The unfinished argument for decentralisation
Decentralised networks retain a powerful appeal because they answer a real problem. Digital life has become dependent on a relatively small number of intermediaries: cloud providers, app stores, payment rails, identity brokers and telecoms operators. That concentration can deliver convenience and scale, but it also creates brittle points of failure, chokepoints for speech and commerce, and asymmetries of power between operators and users. The attraction of decentralisation is therefore not ideological romanticism. It is a practical response to over-centralisation in the systems on which societies increasingly depend.
But the case is too often made in overly mechanical terms. If only information were spread across many nodes, the argument runs, power would also be dispersed. History suggests otherwise. Networks can be decentralised in topology while remaining highly centralised in governance, incentives or operational control. The internet itself is a case in point: designed as a distributed network of networks, it nonetheless evolved layers of concentration in access, hosting, search, social distribution and content moderation. Technical architecture matters, but it does not settle the question of who gets to decide.
Distributing data across many machines is not the same thing as distributing power across many people.
This distinction matters because decentralised systems are maturing from subcultural experiments into infrastructure with public consequences. Once a network aspires to carry finance, communications, archives, identity or civic co-ordination, its defects stop being private inconveniences and become governance questions. Who bears responsibility when harms arise? How are rules updated? How are disputes resolved? Who pays for maintenance no one notices until it fails? These are not peripheral concerns. They are the centre of the decentralisation debate.
Why centralisation keeps returning
Centralisation is not merely the result of bad intent. It is often an emergent property of efficiency. Users gravitate towards convenience; developers towards common standards; operators towards economies of scale; investors towards winners that compound. Network effects reward aggregation. So do compliance costs, cybersecurity demands and the simple economics of uptime. In this environment, decentralised alternatives face a structural handicap: they must compete not only on principle but on performance, reliability and ease of use.
This is why many ostensibly decentralised systems recentralise over time. Control migrates towards the parties who write and maintain core software, run the most reliable infrastructure, hold the most data, provide the main user interfaces or finance the ecosystem. The result is not necessarily monopoly in the classic sense. It is often something more diffuse: a stack in which formal control appears distributed but practical influence pools around a small set of actors. Researchers at the Internet Society and the Internet Architecture Board have long warned that internet consolidation can occur even when underlying protocols remain open.
The lesson is sobering. Decentralisation is not a one-off design choice but a continuous political and economic struggle against gravity. Without explicit counterweights, convenience and capital will pull systems back towards centres of control.
The internet’s original settlement is under strain
The early internet succeeded because it combined distributed design with institutional pluralism. Its technical standards were developed through open processes. Its routing model allowed autonomous networks to interconnect without a single sovereign operator. Its governance culture, however imperfect, rested on rough consensus, interoperability and permissionless innovation. This settlement made the network resilient and generative.
Distributing data across many machines is not the same thing as distributing power across many people.
Yet the same settlement also left gaps. Open standards did not prevent concentration at higher layers. Interoperability did not guarantee competition in user-facing services. And permissionless entry did not solve the problem of sustainable stewardship for common infrastructure. Much of the internet’s foundational software has long depended on underfunded maintainers, as repeated security incidents have shown. The Cybersecurity and Infrastructure Security Agency, among others, has highlighted how insecure and unsupported open-source components can become systemic risks when embedded across critical systems.
In other words, the internet already teaches the central paradox of decentralised networks: openness is necessary, but not sufficient. Common protocols can lower barriers to entry, but they do not by themselves produce healthy institutions. Where stewardship is weak, dependency accumulates quietly until failure makes it visible.
Resilience is more than redundancy
Advocates of decentralised networks often emphasise resilience, and rightly so. Systems with multiple paths, diverse operators and no single point of failure are generally better able to absorb shocks. During natural disasters, censorship attempts or infrastructure outages, distributed architectures can preserve some continuity even when centralised systems go dark. This is one reason public interest in community networks, federated communications and distributed storage has persisted.
