Reed's Law

Reed's Law, formulated by David P. Reed in 1999, states that the value of a network that supports group formation scales exponentially — specifically as 2ⁿ, where n is the number of participants. This dramatically outpaces Metcalfe's Law (which predicts value scaling as n²) because Reed's Law counts not just the connections between individuals but the number of possible subgroups those individuals can form. With even modest network sizes, the number of possible subgroups becomes astronomically large.

From Connections to Communities

The three classical laws of network value describe increasingly powerful scaling dynamics. Sarnoff's Law says the value of a broadcast network scales linearly with the number of viewers — each new viewer adds the same marginal value. Metcalfe's Law says the value of a communications network scales as the square of the number of users, because each new user can connect with all existing users. Reed's Law goes further: in a network where users can easily form groups, the combinatorial explosion of possible subgroups means value grows exponentially.

The mathematical intuition is straightforward. In a network of n people, the number of possible subgroups (of any size, from pairs to the full network) is 2ⁿ − n − 1. For 10 people, that's about 1,000 possible groups. For 50 people, it's over a quadrillion. In practice, not every possible subgroup forms — but the potential for group formation is what creates value, because the network enables communities that participants couldn't have assembled on their own.

Why Group Formation Matters

Reed's Law explains why social platforms and community-driven networks are so much more valuable — and so much stickier — than simple communications networks. A telephone network connects individuals; a social network like Discord or a metaverse platform enables communities, guilds, creative collaborations, and movements that take on lives of their own.

As Jon Radoff has argued in Network Effects in the Metaverse, the degree to which a network facilitates interconnections determines the extent of its emergent creativity, innovation, and wealth. Hub-and-spoke networks (where a central authority mediates all connections) capture less of Reed's Law potential than scale-free networks, where participants can form direct connections and self-organize into subgroups. The distinction matters enormously: hub-and-spoke architectures concentrate value at the center, while scale-free architectures distribute it across the network, generating emergent properties — groups of nodes that produce exponentially greater creativity and problem-solving than any central node envisioned.

This is why multiplayer games function as social networks: players join guilds, form teams, build communities around shared experiences, and develop friendships that outlast any individual game session. The game provides the substrate; Reed's Law dynamics provide the retention and value. Roblox's stickiness comes less from any individual experience and more from the social graph — the subgroups — that players have built within it.

Open vs. Closed Network Architectures

Reed's Law has a crucial implication for platform design: networks that make subgroup formation easy and permissionless capture more value than those that constrain it. Open-source communities, Discord servers, and the broader internet all exhibit strong Reed's Law dynamics because anyone can form a group around any interest without asking permission. Walled gardens — platforms that restrict how users can organize, communicate, and create — artificially suppress the exponential potential.

This creates a tension with platform taxes. Platforms that maximize Reed's Law value (by enabling permissionless group formation) also maximize the surface area for value creation that occurs outside the platform's direct control. The most successful platforms resolve this by providing infrastructure and taking a sustainable cut, rather than trying to control every interaction — what Bill Gates described as creating a platform where "the economic value of everybody that uses it exceeds the value of the company that creates it."

For the agentic web, Reed's Law suggests that the most valuable AI ecosystems won't be closed model providers serving individual users, but open platforms where agents can form dynamic multi-agent collaborations — subgroups of specialized agents assembled on demand to tackle complex tasks. The combinatorial explosion of possible agent teams mirrors the combinatorial explosion of possible human communities, and the economic implications scale accordingly.