I’ve built systems for places where downtime means a missed shift at a hospital or a gap in evidence at a prison, not a bad tweet. That kind of work teaches you to ask an unglamorous question before anything else: where does this actually break, and who is responsible when it does?
Blockchain has never given me a satisfying answer, and the longer I sit with it, the more convinced I am that decentralisation, in the pure form it’s sold to us, isn’t achievable. The cryptography is strong. The trouble is that somewhere along the chain you’re always leaning on a centre, whether political, monetary, or societal. The real question is whether anyone’s willing to admit it.
Take Bitcoin, the system built specifically to prove you don’t need a bank or a government watching over your money. Mining is meant to be distributed across thousands of independent operators, and technically it is. In practice, a handful of mining pools have controlled a majority of the network’s hash rate for years at a stretch. When two or three organisations can theoretically coordinate more than half the computing power securing the world’s best-known cryptocurrency, you don’t have decentralised trust. You have a small committee with better marketing.
Ethereum has told a similar story since its move to proof of stake, where the network is secured by people locking up ether rather than by mining. A single liquid staking provider has, at various points, controlled close to a third of all staked ether, meaning a significant share of the network’s validation now runs through a single protocol and the decisions of whoever governs it.
Then there’s the plumbing underneath the apps people actually use. Most wallets and decentralised applications don’t talk to the blockchain directly. They go through a handful of infrastructure providers who run the nodes (the computers that keep a copy of the chain) on their behalf. When one of those providers has an outage, large parts of the supposedly decentralised ecosystem stop working, because the decentralisation was theoretical all along and the real architecture routed through a couple of companies.
Add the fact that most of that infrastructure runs on the same two or three cloud providers as everyone else’s software, that its code lives on GitHub, and that it gets installed through the same package registries every other piece of software depends on, and the pattern is hard to miss. The chain is public and distributed. Almost everything that makes it usable by an actual human being is not.
Money makes the dependency even harder to escape. A cryptocurrency’s value is still measured against pounds, dollars and euros, so its usefulness is still tethered to the monetary systems it was meant to route around. Stablecoins, which make a huge share of crypto trading and lending possible, are backed by real cash in ordinary bank accounts, managed by companies that can freeze funds, answer to regulators and occasionally turn out not to have quite as much cash as they claimed. Getting money in or out of crypto at any meaningful scale still runs through an exchange, and that exchange still needs a banking relationship, a licence, and a government willing to let it operate.
The 2022 collapse of FTX, one of the largest exchanges in the industry, trapped billions of dollars of customer money that people had been told, implicitly and sometimes explicitly, was safely outside the reach of any single point of failure. It wasn’t. It was sitting in one company’s accounts, subject to one company’s decisions, the same as money in any bank, minus the deposit protection.
And when the system’s own rules produce an outcome the community doesn’t like, the centre reveals itself completely. In 2016 a bug in the smart contract behind a project called The DAO let an attacker drain a large chunk of its investors’ funds. The DAO had been built on the promise that code, not people, would govern the outcome, but the code’s outcome was not allowed to stand. Developers and the wider community deliberated, then rewrote history with a hard fork, a coordinated change to the rules that reversed the theft and split the chain in two.
I don’t think that was the wrong call, and it proves the point. When it actually mattered, a system built to remove human judgement from the equation was saved by human judgement, exercised by a small group of people with enough influence to make it stick. That isn’t a bug you can patch out. Somewhere, a person or an institution has to be able to say what’s true, settle a dispute, or step in when the automated version of justice produces a monstrous result. Courts, governments, and social trust exist because someone has to be the backstop, and no amount of clever cryptography removes the need for a backstop. It just moves the argument about who gets to be one.
None of this makes blockchain worthless. Transparent audit trails, provenance tracking for goods moving through a long supply chain, and cheap, fast cross-border settlement are genuinely useful applications of the technology, and I’d rather see engineering effort spent making those work well than arguing the whole industry into a trustless utopia it was never going to reach. My objection is to the story wrapped around the tool: the idea that removing a visible centre also removes the need for one, when in practice it relocates the centre somewhere less visible and considerably less accountable.
A government or a bank at least has to answer to courts, regulators, elections and journalists. An anonymous founding team, a foundation registered somewhere convenient, or a handful of early investors holding most of a project’s governance tokens answer to far less.
Many years ago I built a system for environments where failure really wasn’t an option: asset capture and reporting across military bases, prisons and hospitals, places where the network drops out constantly and the data still has to be right when it comes back. It’s about as far from crypto’s founding myth as software gets. It’s centrally administered. There’s a team responsible for it, an organisation whose name is on the contract, and a chain of accountability that runs all the way up if something breaks. We didn’t hide any of that.
Instead, we designed for the specific ways centralised systems tend to fail: offline-first architecture so a bad connection doesn’t stop the work, strong encryption and access control so a breach doesn’t become a catastrophe, and redundancy so no single server going down takes the whole thing with it. The system has a centre, and everyone using it knows exactly where that centre is, who’s behind it, and what happens if it lets them down. That’s the engineering discipline blockchain’s marketing quietly assumes it can skip.
I keep coming back to the same point. Every system, decentralised or otherwise, has a point where trust has to sit, whether that’s a validator cartel, a stablecoin issuer, a cloud provider, a court, or a government. The interesting engineering question is what you do once you’ve found that point, and whether you’re willing to say so out loud. A central system that’s upfront about its weak points, that lets you see who’s accountable and what happens when things go wrong, will beat a system that markets itself as trustless while quietly reassembling all the same power structures out of sight, every time.
We don’t have a decentralisation problem to solve. We have a trust and accountability problem, the same one we’ve always had, and no ledger, however distributed, was ever going to make it disappear.