The difference between a service that dies and a protocol that survives is architectural. A service has an operator and a legal jurisdiction that can be pressured. A protocol has none of these, which is why a single injunction killed Napster while two hundred thousand lawsuits did nothing to BitTorrent. The design decision that separates them is whether the system requires any specific human being to continue operating. If it does, that human is the attack surface.
This principle extends to every layer of the stack. A smart contract deployed to a blockchain that no one can modify or withdraw falls outside the reach of sanctions law, because what belongs to no one cannot be sanctioned. Code that requires an operator to run a server or coordinate transactions creates a pressure point that the state will eventually find and squeeze. The architectural question every protocol designer must answer: after I walk away, does this keep running?
Governments have converged on five strategies for attacking systems they cannot shut down directly.
The first is prosecuting developers. When code itself is beyond reach, the people who wrote it become targets. Privacy tool developers have faced criminal charges for money laundering and operating unlicensed money transmitting businesses, with sentences measured in years and forfeitures measured in hundreds of millions. The legal theory that building a tool makes you responsible for its misuse has been applied even when the tool operates autonomously after deployment. For builders, the practical implication is clear: legal exposure correlates with operational involvement. Writing and deploying immutable code carries less legal surface area than running a service built on that code.
The second is controlling distribution. A protocol that cannot reach users exists only in theory. App store gatekeepers control software distribution for billions of devices and have removed privacy-focused applications without explanation or recourse. Any system depending on a single distribution channel has a single point of failure, regardless of how decentralized its protocol layer might be. Sideloading, alternative stores, progressive web apps, and direct binary distribution all provide redundancy, but each one narrows the potential user base. Building for multiple distribution channels from the start is cheaper than retrofitting after your primary channel disappears.
The third is attacking the network layer. Deep packet inspection identifies protocol traffic by its byte patterns, and governments with sufficient infrastructure block identified traffic at the ISP level. The most aggressive approach blocks all fully encrypted traffic that fails to match known protocol fingerprints, accepting collateral damage to legitimate services. The countermeasure is encrypted transport that makes protocol traffic indistinguishable from ordinary web traffic. When blocking your protocol means blocking HTTPS itself, only a government willing to break its own internet can sustain the censorship.
The fourth is exploiting centralization. Decentralized protocols tend to centralize in practice. When two mining pools control over fifty percent of block production under one regulatory jurisdiction, or one relay handles seventy percent of all posts, the protocol's theoretical decentralization is operationally meaningless. These concentration points create pressure that regulators can exercise at trivial cost. Dominant relays and large mining pools exist because users and operators choose convenience over distribution, and every one of them is a target.
The fifth is economic attrition through compliance costs. The global anti-money laundering system costs $275 billion per year and catches less than one percent of criminal proceeds. Banks spend hundreds of millions each on compliance to avoid regulatory penalties; crime prevention is incidental to the calculation. Each enforcement action against a protocol creates a new compliance category, generating demand for new monitoring tools. The compliance industry grows with each demonstration of its own ineffectiveness. The cost falls on every entity that interfaces between open protocols and the traditional financial system, creating a buffer zone of institutions too afraid of regulatory liability to touch anything novel.
The protocols that survive these attacks share a set of design properties, each one a direct answer to one of the vectors above.
Eliminate operator dependency. If the system requires a specific person or company to function, that person or company will eventually be pressured or arrested. The design target is a system that runs on any willing machine with no coordination from its creators. The test: if every original developer disappeared tomorrow, would the protocol keep running without them?
Build redundant distribution from day one. Satellite broadcasts requiring only a receiver dish and mesh networks operating on unlicensed radio spectrum both eliminate potential chokepoints at layers the state does not control. No single distribution channel should be necessary for the protocol to reach its users. The projects that treat distribution as an afterthought discover its importance when their primary channel vanishes overnight.
Encrypt all transport by default. Traffic that is identifiable is traffic that can be blocked. Encrypted transport that makes every byte appear uniformly random forces censors into an expensive choice: block all unidentified encrypted traffic and accept massive collateral damage, or allow the protocol through. For protocols operating in adversarial environments, this is a survival requirement.
Treat decentralization as a discipline. The natural tendency of every network is to concentrate around a few high-capacity nodes, and this tendency must be actively resisted through relay diversity, geographic distribution of infrastructure, multiple independent client implementations, and low barriers to running nodes. A protocol that launches decentralized and drifts toward concentration has a shelf life determined by how long it takes regulators to notice.
Intellectual honesty demands acknowledging the limits. Governments do not need to break the cryptography. They need to make the human interfaces painful enough that ordinary users never encounter the protocol at all. A system confined to sideloaded apps and terminal commands will serve its committed users, but it will not become the default communication layer for a billion people. The UX gap between centralized services and decentralized protocols remains the most effective form of soft censorship, and no architectural principle solves it alone.
The economic asymmetry cuts both ways. Protocol adaptation is cheap, but building usable software is expensive. The open-source projects that sustain these protocols operate on budgets that would not cover a single bank's compliance department. Survival and mass adoption are different problems requiring different resources, and most protocols will achieve the first without ever reaching the second.
The humans who build these protocols pay real costs: years behind bars and millions in forfeiture. The code they write pays nothing. It copies itself across every willing machine on the network, indifferent to the legal theories arrayed against its creators. The state can punish the hand that types, but it has never succeeded in erasing what was typed. For builders, the lesson is structural: design for a world where you are not around to run it, distribute through channels no single entity controls, encrypt so censors cannot identify it, and resist the centralization that makes all other defenses irrelevant.
