An engraved plate on slate blue: three dark woolly spheres are lashed into an orange web of threads around a spiked wheel, a face ringed by a black corona sits wrapped in swirling filament coils, concentric ring-bursts fill the corners, and faint fish, an insect and a jellyfish drift at the edges.
15 July 2026 · 10 min read

The Ban Question

A ban is a law, and a law reaches what's inside a government's borders: the bridges, the custodians, the middlemen, and you, if you live there. Bitcoin's design puts the rules and the supply outside that reach, and the keys beyond anything a state can do to millions of holders at once. What a ban can and can't touch, and what the largest state attack to date, China's 2021 mining ban, actually changed.

Any government can pass a law against bitcoin. The question is what that law reaches, what it can’t, and what happened when China banned mining.

Everything else depends on the answer, because the thesis needs bitcoin to be neutral money, money whose rules no one can change alone. If a government could turn the network off or rewrite its rules, bitcoin would be a licensed product waiting for its licence to be revoked.

A ban is a law

A ban is a law, and a law binds the people, companies, and property inside a government’s borders, but a state whose currency other countries must use to trade reaches much further than its own borders through the banks and the payment system.1 Bitcoin’s design puts what matters most outside both. The rules, the supply, and the keys, which a state can only reach one holder at a time. What’s left inside the reach is the bridges, the custody, the domestic industry, and you, if you live there. So a government can make bitcoin slower, more expensive, and legally risky to use. It can’t turn the network off, change its rules, or print more of it. And China ran the biggest live test of this in 2021.2

What a ban can reach

The bridges are licensed companies. Exchanges [the businesses that convert pounds into bitcoin and back] and the banks that serve them have offices and directors, so a government can fine them, close them, or cut them off from the banking system. A state whose currency other countries must use to trade reaches exchanges abroad too, through the banks those exchanges need. Every crossing gets slower and costlier.

Custody is reachable for the same reason. When one regulated company holds the coins behind funds that millions of people own shares in, that pile sits at one legal address, with a board and a regulator. An ETF [a stock-market fund that has a custodian hold bitcoin on your behalf] is the clearest case. The pile can be pressured, frozen, or handed over. That’s close to how gold was reached. The order of 5 April 1933 required holders to deliver gold coin, gold bullion, and gold certificates to a bank, and the collection ran through the banks rather than through searches of homes.3 The Gold Reserve Act of 30 January 1934 then made the Federal Reserve’s gold the property of the United States, and that step was easy because the gold had already gathered in a few vaults.4 Between them, the two steps reached everyone, individuals included. What the pooling did was make the order cheap to enforce. Custody was one of gold’s two weak points. The claim on the metal was the other, because a promise to pay can be refused by whoever made it.

Domestic companies and jobs are inside the border. Miners, wallet developers, shops that accept it. A government can shut them down or drive them abroad.

A ban reaches you, if you live there. A state can criminalise spending, watch the bridges, tax punitively, prosecute publicly to spread fear. A ban doesn’t have to kill the network to make your life inside that border worse. If a state turns openly hostile, the advice is to cut your exposure to that state, up to and including leaving it. Leaving isn’t possible for most people, but reducing what that state can reach is possible for almost everyone.

The story is reachable as well. A government can push the line that holding your own keys is dangerous, and steer people toward supervised wrappers or a central bank digital currency [state-issued digital money that can be programmed, monitored, and frozen account by account].

What a ban can’t reach

The rules are outside every border. There’s no company, no chief executive, no headquarters, no switch. But a state can reach named people, the handful who hold the right to merge changes into the main software and the few large pools that direct much of the mining.5 Reaching them doesn’t change what counts as a valid block. Everyone running a node [software anyone can run that checks every transaction and block against the rules] enforces the rules. In one place the software takes the developers’ word by default.6 It skips checking the owners’ signatures on the transactions in blocks older than a point the developers mark when they ship it, and you can turn that off.7 A law that binds one country’s node operators changes nothing for the rest, and nodes are globally distributed, many of them unnoticed.

