Metcalfe’s Law: Why 200 Members Feels Different Than 20

Something happens between member 20 and member 200 that has nothing to do with your content getting better. Metcalfe’s Law is the math behind it: the value of a network scales with roughly the square of its members, not the count itself. Twenty members give you about 190 possible connections. Two hundred members give you almost 19,900. You didn’t get ten times better at running a community. You got a hundred times more ways for people to find each other.

The math nobody does on purpose

Robert Metcalfe worked it out for Ethernet networks in the 1980s, but the formula doesn’t care what kind of network you’re counting. It’s n(n-1)/2, where n is the number of members and the result is the number of possible pairwise connections between them.

Run it yourself. Ten members: 45 possible connections. Fifty members: 1,225. A hundred: 4,950. Two hundred: 19,900. The curve isn’t gentle. It’s not even close to a straight line. Every member you add isn’t one more person in the room, they’re one more person everyone already in the room could now talk to.

This is why a ten-person community can feel like a ghost town and a two-hundred-person one can feel like it runs itself. Nobody redesigned anything. The room just crossed a threshold where enough connections existed for something to always be happening.

It’s worth being honest about where the analogy strains. Metcalfe worked this out to sell Ethernet cards, where every connection is genuinely identical and genuinely usable, a cable is a cable. People aren’t cables. A connection between two members who never actually talk contributes nothing, no matter what the formula says exists on paper. Some economists have argued real network value grows closer to n log n than n-squared, precisely because most possible connections in a large network never activate. That correction matters more than the original formula, and it’s the reason this piece leans on the shape of the curve rather than treating the exact number as a KPI you should chase to two decimal places. The point isn’t that your 200-member community is worth exactly 104x your 20-member one. The point is that it’s worth an order of magnitude more, and that order of magnitude is why growth compounds instead of adding up.

Why it feels like luck

Most people running a community never do this math, so the growth from 20 to 200 members reads as a mystery. One month nothing happens. A few months later there’s a comment thread on every post, people are tagging each other, someone starts a side conversation that has nothing to do with your original topic. It feels like the community found its own momentum.

It did. But the momentum was sitting in the math the whole time, waiting for enough members to exist for it to show up. You weren’t unlucky at 20 members. You were mathematically underpowered. There were only 190 possible connections in the whole room, and most of those never activate, because most pairs of people never happen to post at the same time, reply to the same thread, or show up in each other’s feed.

At 200 members, even a small percentage of 19,900 possible connections activating is enough to make the whole space feel alive. You don’t need everyone talking to everyone. You need a tiny fraction of an enormous number, and an enormous number was always going to beat a small one at producing that fraction.

Picture the same comment thread at both sizes. At 20 members, a decent post gets two replies, maybe three, and the thread dies there because there’s genuinely nobody left with something to add. At 200 members, that same post has thirty people who might plausibly reply, and even if only one in ten does, that’s still three replies, then a fourth person replying to a reply, then someone quoting it somewhere else. Same post. Same effort. Different pool of possible connections to draw from.

This is easiest to see in a real member directory, where every card is a person who could plausibly connect with every other card on the page.

The counterpoint nobody mentions

Here’s where it gets uncomfortable if you’ve read about Dunbar’s Number already. That piece makes the case that any one person can only hold about 150 stable relationships, no matter how big the group around them gets. Metcalfe’s Law says the network’s total value keeps compounding well past that. Both are true at the same time, and they’re not actually arguing with each other.

The network’s value and any one member’s usable slice of that value are different things. Total possible connections in a 500-person community is over 124,000. No single person is going to use even a fraction of a percent of that. What they will use is their own 150-ish, drawn from whichever smaller room inside the larger network they actually spend time in.

Sit with that for a minute. The n-squared math is real and it’s why a bigger community is worth defending. But it compounds at the network level, not the member level. A member doesn’t need access to all 124,000 connections to benefit from a network that has them. They need their own well-chosen 150, and a structure that makes those 150 easy to find inside a much larger space. That’s what sub-spaces are for: they let a community bank the compounding value of scale while keeping any one member’s actual experience inside a size a person can hold in their head.

The other half: how n grows in the first place

Metcalfe’s Law explains why each new member is worth more than the last one. It doesn’t explain how you get from 20 to 200 in the first place. That’s a different piece of math, closer to what the viral coefficient measures: how many new members each existing member actually brings in.

The two ideas stack cleanly. A low viral coefficient means n grows slowly, member by member, mostly through your own effort. That’s fine, and it’s normal. But it means you’re climbing the compounding curve one linear step at a time while the payoff is quadratic, which is exactly why the early months of any community feel so disproportionately slow compared to what comes later. You’re not doing anything wrong. You’re just early on a curve that looks flat until it very suddenly doesn’t.

What this actually changes about what you do next

If you take the math seriously, the obvious mistake becomes obvious too: treating headcount as the goal. A member who joins and never connects to anyone else adds exactly zero possible new value to the network, because the value was never in their existence, it was in the connections attached to it. A hundred members who never interact are worth less than fifty who do, even though the roster looks bigger.

So the real job isn’t recruiting. It’s making the first connection easy to find, fast to make, and likely to happen in the first session someone shows up. Every member who joins and leaves without connecting to a single other person is a data point the math already told you not to celebrate.

The n-squared curve is patient. It will reward you for getting from 20 to 200 members whether or not you understand why. But understanding why changes what you optimize for on the way there, and that’s the difference between a community that eventually gets lucky and one that gets there on purpose.