Network Effects: Strong Ties Beat Node Count

Reframing where network value lives

The conventional account of network effects, often attributed to Metcalfe's Law, holds that a network's value scales with the square of its number of users. The strategic conclusion that follows is to maximize total user count as quickly as possible. Evan Spiegel has described a different reading of where the value actually sits, drawn from building Snapchat against competitors with far larger user bases.

In his account, value comes from whether the specific people a user communicates with are present, especially the people they communicate with most often, rather than from the raw count of nodes. "You don't need 500 friends on Snapchat. You just need your best friend on Snapchat. And that's what helped the service really grow and take on these much larger competitors."¹ A single close friend, in this framing, can represent as much as half of a person's communication value. Once that relationship is on the platform, most of the potential value has been captured without the remaining connections. A small, communication-dense network can therefore be worth more to its members than a large, sparse one.

How Spiegel applied the idea

Spiegel treats several strategic consequences as following from this reframe. The first is that a young product can compete against a much larger incumbent early, because it does not need to beat the incumbent on total users to be valuable to a person whose five closest friends are already on it. The second is that organic growth tracks the close-friend graph rather than the full social graph: users invite the people they actually talk to, which he treats as a more efficient growth vector than seeding from a full address book or friend list.

The third consequence he draws concerns switching costs. In his telling, stickiness is relationship-specific rather than platform-wide. A user loses value when the particular people they communicate with leave, and retains it when those people stay, largely independent of activity at the edges of the network.¹

The Poke case as evidence

Spiegel points to Facebook's Poke, a direct clone of Snapchat's features built on Facebook's existing graph, as the clearest illustration. Facebook had more users and stronger distribution and still did not displace Snapchat. In his reading, the two graphs were optimized for different things: Facebook's was a weak-tie network built around acquaintances and public signaling, while Snapchat's was a strong-tie network built around private conversation between close friends. The clone copied the features but not the communication context in which the features mattered.¹ Spiegel connects this to his broader view that features alone do not defend a product, a claim developed further in No Moat in Software.

Attribution and open questions

This is Spiegel's own account of Snapchat's early advantage, given in a single interview after the outcome was known, and it is framed as an explanation of a company that succeeded rather than a tested prediction. It leaves open how far the strong-tie framing continues to apply as a platform matures, since Snapchat itself now has roughly 500 million daily active users, at which point the behavior may revert toward the node-count model it was set against. It also leaves open whether close-tie density and communication frequency can be measured directly enough to serve as a forward indicator of defensibility rather than an after-the-fact story.