So is the supply. Twenty one million coins, issued on a fixed schedule.8 That counts coins, not claims to coins. Anyone holding coins for other people can write more claims than they hold, and nothing in the rules stops them. No parliament can vote the supply higher, because supply is a rule, and the nodes enforce the rules.

A law can’t round up keys people hold themselves. Self-custody [keeping the secret keys to your own coins rather than leaving them with a company] reduces the asset to information. Twelve memorised words, the seed phrase [a list of words that can recreate your wallet], walk through any airport.9 There’s no vault to raid, because the vault is tens of millions of separate heads and drawers spread across jurisdictions. A state would have to knock on every door. Mass seizure becomes far costlier, not impossible.

And no single country’s law stops the network running. As long as one jurisdiction anywhere lets miners and nodes run, blocks keep arriving for anyone who can reach the network, including the citizens of the country that banned it, near ten minutes apart, and when a shock slows them the difficulty adjustment pulls them back.10

That leaves an asymmetry. A ban has to work everywhere, forever, to beat the network. The network only has to keep working somewhere.

China, the live experiment

By spring 2021 roughly half the world’s bitcoin mining ran in China, and not long before it had been closer to two thirds.11 China’s mining was built on coal in the north and seasonal hydro in Sichuan. In May 2021 the State Council ordered a crackdown, and through May and June the provinces shut the farms.12 Around half the network’s total computing power went dark within weeks.13 This was the largest state attack on bitcoin to date. The world’s second-largest economy switched off its own dominant share of the network’s security.

At the network level, blocks slowed, because fewer machines were working on puzzles set at the old difficulty. Then the difficulty adjustment [an automatic rule that retunes how hard mining is, roughly every two weeks, so blocks keep arriving near ten minutes apart] pulled the difficulty down in steps.14 The largest single step, about 28%, was the biggest downward reset in the network’s history, and the steps together came close to matching the fall in machines.15 Blocks returned to schedule. No transaction was reversed. No balance changed. Not one extra coin was issued and not one was destroyed. Every coin stayed exactly where its keys said it was, including inside China. On the ledger, the whole event shows up as slower blocks for about nine weeks and the difficulty steps that fixed them.16 Nothing else.

At the market level, the ban paid the survivors. With half the machines gone, every remaining miner earned roughly twice as many coins per machine, because the same rewards went to fewer of them. That profit pulled the crated Chinese hardware into new homes in Texas, Kazakhstan, Canada, and elsewhere.17 Within months the network’s computing power was climbing steeply, as the relocated machines restarted and new ones came online, and inside a year it had passed its old peak and kept going.18 The United States ended up the largest mining country.19 And China came back.20 The best published estimate puts roughly a fifth of the network’s computing power inside China within a year, and that estimate is contested because miners can hide where they are.21 The ban didn’t fully hold at home.22

So what did China get? The supply rules didn’t change and China won no say over the network. It handed its rivals most of the mining, with the sites, the newly issued coins, and the fees that go with it.23 China kept the business of designing and selling the machines, and sold those machines on to the countries that took the work.24 A national ban works on the network the way a road closure works on a city. The traffic reroutes, and the closed road loses the traffic.

What a state does instead

A ban fails against the network and gifts the industry to whoever defects, so the incentives run against prohibition. A ban that worked would need every jurisdiction on earth to join and hold the line forever, when each of them profits by breaking it. And a ban advertises what the banned tool can do, and often speeds adoption outside the banning border. Prohibiting a savings technology tells citizens it has power their own currency lacks. Adoption breaks through first where people have the least control over their own money. A hostile state doesn’t lead the breakthrough. It pushes use away from the regulated middlemen and into trades between people, which is a second and separate force. Weak money sends people looking for an exit. The ban points at one.

