The invisible politics of Bitcoingovernance crisis of a decentralised infrastructure

Bitcoin was designed so that no one would be in charge. That was not an accident — it was the entire point. So here is the question that De Filippi and Loveluck spend their paper unraveling: if no one is in charge, then who decides when something needs to change? The ideological roots of Bitcoin run deep. The cypherpunk movement of the late 1980s and 1990s — a loose collective of hackers, mathematicians, and activists — argued that strong cryptography could protect privacy and enable direct, self-organized transactions without any third-party oversight. Bitcoin was the fullest technical expression of that belief. Supported by many libertarians, it was also read as a continuation of Hayekian ideas to "denationalize" money and strip states of their control over monetary issuance. The goal was radical: replace the social authority of banks and central banks with cryptographic rules. Two technical innovations made that possible. The first is the blockchain — a decentralized, append-only public ledger in which each block contains a cryptographic reference to the previous one. This means tampering with any past entry produces a visible break in the chain. The second is Proof-of-Work, a consensus mechanism where a network of validators called miners commit computational power to solve cryptographic puzzles, earning the right to record the next block of transactions and collecting newly generated bitcoins as a reward. The total supply is fixed at 21 million bitcoins, built directly into the protocol. The very first block, the Genesis block, was created in January two thousand nine and released fifty bitcoins. By the time De Filippi and Loveluck were writing, Bitcoin had reached a market capitalization of almost seven billion dollars. The system was explicitly designed to be "trustless." That word matters. It doesn't mean untrustworthy — it means trust in a central authority is replaced by trust in mathematics and publicly verifiable history. Anyone can check the ledger. No one can secretly alter it. That promise is real, and it is genuinely powerful. But here is where the paper's central argument begins. De Filippi and Loveluck draw a distinction that sounds simple and turns out to be everything. There are two layers of governance operating in Bitcoin simultaneously. The first is governance by the infrastructure — the rules encoded in the protocol itself, the algorithmic constraints that determine who can transact, how transactions are validated, and what behaviors earn rewards. This layer is visible, celebrated, and treated as Bitcoin's revolutionary contribution: social coordination handled by code. The second layer is governance of the infrastructure — the social process by which that protocol is actually designed, changed, and maintained. This layer is largely invisible. And invisible does not mean absent. When Satoshi Nakamoto, Bitcoin's pseudonymous creator, eventually disengaged from the project, the authority to commit code to the main repository passed to a small group of core developers. Anyone can submit a proposal. Only a handful of people can merge it. The formal mechanism for this is the Bitcoin Improvement Proposal, or BIP — a design document modeled on Python's PEPs and the internet's RFCs, intended to collect community input and document technical decisions. In principle, the process is open. In practice, De Filippi and Loveluck show that final acceptance rests with core developers assessing how much public support a proposal has gathered. One piece of project governance guidance put it this way: "ultimately consent may rest with the consensus of the Bitcoin users" — but the path to that consensus runs through a small technocratic core. The paper is precise about what this means: not a conspiracy, but a structure. Technical expertise creates natural gatekeeping. The people who understand the code well enough to evaluate a proposal are, by definition, few. But that structural concentration carries political weight that the open-source framing tends to obscure. Many disputes that look like engineering debates are, on closer inspection, deeply political choices about what Bitcoin is for, who it serves, and what trade-offs are acceptable. Nothing made this clearer than the block-size war. The Bitcoin blockchain was originally configured with a one-megabyte block-size cap. That parameter directly constrains transaction throughput: since a block is produced roughly every ten minutes, the cap determines how many transactions the network can confirm in a given period. As adoption grew, many participants argued the one-megabyte limit was preventing Bitcoin from scaling. Others insisted that raising it would centralize the network because larger blocks would require more powerful machines to store and validate, pricing out ordinary participants. When consensus failed, two core developers — Gavin Andresen and Mike Hearn — released Bitcoin XT on August fifteenth, two thousand fifteen. The XT client would accept blocks up to eight megabytes. The upgrade had a threshold mechanism: starting January eleventh, two thousand sixteen, if seventy-five percent of the most recent thousand blocks were signed by XT nodes, the larger block size would activate, with the cap then doubling every two years. The threshold was never reached. As of that date, only about ten percent of blocks had been signed by XT nodes. Bitcoin XT was abandoned on January twenty-third. The response was fierce. Gregory Maxwell warned that larger blocks would favor large mining operations over home computers, increasing centralization. Nick Szabo raised security concerns. Online debate escalated into censorship, personal attacks, and distributed denial-of-service incidents. Coinbase was removed from Bitcoin.org after announcing it would experiment with Bitcoin XT. Mike Hearn resigned, publicly declaring the project in crisis. Bitcoin XT's failure didn't settle anything. Bitcoin Classic emerged next, proposing a two-megabyte cap using the same seventy-five percent miner endorsement scheme, and at one point accounted for twenty-five percent of the network's nodes. What the whole episode revealed was that miners — specifically, large mining pools — had become a decisive political constituency that the original decentralized ideal had not anticipated. Here is the structural problem. Most of the network's hashing power is concentrated among a small number of mining pools, mostly based in China, which together hold over seventy-five percent of the total. In two thousand fourteen, a single pool briefly controlled more than half of the network's hashing power — creating the realistic possibility of a so-called fifty-one percent attack, where a majority actor could theoretically rewrite recent transaction history. The decentralized system had produced a highly concentrated power center, and that center had leverage over which version of the protocol the network would actually follow. This is the core irony the paper surfaces. Bitcoin was designed to eliminate the need for trusted intermediaries by encoding rules in mathematics. But the process of writing and changing those rules is deeply social, and the social process has its own centers of power — they are just less visible than a central bank. So what would legitimate governance of Bitcoin actually look like? De Filippi and Loveluck are careful here, and their caution is itself instructive. They point to internet governance precedents — the Internet Engineering Task Force's model of "rough consensus and running code," and contested bodies like ICANN — as evidence that sociotechnical systems cannot ensure their own legitimacy through technology alone. The Bitcoin Foundation, they note, was a first attempt at institutionalizing governance but never managed to establish itself because it lacked legitimacy and accountability. A more productive path, they suggest, is to treat the blockchain as a platform on which diverse rules can be encoded while deliberately constructing institutional frameworks on top. These institutions should acknowledge political trade-offs, broaden participation beyond core developers, and provide transparent deliberation. That last point is the one worth carrying out of this episode. The lesson is not specific to Bitcoin. It applies to the whole category of systems that promise to replace social trust with algorithmic rules. The code can do extraordinary things — eliminate certain kinds of corruption, enforce contracts without courts, and enable transactions across borders without banks. But it cannot deliberate. It cannot represent competing interests. It cannot hold itself accountable. Those functions require institutions — messy, imperfect, political ones. De Filippi and Loveluck's contribution is to show that pretending otherwise doesn't make the politics disappear. It just makes them invisible, and invisible power is the hardest kind to challenge. This lecture was created by ennepō. Go to https://ennepo.ai to Discover, Create and Follow the latest research in your field. Read when you can. Listen when you want to.