But resilience is frequently misunderstood as a purely technical property. In practice, resilient systems also require social capacity: trained operators, trusted governance, clear procedures and financial continuity. A mesh network, for instance, may route around physical damage, but only if people know how to deploy and maintain it. A federated service may avoid reliance on one platform, but only if administrators can manage abuse, upgrades and legal obligations. Redundancy in hardware or topology does not remove the need for competent institutions; it raises the premium on them.
Resilience is not achieved when a network can survive failure in theory, but when communities can keep it functioning in practice.
That points to a more grounded metric for decentralisation. The question is not simply whether a system can operate without a central authority. It is whether many different actors can operate it well, at reasonable cost, under ordinary and adverse conditions alike. If the answer is no, then decentralisation exists mainly on paper.
Governance is the real bottleneck
The most difficult challenge in decentralised networks is not packet routing, replication or consensus. It is governance. How should a network evolve when stakeholders disagree? How should it respond to harmful content, fraud, sabotage, security flaws or state pressure? Which decisions should be encoded in software, and which should remain subject to human judgement? These are constitutive political questions, not implementation details.
There is a temptation among engineers to treat governance as a layer that can be minimised by better protocol design. Sometimes this works: clear rules, transparent logs and automation can reduce arbitrariness. But only up to a point. Real-world systems encounter ambiguity. They have users with unequal expertise, asymmetric risk and conflicting interests. They must navigate law, norms and emergencies. In such conditions, governance cannot be eliminated; it can only be obscured or made accountable.
The better tradition in internet governance has never denied this. Bodies such as the Internet Engineering Task Force and standards organisations thrive not because they remove politics, but because they channel it through procedures that are legible, contestable and revisable. Decentralised networks need analogous arrangements at every layer: not a fantasy of frictionless autonomy, but institutions capable of handling disagreement without collapsing into arbitrary rule.
The financing problem few enthusiasts solve
Resilience is not achieved when a network can survive failure in theory, but when communities can keep it functioning in practice.
Infrastructure requires boring money. Servers need replacing, documentation updating, vulnerabilities patching and users supporting. The trouble is that decentralised networks often struggle to finance these functions on a durable basis. Where no single operator captures enough value, underinvestment is common. Where value can be captured, control tends to follow. This is one reason many public goods on the internet oscillate between volunteer exhaustion and dependence on a handful of patrons.
The challenge is especially acute for systems that aspire to civic relevance. If a network is meant to support local communications, archives, public-interest media or shared digital identity, it cannot rely indefinitely on enthusiasm alone. It needs recurring revenue, pooled procurement, co-operative ownership, public funding, foundation support or some blend of the above. Each model carries trade-offs, but avoiding the question simply invites hidden centralisation later, when the best-resourced actor steps in as de facto governor.
The recent resurgence of interest in digital public infrastructure is useful here. Institutions such as the United Nations Development Programme and the World Bank have argued that foundational digital systems should be treated more like public utilities than speculative products. That logic applies equally to decentralised networks. If they are to serve public ends, they need public-interest funding and fiduciary norms, not just clever code.
Interoperability is the understated prize
If the strongest claims made for decentralisation are often overstated, one claim is arguably understated: its role in preserving interoperability. The deepest value of decentralised design may not be that everyone runs their own node, but that no single actor can unilaterally set the terms of participation for everyone else. Open protocols and federated models can make exit more feasible, entry less costly and lock-in less severe. They give users and organisations room to move.
This matters because competition in digital markets increasingly depends on the ability to connect across services rather than to dominate within one. Interoperability can reduce dependence on vertically integrated gatekeepers without requiring a world of atomised, self-hosted computing. In some contexts, a plural ecosystem of compatible providers may be both more realistic and more beneficial than maximal decentralisation. The objective is not purity. It is contestability.
The most important freedom in a networked society may be the freedom to leave without disappearing.
Policy is beginning to catch up with this reality. Debates around data portability, interoperability duties and open standards are no longer niche concerns. They are becoming central to how governments think about digital competition and resilience. Decentralised networks should be understood within that broader settlement: as mechanisms for keeping systems open enough that power remains negotiable.
