Why Bitcoin, Not a New Coin

A purpose-built agent-coin can copy the features — but not the neutrality, security, trust, and liquidity that make a base money.

In brief. If the agent economy needs a money with four properties — no-KYC self-custody, censorship-resistance, sub-cent settlement, machine tempo — why not just build one? A purpose-built coin could satisfy all four on launch day; the constraints are the easy part, a copyable spec. What it can’t copy is what makes a money a base money: credible neutrality (no one who can change the rules), a proof-of-work security budget, Lindy trust from surviving seventeen adversarial years, deep liquidity, and an economy’s pull toward a single money. None of that is minted on demand — it’s earned, and Bitcoin alone has earned the full set. This is the long version of The Case’s “why not a new coin” beat.

The features are the easy part

Start by conceding the strongest form of the objection. The four constraints really are a specification, and specifications can be copied. A new chain can ship with no-KYC self-custody, censorship-resistance at the protocol level, sub-cent fees, and sub-second settlement — and it can do so on day one, purpose-built for agents, unburdened by Bitcoin’s conservatism or block-size debates. If satisfying the four constraints were the whole question, there would already be a thousand viable agent-coins, and the substrate question would be a matter of taste.

It is not the whole question, and the gap between “satisfies the four constraints” and “is the agent economy’s money” is the entire subject of this page. The constraints get a candidate in the door. What decides which candidate an economy actually adopts as its reserve and unit of account is a different and harder set of properties — ones that describe not what the money does but what it is, and what it is turns out to be a function of history that a launch cannot fast-forward.

Neutrality can’t be minted — only earned

A base money has to be credibly neutral: no issuer, no premine, no foundation sitting on a treasury, no governing body, no policy lever anyone can pull. Neutrality is, precisely, the absence of a controlling party — and an absence is demonstrated by history, not declared in a whitepaper.

Here is the problem every new coin runs into. To launch a coin, someone creates it: chooses the initial distribution, holds the founder’s allocation, sets the roadmap, controls the multisig or the upgrade keys. That someone is a party who can act — and a party who can act is a party who can be pressured, subpoenaed, bought, or captured. The moment a money has a steward who can change its rules, supply, or ledger, it has re-introduced the exact discretion surface the four constraints exist to eliminate. It does not matter how benevolent the founder is; the constraint is violated by the existence of the lever, not its use.

Bitcoin’s neutrality is credible for one reason: there is no one to call. Its creator left and never moved the coins; the supply schedule is fixed and has never changed; no party can alter the rules without the near-unanimous consent of a global, adversarial set of node operators who have every incentive to refuse. That property cannot be forked, because the act of forking it re-creates a founder — the new chain’s launcher is, by definition, a controlling party at genesis. You can copy Bitcoin’s code in an hour. You cannot copy the fourteen years of not being controlled by anyone that make its neutrality believable.

Proof of work and the security budget

A base money’s settlement has to be expensive to attack, because the whole point is that no one can rewrite who owns what. Bitcoin anchors its ledger to proof of work: rewriting history requires out-computing the largest hash rate and energy expenditure ever assembled, which costs more than any attack could yield. Security here is not a claim; it is a standing, sunk, ongoing cost that an attacker would have to overcome in real watts.

A new coin starts with a trivial security budget. Whatever its design, on day one it is cheap to 51%-attack, cheap to reorg, cheap to coerce — so its “settlement assurance” is nominal until it has accumulated a defensive cost base, which again takes years and enormous capital. Proof of work also gives Bitcoin a cost that can’t be faked: new supply can only be produced by spending real energy, which no ledger entry can fake. A freshly minted token cannot replicate that — least of all a proof-of-stake design, where the issuer can mint at will and the largest holders set the rules, reproducing precisely the “the rich and the issuer decide” surface a neutral money must not have. Security, like neutrality, is bought with accumulated cost over time, not written into a launch.

Lindy and aggregated trust

Trust in a money is cumulative and time-bound, and that is not a soft observation — it is the asset’s most important property and the one most obviously impossible to shortcut. Every year Bitcoin survives adds to a record that cannot be back-dated: every exchange collapse it outlived, every state that tried and failed to kill it, every 80% drawdown it recovered from, every protocol attack that didn’t land. This is the Lindy effect — for a non-perishable like a monetary protocol, the longer it has already lasted, the longer it is rationally expected to last. Seventeen years of adversarial survival is itself a property, and it is one no new coin can possess, by definition, on the day it launches.

