Starting with the nodes, starting with the network

A reflection on Medium’s early strategy

One of the best decision the Medium team ever took at the early stage of the product was to focus on the network, rather than the particular nodes (the pieces of content produced by users).

Focusing on the nodes generally means focusing on raw numbers (the more content, the better) while focusing on the network means focusing on the strength and density of the links between nodes.

  1. Medium was invite-only when it launched. This is probably one of the best and well known strategies for building strong networks because it ensures that each additional user is already connected to an existing one.
  2. They first added an existing social layer to the platform, Twitter. It may seem obvious given Medium’s founder history, but actually make a lot of sense because both products are primarily driven by the interest graph (They added Facebook and Google later on, though).
  3. They built a recommendation engine before even giving users basic personalization options that most of the competing products already had.

There was a time where the internet was all about content, creating content, making it easier for people to produce content. Today we’ve reached a point where content is ubiquitous and the problem becomes relevance.

With a nodes focused blogging platform, you would discover relevant content through search engines or existing social network. But none of these solutions is really good.

Existing social network (driven by the interest graph) can be a good recommendation engine for users — I definitely discover a lot of Medium posts on Twitter — but falls short for the company. They’ll get really poor data.

As for the search engines, the massive amount of content, the fact that most people never go further than the first page of results, plus the way page rank algorithms works: You’ll get a lot of “institutional” content, not really fit for discovering dynamic new content.

A strong network gives a few unfair advantages

  • It provides algorithms with more and better training data, which means more relevant content for users.
  • It creates networks effect, networks effect may be hard and take long to start, but when they pick up they are an incredible force for growth.

But this is kind of a bet for early stage companies, where growth metrics are religiously checked every day (um, hour?). Focusing on the network will sometimes mean sacrificing potential growth sources for better network density and node relevance (just read about Facebook early expansion).

It may have become obvious how much strong networks matter, but the path to get there seems not yet well understood.

Focusing on nodes will often mean trying to build the most complete product. You always want a good product, good enough that people will come for the tool, stay for the network. But you also need to look up from you code editor and think more about the big picture. Focusing on the network means being thoughtful about what you build.