How Mesh Networks Could Replace Centralized Communication

The illusion of a distributed internet is maintained by the experience of connecting to any website from any device. What's obscured is the physical and organizational reality underneath.

Approximately 95% of intercontinental internet traffic travels through roughly 400 undersea fiber-optic cables. These cables surface at a small number of landing stations — often in politically and geographically predictable locations. The United States, the United Kingdom, and a handful of other states have intelligence agency access to major cable landing points. This is not conspiracy theory; it is documented fact, confirmed by the Snowden revelations and subsequent reporting.

Above the physical layer, routing is managed through a system of Autonomous System Numbers (ASNs) — essentially large networks that agree to exchange traffic. Most internet users' traffic passes through a small number of Tier 1 providers: AT&T, Cogent, Lumen, NTT, Telia. These providers peer with each other at Internet Exchange Points (IXPs), physical locations where networks connect. There are fewer than 1,000 significant IXPs globally, concentrated in major cities.

Above that: the concentration of web services. As of 2023, Amazon Web Services, Microsoft Azure, and Google Cloud together host an estimated 60-70% of global cloud workloads. This means a significant portion of all internet content and services runs on infrastructure owned by three American companies subject to American law, American intelligence oversight, and the business decisions of American executives.

The architecture described as the internet is, for most practical purposes, a highly centralized system with choke points at the physical layer (cable landing stations), the routing layer (Tier 1 providers and IXPs), and the application layer (major cloud providers and platforms). Anyone who can coerce or compromise these choke points can surveil, filter, or terminate communication for large populations.

A mesh network eliminates the choke point architecture by making every node both a consumer and a relay. In a radio frequency mesh network, a device receives a signal, processes the portion addressed to it, and rebroadcasts the rest to neighboring nodes. In an IP mesh network, routing protocols continuously map available paths and reroute data around failed or compromised nodes.

The topological consequence: there is no single point the adversary can target to bring down the network. Shutting down a mesh network requires shutting down every device. In a sufficiently large and distributed mesh, this is practically impossible for any actor short of physical invasion.

This is not merely a technical design choice. It is a political statement embedded in architecture. Who controls communication determines who controls information. Centralizing communication is a power concentration mechanism. Distributing it is a power distribution mechanism. The design of the network encodes a theory of governance.

The internet's original designers understood this. Paul Baran's 1964 RAND memoranda on distributed communications explicitly identified centralized and decentralized networks as politically distinct, not merely technically different. Baran's advocacy for distributed networks was motivated by resilience against Soviet attack, but the political logic extends further: any actor who wants to dominate communication benefits from centralization; any actor who wants to resist domination benefits from distribution.

Several mesh networking approaches have demonstrated real-world viability at different scales.

Low-range device-to-device mesh (Bluetooth, Wi-Fi Direct): Applications like Bridgefy, Meshtastic, and Briar create short-range mesh networks between smartphones without requiring any internet infrastructure. Range is limited — typically 100-300 meters per hop — but chains of devices can extend coverage significantly. These protocols have been used in protest environments (Hong Kong, Belarus) and disaster scenarios (earthquake response in Nepal) where infrastructure was unavailable or compromised.

Long-range radio mesh (LoRa, HAM radio mesh): Meshtastic uses LoRa radio modules to create mesh networks with ranges of several kilometers per hop in open terrain, enabling communication across significant distances without internet infrastructure. Amateur radio operators have built AREDN (Amateur Radio Emergency Data Network) mesh networks in numerous countries for emergency communication. These networks operate outside the commercial internet entirely.

Community broadband mesh (802.11 Wi-Fi, 60GHz point-to-point): Community networks like Guifi.net in Catalonia, Freifunk in Germany, and NYC Mesh in New York City build distributed broadband infrastructure using standard Wi-Fi equipment. These networks are community-owned, governed by their users, and do not depend on commercial ISP infrastructure for internal communication. They typically peer with the commercial internet for external access, but their internal routing is fully mesh.

Satellite-assisted mesh: Starlink and similar low-earth-orbit satellite constellations introduce a new variable. They are not mesh networks, but they provide redundant physical layer connectivity that can be combined with mesh distribution at the ground level. Communities that install local mesh networks with a single Starlink uplink gain both local resilience and global connectivity.