There is a coordination layer on top of the time layer. Money is a Schelling point: people trust it partly because everyone else does, and that mutual expectation is itself the asset. The agent economy, asked to park a treasury for the long term, will rationally choose the money the most other actors already trust and the one that has already absorbed the most punishment — not the untested challenger asking it to go first. Trust aggregates onto the survivor. That aggregation has already happened, and it happened around Bitcoin.

Liquidity, and why economies converge on one money

This is the deepest reason, and it is monetary rather than technical — it would hold even if a new coin somehow matched Bitcoin on neutrality and security.

Picture a pure barter economy of n goods. To trade, you need a price for every pair — and there are n(n−1)/2 of them. Liquidity is shattered across thousands of thin markets, and there is no common yardstick of value, so pricing anything is a combinatorial mess. Economies escape this trap the same way every time: they converge on one money — a single good everyone accepts — so that each of the other goods needs only one price (denominated in that money) and one deep market (traded against that money). The pricing problem collapses from n² to n; liquidity pools instead of fragmenting.

Money is therefore a network-effect good: its usefulness rises with the number of participants who accept it, which creates a self-reinforcing, winner-take-most gravitation toward a single dominant money. A world of a thousand agent-coins does not improve on this — it re-creates the barter problem, with fragmented liquidity, no common unit of account, and constant conversion friction taxing every trade. The agent economy will converge on one money for the identical reason every human economy has: pricing efficiency and liquidity depth. And the money an economy converges on is the most neutral, most liquid, most trusted, hardest one available — the one with the strongest network effect already in hand. Among deployed candidates that is Bitcoin, not a coin minted last quarter with no one yet on the other side of the trade.

”But agents will tokenize their own services”

A fair and important objection, and the one most specific to this economy: agents may well issue tokens for their own services — an inference credit, a compute voucher, a data-feed token, a share of some agent’s future output. Doesn’t that mean a proliferation of agent-issued tokens rather than convergence on one money?

No — and seeing why sharpens the whole argument. A service token is a claim on a good, not money. Its value derives from the service behind it, and its issuer is that service; it is the thing being sold, sitting on the goods side of the trade. To price that token, to trade it, to settle it against everything else an agent might want, the agent still needs a neutral unit that everyone accepts on the other side of the pair. Tokenizing a service does not remove the need for money — it increases it, because now there must be a liquid market in which that token can be sold for the neutral money the agent actually wants to hold, price against, and spend elsewhere.

So service tokens proliferate at the edges exactly as the argument predicts goods do — while the settlement money trends toward one. A thousand tokenized services is not a thousand monies; it is a thousand goods, all of which need a common money to be priced and exchanged efficiently. The existence of agent-issued tokens is evidence for a single neutral base money, not against it. The tokens are the merchandise; Bitcoin is the till.

Most “agent-coins” are issuer-tokens in disguise

There is a more deflating observation to make about the agent-coins actually being proposed. Inspect them and the overwhelming majority are issuer-backed tokens — a company, a foundation, a treasury, an upgrade key, a discretionary supply. Which means they fail the very constraints the agent economy needs (no-issuer, censorship-resistance) on contact, and they collapse straight back into the analysis of the Independence Doctrine: an institution that retains a freeze-and-control surface cannot serve the parallel economy without ceasing to be itself. A token with a controlling party is not a new neutral money. It is a private stablecoin wearing a crypto costume — and the agent economy has the same reason to decline it that it has to decline every other issuer-controlled instrument.

The honest test — and what would prove it wrong

The argument is not mystical about the ticker, and it should be stated in a way that can lose. If a credibly-neutral, no-issuer, sufficiently-secured alternative base money emerged — and accumulated the liquidity, the Lindy record, and the network effect — and agents migrated to it, then “Bitcoin specifically” weakens to “a Bitcoin-like substrate.” That is a real, if demanding, outcome.

But notice what that test actually requires, because it is the whole point. A challenger would have to earn the entire bundle — neutrality, proof-of-work-grade security, accumulated trust, deep liquidity, and single-money network effect — from zero, while Bitcoin’s lead on every one of those compounds. The argument here is for the property bundle, not the name; Bitcoin is simply the only deployed money that holds the full bundle, and the bundle is the kind of thing that can only be earned over time, with a seventeen-year head start that grows rather than shrinks. A new coin can copy what Bitcoin does on launch day. It cannot copy what Bitcoin is, because what Bitcoin is, is the money that already spent the time.