PhronesisThe kernel
Grant the case first, in full, because it is documented and because the correction depends on it.
In 2008 Satoshi Nakamoto published a nine-page paper, "Bitcoin: A Peer-to-Peer Electronic Cash System," and the complaint on its first page still reads clean: "Commerce on the Internet has come to rely almost exclusively on financial institutions serving as trusted third parties to process electronic payments." The fix is stated just as plainly — "an electronic payment system based on cryptographic proof instead of trust, allowing any two willing parties to transact directly with each other without the need for a trusted third party." Every convenience the technology genuinely has follows from that one removal.
Permissionless: no account application, no approval, no hours — the ledger does not close on weekends and does not ask who you are. Borderless: a transfer to anywhere the network reaches settles without the correspondent-banking chain that makes an international wire a multi-day diplomatic event. Fast finality: settlement in minutes to an hour, against days for cross-border wires. Censorship-resistant: there is no administrator positioned to freeze or refuse the transaction. Programmable: money that can be addressed like an API.
And the use is not hypothetical. Some remittance corridors are underserved by banks — slow, expensive, or simply absent — and people move value on permissionless rails there because the alternatives are worse. That deserves sober acknowledgment, not promotion and not a sneer. The kernel is true.
The electric part
What replaced the trusted third party is electricity. Proof-of-work secures the ledger by having miners spend energy on trillions of guesses per second; the chain carrying the most spent work wins, and rewriting history would mean re-spending the work. The security is not like paid-for energy — it is paid-for energy, denominated in kilowatt-hours.
The scale is measured. The Cambridge Bitcoin Electricity Consumption Index puts Bitcoin's annualized draw at roughly 148 TWh as of mid-2026, with theoretical bounds of about 80 to 231 TWh — the estimate moves with hashrate and with the electricity-price assumption underneath it. By Cambridge's own comparison set, that is roughly the annual electricity consumption of Malaysia, slightly less than Poland's, more than Norway's or Sweden's.
In this site's terms, and as a proposal: a proof-of-work ledger is order rented from a flow — the eddy with a market attached. Stop paying the energy and the security dissolves. "Electric" is not a metaphor here; it is the balance sheet.
The correction
Here is the spine of the note. Nakamoto did not stumble into irreversibility; it was the point. Section 1 of the paper: "Completely non-reversible transactions are not really possible, since financial institutions cannot avoid mediating disputes." Mediation costs money, the costs price out small transactions, and "with the possibility of reversal, the need for trust spreads." His remedy: "Transactions that are computationally impractical to reverse." Note the engineer's precision — impractical, not impossible — though at network scale the distinction rescues no one. Finality is the product. Everything in the kernel above is a description of finality wearing its good clothes.
Now the same property, wearing its other clothes.
Error is unrecoverable. Chainalysis estimated in 2017 that between 2.78 and 3.79 million bitcoin — roughly 17 to 23 percent of the bitcoin then in existence — were lost forever; a 2020 follow-up put it near 20 percent. Lost keys, dead drives, mistyped addresses. On a reversible rail these are phone calls. On this rail they are geology.
Theft is unrecoverable by construction. The FBI's Internet Crime Complaint Center logged $11.37 billion in cryptocurrency-linked losses across 181,565 complaints in 2025 — up 22 percent from 2024, an average loss of $62,604, and more than half of all reported internet-crime losses that year ($20.88 billion total). The FBI's own explanation of why criminals prefer the rail includes "the speed of irreversible transactions"; it adds that using cryptocurrency "makes it harder for victims to recover stolen funds." The scammer and the remittance sender are using the same feature.
The contrast is legal, not just technical. A U.S. card payment carries statutory dispute and reversal rights — Regulation Z for credit, Regulation E for debit — with error-resolution processes and caps on unauthorized-use liability written into law. The FTC states the other side flatly: "Cryptocurrency payments typically are not reversible" and "do not come with legal protections." One rail has an undo mechanism required by statute; the other was designed so that no such mechanism can exist.
Self-custody concentrates the failure. The pitch is "be your own bank." The full accounting: you are also your own fraud department, your own insurance fund, your own recovery process, and your own night shift. Every protection a bank distributes across institutions and law lands on one person's operational security, forever, with no failover. One weak hour, one final mistake.