So the realistic attack is the squeeze. Lean on the bridges, concentrate coins into regulated custody, frame self-custody as dangerous, keep bitcoin as an “investment” inside the old rails while people spend state money. What the squeeze takes from the bridges and the custodians first is records and control of the rails, not the coins. The ban is loud and fails. The wrapper is quiet and might not. Which is why self-custody and actual use are what the defence rests on.

What the asymmetry doesn’t settle

“1933 proves states win.”

Gold lost its monetary job because it pooled into vaults and the rules on the claims were then changed. Bitcoin was designed to remove the need for both. But the point cuts both ways, and I’ll be straight. The protection is conditional. Every coin that migrates into pooled custody rebuilds the 1933 target. A ban can’t round up keys. It can absolutely reach an ETF. So whether governments “can ban bitcoin” depends partly on how people choose to hold it.

“The network surviving isn’t the same as you being fine.”

Correct. A hostile state can’t kill the ledger, but it can hurt its own citizens with friction, fear, and prosecution. Venezuelans and Nigerians used bitcoin under hostility because the alternative was worse, and it was never painless.

“Never is a long time.”

Also fair. The claim is about cost, not impossibility. Each escalation buys less and costs more, in savings and talent leaving, an industry gifted to rivals. Each attempt to ban it so far has ended up showing what the attacker couldn’t do, and each survival left the network stronger.

A ban reaches the middlemen, the custody, and you if you live there, never the rules or the supply, and the keys only door by door. China tested what a ban can’t reach at the largest scale any single state could, and the result is written on the chain. What a ban can reach is still being tested, and how people choose to hold their coins is part of that answer.

Sources 24

What this essay rests on, as checked on 12 November 2026.

