "coins whose keys are gone", because on-chain there is no difference between a patient owner and a lost password.

That last limitation is usually treated as a data problem. I think it is the finding.

Roughly three quarters of this asset sits with holders who cannot be enumerated, subpoenaed as a group, or negotiated with. No other reserve asset works like that. Gold has vaults with addresses and auditors who visit them. Government bonds have registries with names attached. Even the shares of a widely held company have a transfer agent.

Bitcoin's largest holder class is a statistical residual. Nobody was watching when this network was small enough to kill, and nobody can watch who holds it now that it is not. Those are the same property at two different ages.

A network with a governing centre can be switched off by capturing the centre. A network with a registered ownership base can be pressured through the register. Bitcoin has neither, and the table above is the clearest evidence of it. This is not a claim that individuals are trapped, because any individual can sell tomorrow morning. It is a claim that there is nobody to negotiate with, because there is no body.

The Three Objections That Deserve Answers

Network effects only delay, they do not bind. This is the strongest objection and the evidence for it is in our own research. The American landline network reached 95 percent of households in 2002 and 41.7 percent by 2018. Earlier, it lost a quarter of its households in the Great Depression, falling from 41 percent in 1929 to 31 percent in 1934. Symbian led the smartphone market in 2010 with 36.6 percent and does not appear in the 2026 statistics at all. Every one of those was a dominant network. Every one of them lost.

The answer is not that Bitcoin is different by nature. It is that the switching cost is different in kind. Moving from a landline to a mobile cost a consumer nothing except the effort, because the two networks carried the same thing: a conversation. Moving from one money to another means persuading everyone you transact with to move too, and money is the only network where the thing being carried is the agreement. That makes displacement slower and much less likely, not impossible.

States can walk away too. They can, and one has. El Salvador made bitcoin legal tender in 2021 and then amended the law in early 2025 under an arrangement with the International Monetary Fund, dropping the obligation on merchants to accept it. The first state to adopt bitcoin as legal tender partially reversed the step when it needed credit. Anyone arguing that states will be pulled toward Bitcoin has to hold that case in view, because it shows the pull running the other way when the incentives change.

The first-mover advantage was luck. Yes. Arthur established that in 1989, and I said so above rather than waiting to be caught. It does not weaken the conclusion, though. An advantage does not have to be deserved to be real, and an accident that cannot be repeated is still an accident that cannot be repeated.

What Would Show Me I Am Wrong

Three things, all observable.

A network launched after Bitcoin surviving to meaningful size without a central point of control during its early years would break the core claim of this essay, because it would prove the trap can be escaped by design rather than obscurity.

Bitcoin's share of the store-of-value group falling structurally over several complete cycles, rather than oscillating, would show the concentration is a phase rather than an equilibrium.

And addresses holding a balance declining across a full cycle would mean the participation curve has turned. That single series is the one I would watch. It has fallen only once, by 16.6 percent between January 2018 and January 2019, and it recovered within a year. A drop that did not recover would be a different signal entirely.

None of that says anything about price, and this essay deliberately does not. What it says is narrower. Money's value has always been other people, Bitcoin arrived at that position through an opening that has since closed, and the largest part of what it has become is held by people no institution can name. Whether that ends up mattering as much as I think it does is not something the data can settle yet.

For the mechanics behind all of this, start with what the network effect is and why Bitcoin has one. For the evidence and the argument about how it is measured, see our analysis of where the network effect comes from. And for the deeper reason Bitcoin can hold an agreement without an authority to enforce it, The Clock No One Owns goes further than this essay does.