The convenience and the danger are the same property.
That is the whole correction. There is no version of the fast, permissionless, unstoppable transfer that is also the recoverable one. You cannot keep the hinges and lose the door.
Where it lands
This site keeps a frame for exactly this class of thing: some doors are one-way, and the question that governs conduct near them is whether a mistake is recoverable. Irreversible-by-design money answers that question in advance, for every mistake, uniformly: no. It is the rare domain where all errors — typo, theft, coercion, a persuasive stranger, a moment of certainty — land in the unrecoverable class.
For an institution, the trade can be priced. Custody arrangements, multi-signature process, insurance, legal staff — finality becomes a line item, wrapped in enough procedure that no single tired human is ever the last check. That is a defensible engineering posture, and it is how the serious operators treat the property.
For a person there is no such wrapper. The judgment call would be exercised at exactly the wrong times — tired, rushed, frightened by a scammer's deadline, or worst of all, sure. The site has argued that intuition feels most certain precisely in the domains that give it least to learn from, and a 3am transfer decision is a low-validity domain with a countdown attached. The convenience is a one-way door with excellent hinges: it opens most smoothly in the states least fit to walk through it.
So the proposal: the rational personal policy is a wall, not a judgment call — a standing rule, made once, in a fit state, that removes the per-case decision entirely, because the per-case decision is the failure mode. This is not a claim that the technology is worthless. It is the claim that a domain where every error is final should be governed by policy, not by in-the-moment judgment.
The plain fact, stated as the conclusion it is: this site takes no crypto payments. The wall is the output of the analysis, not a prejudice against the ledger.
Limits
Three limits bound the argument's reach; none touch its spine.
The electric part is contingent. Ethereum's switch from proof-of-work to proof-of-stake — the Merge, September 15, 2022 — cut its energy use by roughly 99.95 percent on the Ethereum Foundation's estimate; CCRI later measured the electricity reduction at 99.988 percent. The energy objection is an objection to proof-of-work, not to ledgers as such. The finality objection survives the Merge untouched.
Custody restores reversibility — by undoing the premise. A custodial service can freeze accounts, reverse internal transfers, and answer subpoenas, which is to say it is a financial institution serving as a trusted third party: the exact thing the 2008 paper was written to remove. The industry's fix for finality is Nakamoto's own trade, run backward. What remains is a slower bank with thinner statutory protection.
"Convenient" varies by corridor. Where banking rails are genuinely bad, the finality risk is a trade some people take with open eyes, and the analysis should price that trade, not mock it. The wall proposed here is a personal policy for someone with working alternatives, not a decree for someone without them.
None of these limits reverse the correction. At the protocol layer, finality is the product — and a product whose every failure is permanent belongs, for a person, behind a wall built in advance.
A companion to Money, the Sport with No Referee — this note is in the same genre: a warning, not a compliment, and not an on-ramp.
Kin to Money, the Sport with No Referee, the one-way door, order rented from a flow, and the gut's domain.
Rests on: Satoshi Nakamoto, "Bitcoin: A Peer-to-Peer Electronic Cash System" (2008), for the design intent and the non-reversibility passages, quoted verbatim; the Cambridge Bitcoin Electricity Consumption Index (Cambridge Centre for Alternative Finance), ~148 TWh/yr annualized as of mid-2026, bounds ~80–231 TWh, with Cambridge's own country comparisons; Chainalysis lost-coin estimates (2017 study, 2.78–3.79M BTC, 17–23% of bitcoin then in existence; 2020 follow-up near 20%); FBI IC3 2025 Annual Internet Crime Report ($11.37B in cryptocurrency-linked losses, 181,565 complaints); FTC consumer guidance on cryptocurrency and scams, with Regulation Z (12 CFR § 1026.13) and Regulation E (12 CFR §§ 1005.6, 1005.11) for the statutory reversal rights on cards; ethereum.org and CCRI for the Merge's energy reduction (September 15, 2022; ~99.95% estimated, 99.988% measured). Those results are established and credited. Reading finality through the one-way-door frame — and concluding that the rational personal policy is a wall rather than a judgment call — is the proposal, offered to be argued with, not a proven result.