  1. 1 Cambridge University Press, An Introduction to Public International Law, Chapter 6: Jurisdiction, pages 101–119: "full jurisdiction over events and persons within its territory"
  2. 2 Cambridge Judge Business School, Geographic shift: "The immediate effect of the government-mandated ban on crypto mining in China was a 38% drop in global network hashrate in June 2021" Cambridge Centre for Alternative Finance, Cambridge Blockchain Network Sustainability Index: Mining Map: Methodology: "Following the government ban in June 2021, reported hashrate for Mainland China effectively plummeted to zero during the months of July and August but suddenly surged back to more than 20% in September." Cambridge Judge Business School, New data reveals timeline of China's bitcoin mining exodus, 15 July 2021: "This has effectively led to all of China's hashrate disappearing overnight." China's share of global hashrate: 75.5% in September 2019, 46% in April 2021, then effectively 0%. The page makes no comparison with any other state action, so it does not show this was the biggest.
  3. 3 The American Presidency Project, Executive Order 6102, Forbidding the Hoarding of Gold Coin, Gold Bullion and Gold Certificates, 5 April 1933: "All persons are hereby required to deliver on or before May 1, 1933, to a Federal Reserve Bank ... or to any member bank of the Federal Reserve System all gold coin, gold bullion and gold certificates". Dated April 5, 1933.
  4. 4 Federal Reserve History, Gold Reserve Act of 1934: "Section 2 of the act transferred ownership of all monetary gold in the United States to the US Treasury. Monetary gold included all coins and bullion held by individuals and institutions, including the Federal Reserve." The Act was signed January 30, 1934. The source prints a transfer of all monetary gold, coins and bullion held by individuals and institutions as well as by the Federal Reserve, not of the Federal Reserve's gold alone.
  5. 5 Bitcoin Core, Contributing guidelines, CONTRIBUTING.md, bitcoin/bitcoin repository: "Whether a pull request is merged into Bitcoin Core rests with the project merge maintainers." The page gives no number of maintainers, names no one, and says nothing about mining pools.
  6. 6 Bitcoin Core, Release announcement, Bitcoin Core 0.14.0, 8 March 2017: "Bitcoin Core 0.14.0 ships with a default assumed valid block that is set during the release process by having multiple well-known developers each confirm that a certain block is known to be valid."
  7. 7 Bitcoin Core, Release announcement, Bitcoin Core 0.14.0, 8 March 2017: "the software will skip verifying signatures in the blocks before the assumed valid block." "The default assumed valid block in Bitcoin Core 0.14.0 is #453354, 16 Februrary 2017"
  8. 8 Bitcoin.org, Frequently Asked Questions: "Bitcoin is unique in that only 21 million bitcoins will ever be created." "it is not possible to generate uncontrolled amounts of bitcoins out of thin air" Bitcoin Core, Bitcoin Core source code, src/validation.cpp, master branch: "CAmount nSubsidy = 50 * COIN;" and "Subsidy is cut in half every 210,000 blocks which will occur approximately every 4 years."; "if (halvings >= 64) return 0;" Bitcoin Core, Bitcoin Core source code, src/test/validation_tests.cpp, subsidy_limit_test: The test asserts the summed issuance as CAmount{2099999997690000} satoshis, that is 20,999,999.9769 BTC. The code's total is a little under the twenty one million coins the sentence states.
  9. 9 Bitcoin Improvement Proposals, BIP 39, “Mnemonic code for generating deterministic keys”, assigned 10 September 2013: "The allowed size of ENT is 128-256 bits." The table row "| 128 | 4 | 132 | 12 |" gives 12 words for 128 bits of entropy. Twelve is one of five permitted lengths: 12, 15, 18, 21, 24.
  10. 10 Bitcoin Core project, Source file src/kernel/chainparams.cpp, bitcoin/bitcoin repository, master branch: "consensus.nPowTargetTimespan = 14 * 24 * 60 * 60; // two weeks"; the file also sets nPowTargetSpacing to 10 * 60 seconds. That is the protocol's design target, not an observed average arrival time for blocks.
  11. 11 Cambridge Centre for Alternative Finance, Cambridge Judge Business School, New data reveals timeline of China's Bitcoin mining exodus: "China's share of total Bitcoin mining power has declined from 75.5% in September 2019 to 46% in April 2021." The earlier share printed here is 75.5%, not nearer two thirds.
  12. 12 Government of China (gov.cn), Report on the 51st meeting of the State Council's Financial Stability and Development Committee, chaired by Liu He, 21 May 2021: "打击比特币挖矿和交易行为" (crack down on bitcoin mining and trading), published 21 May 2021. The body is the State Council's Financial Stability and Development Committee, not the full State Council, and the page prints nothing about provincial shutdowns.
  13. 13 Blockchain.com, Charts — Hash Rate: Peak about 185,429,301 TH/s on 4 May 2021; trough about 87,820,755 TH/s on 7 July 2021, a fall of about 53%; that peak-to-trough window is about nine weeks.