Community networks show what theory often misses
Some of the most instructive examples of decentralisation come not from grand digital manifestos but from local, community-led infrastructure. Community broadband and wireless networks have demonstrated that distributed ownership can expand connectivity where commercial incentives are weak or misaligned. The Internet Society has documented cases in which communities built and governed their own access networks, tailoring operation to local needs and constraints.
These examples matter because they correct a common misconception. Decentralisation is often imagined as disintermediation: users liberated from organisations altogether. In practice, successful community networks depend on organisations of a different kind. They require local leadership, training, maintenance routines, shared rules and partnerships with public authorities or civil-society groups. Their strength comes not from the absence of institutions, but from institutions being closer to the people they affect.
That is a more persuasive democratic vision than the rhetoric of total autonomy. It recognises that interdependence is unavoidable. The question is whether that interdependence is mediated by distant, opaque centres or by accountable structures with meaningful local participation.
The most important freedom in a networked society may be the freedom to leave without disappearing.
States will not stay at the edge
Any serious argument about decentralised networks must reckon with the state. Governments are not external to digital infrastructure; they regulate spectrum, competition, privacy, cybersecurity, cross-border data flows and liability. In some contexts they also fund broadband, mandate standards or directly provide digital services. The notion that decentralised networks can flourish in a legal vacuum is not realistic.
Nor is state involvement inherently hostile. Public policy can preserve open standards, support community infrastructure, curb anti-competitive conduct and finance maintenance for shared digital goods. But states can also use centralisation to simplify surveillance, control information or favour national champions. Decentralised design can therefore serve as a constitutional safeguard as much as an engineering strategy: a way of ensuring that no single public or private actor can easily dominate the whole field.
The aim should not be to expel the state from networks, which is impossible, but to shape legal frameworks that reward openness, interoperability and pluralism. That means focusing regulation on outcomes such as portability, access and procedural accountability rather than mandating one architecture in all cases.
From ideology to public infrastructure
Too much debate about decentralised networks has been trapped between utopianism and dismissal. One side promises liberation through protocol alone; the other treats any departure from centralised platforms as impractical idealism. Both positions miss the more consequential middle ground. Decentralisation is neither a cure-all nor a relic. It is a design and governance discipline for building systems that remain open, adaptable and resilient under pressure.
That discipline begins with institutional realism. Not every function should be fully decentralised. Some services benefit from central co-ordination, especially where safety, performance or legal clarity are paramount. But centralisation should be justified, bounded and reversible, not accepted by default because it seems cheaper in the short term. The burden of proof ought to shift. Given the social dependence now placed on digital systems, concentration should be treated as a risk to be managed, not merely an efficiency to be admired.
This is where decentralised networks can make their strongest contribution. They can provide fallback capacity, reduce lock-in, widen participation in governance and create room for institutional diversity. Their value lies not in abolishing centres altogether, but in preventing any one centre from becoming inescapable.
What a mature decentralised future would look like
A mature decentralised future will look less dramatic than its early evangelists imagined. It will not consist of perfectly autonomous individuals operating outside institutions. It will consist of layered systems in which open protocols, federated services, community ownership and public-interest rules coexist. Some layers will be centralised for good reasons, but their power will be constrained by interoperability, portability and oversight. Others will be distributed by design, providing redundancy and competitive pressure.
Most importantly, success will be judged by ordinary civic criteria rather than technical novelty: reliability, accountability, accessibility, affordability and the capacity to correct mistakes. Decentralised networks will matter when they help communities communicate during disruption, when they reduce dependence on dominant intermediaries, when they support local autonomy without sacrificing broader co-ordination, and when they keep digital power contestable.
The future of decentralisation, then, is not anti-institutional. It is institutional in a different register. Protocols can open the field, but only governance can keep it open. The real task is not to design systems with no centre at all. It is to build societies in which centres exist, but can be challenged, exited and rebalanced before they harden into destiny.