Delay-tolerant mesh (DTN — Disruption-Tolerant Networking): For environments where continuous connectivity is impossible — remote rural areas, disaster zones, maritime contexts — DTN protocols enable "store and forward" communication: a device carries data until it encounters another node, then passes it along. This is how some postal systems have operated for centuries, now implemented in software.

The technology is achievable. The harder problem is governance — who builds, owns, maintains, and sets policy for a mesh network.

Commercial ISPs solve governance through market mechanisms and regulatory compliance. The operator owns the infrastructure, sets the terms of service, and is subject to the legal requirements of their jurisdiction. This is efficient but produces centralization.

Community mesh networks must solve governance through collective decision-making, which is harder. Guifi.net developed the Procomuns model — a commons-based peer production framework — that specifies rights and obligations for network participants. The network is owned by its users in aggregate, managed by a non-profit foundation, and governed by a set of openly accessible rules. This model has sustained a 35,000-node network for over twenty years.

NYC Mesh uses a simpler model: all rooftop nodes are volunteer-installed, the network is operated by a non-profit, and participation is free. The model depends on volunteer labor and small-donor funding rather than revenue. This limits scaling capacity but eliminates the commercial pressure to monetize user data.

The governance question at civilizational scale is whether mesh communication infrastructure can be maintained as a commons — like roads, parks, and public water systems — rather than as a commercial service. Several countries have begun treating broadband access as a public utility, with state investment in infrastructure. Applying this logic to mesh architecture would require a policy shift that most governments currently resist because centralized infrastructure is easier to regulate (and surveil).

Mesh networks are not historically unprecedented. They are the native architecture of human communication before the invention of centralized systems.

Oral cultures operated on mesh principles: every person was both a receiver and a relay. Information spread through chains of personal contact. The topology was fully distributed. Centralized communication — state postal systems, royal messengers, wire telegraphy — represented a departure from this default, enabling faster long-distance transmission at the cost of centralized control.

The printing press created a semi-distributed information architecture: texts were produced centrally but could be copied and distributed across personal networks. The pamphlet culture of the English Civil War and the American Revolution was a mesh-like information system: no central node, many relay points, high resilience to censorship (burning one press did not stop the information).

Radio in the 1920s began as a distributed medium — anyone with a transmitter could broadcast. Regulatory centralization (frequency licensing, the FCC in the US) transformed it into a centralized medium. The same regulatory pattern could apply to mesh networking if governments choose it. The political contest over mesh network regulation is ultimately a contest over whether the distributed architecture of early radio or the centralized architecture of broadcast television will govern the next generation of communication.

What kinds of political arrangements does communication architecture make possible?

Centralized communication is compatible with both democracy and authoritarianism, but it makes authoritarianism easier. A government that controls communication infrastructure can suppress information about its own actions, prevent coordination among dissidents, and maintain a version of events that serves its interests. China's Great Firewall is the most developed version, but the capacity exists anywhere the infrastructure is centralized enough.

Distributed mesh communication makes information suppression structurally harder. It does not make it impossible — end-to-end encryption, traffic analysis, and physical persecution of network participants can all compromise mesh networks. But the cost and difficulty of suppression rise significantly when there is no central node to target.

This is why authoritarian governments have consistently moved to regulate, register, or prohibit mesh networking equipment and software. Bridgefy was removed from Hong Kong app stores. Russia has moved to require domestic routing of internet traffic through state-controlled exchange points. These are attempts to reimpose centralized architecture on a distributed medium.

The civilizational argument for mesh infrastructure is not primarily about efficiency or cost — commercial centralized networks often win on both dimensions. It is about the structural preservation of communication freedom. A civilization in which communication requires the permission of a central authority is structurally vulnerable to the corruption or capture of that authority. A civilization in which communication is distributed across millions of autonomous nodes is structurally resilient to such capture.

Building mesh network capacity, at community scale, is a hedge against civilizational communication failure — whether that failure comes from authoritarian capture, corporate consolidation, natural disaster, or the fragility of infrastructure no one is responsible for maintaining.