  14. 14 Blockchain.com, Network Difficulty chart: “The difficulty is adjusted every 2016 blocks (every 2 weeks approximately) so that the average time between each block remains 10 minutes.” Difficulty 2.5046e13 (14–29 May 2021) down to 1.3673e13 (18–30 July). Only the difficulty series is printed. mempool.space, Bitcoin difficulty adjustments data feed, retarget records 1 May to 25 August 2021: Retarget entries: 30 May 2021 (block 685440) ratio 0.840347, a 15.97% cut; 3 July 2021 (block 689472) ratio 0.720573, a 27.94% cut. The feed prints retarget records, not block timestamps or intervals.
  15. 15 mempool.space, Full history of bitcoin difficulty adjustments, data feed: “[1625294046,689472,14363025673659.96,0.720573]” — multiplier 0.720573 at block 689472 on 3 July 2021, the largest downward step in the record; equivalent to about a 27.94% fall, though the page prints the ratio, not a percentage.
  16. 16 mempool.space, Bitcoin difficulty adjustments data feed, retarget records 1 May to 25 August 2021: Retarget entries from 1 May to 25 August 2021: 30 May ratio 0.840347 and 3 July ratio 0.720573 (both cuts), then 25 August ratio 1.13236, an increase. The feed prints retarget dates and ratios, no count of weeks.
  17. 17 Cambridge Judge Business School, “Bitcoin mining – an (un)surprising resurgence?”: “the trend quickly reversed as miners began to relocate operations abroad”, after the hashrate bottomed at 57.47 EH/s on 27 June 2021. Cambridge Judge Business School, “Bitcoin mining electricity update: new data”, published 17 May 2022: Global hashrate shares for September 2021 to January 2022: “Kazakhstan (13.22%), Canada (6.48%), and Russia (4.66%)”, with the US at 37.84% and China at 21.11%.
  18. 18 Cambridge Judge Business School, “Bitcoin mining – an (un)surprising resurgence?”: “by the end of the year, total network hashrate had almost fully recovered to pre-ban levels (193.64 EH/s on 21 December 2021)”, up from a 57.47 EH/s trough. It says almost fully recovered, not past the old peak. Blockchain.com, Total Hash Rate (TH/s) chart, seven-day average: Seven-day average: about 180.7 EH/s on 14 May 2021, about 84.8 EH/s on 3 July 2021, about 179.1 EH/s on 28 December 2021 — back near the May level by late December, not above it.
  19. 19 Cambridge Judge Business School, “Bitcoin mining electricity update: new data”, published 17 May 2022: “the US extended its leading position at 37.84% of global hashrate”, for mining map data “spanning the period from September 2021 to January 2022”.
  20. 20 Cambridge Judge Business School, “Bitcoin mining – an (un)surprising resurgence?”, mining map data September 2021 to January 2022: “reported hashrate suddenly surged back to 30.47 EH/s in September 2021, instantly catapulting China to second place globally in terms of installed mining capacity (22.29% of total market)”; “China has re-emerged as a major mining hub (21.11%).”
  21. 21 Cambridge Judge Business School, “Bitcoin mining electricity update: new data”: “In China, following a sudden uptick in covert mining operations after the June 2021 government-mandated ban on Bitcoin mining, the country has re-emerged as a major mining hub (21.11% of global hashrate)”, data to January 2022.
  22. 22 Cambridge Centre for Alternative Finance, Cambridge Blockchain Network Sustainability Index, Mining Map: Methodology: “Following the government ban in June 2021, reported hashrate for Mainland China effectively plummeted to zero during the months of July and August but suddenly surged back to more than 20% in September.”
  23. 23 Cambridge Judge Business School, “Bitcoin mining electricity update: new data”, published 17 May 2022: Global hashrate shares: USA 37.84%, China 21.11%, Kazakhstan 13.22%, Canada 6.48%, Russia 4.66%. The shares are for the data window September 2021 to January 2022 as a whole, not a January 2022 reading. Bitcoin Project, Developer guide, “Block Chain”: The coinbase transaction “should collect and spend the block reward comprised of a block subsidy and any transaction fees paid by transactions included in this block”. The page does not say the output names who was paid.
  24. 24 Canaan Inc., Annual report on Form 20-F for the financial year ended 31 December 2019: “We primarily sell our Bitcoin mining machines to customers in the PRC.” “As a fabless IC design company, we do not own any IC fabrication facilities.” PRC customers 74.8% of revenue in 2019. The filing predates the ban. Marathon Digital Holdings, Inc., Form 8-K, 30 July 2021, entry into a material definitive agreement: “On July 30, 2021, Marathon Digital Holdings, Inc. ... entered into a fully executed contract with Bitmain to purchase an additional 30,000 S-19j Pro ASIC Miners... The purchase price is $126,000,000.00.” A single filing; other miners' purchases are not shown.
Explore
Support the work

Reader-funded, on purpose.

No ads, no sponsors, no paywall — paid for by the readers it’s written for, in bitcoin.

Support the work