I posted this internally to Medium employees on March 5, 2014. See Hatching Inside Medium for more context. We’ve since done a lot of work in this area (with more to come).

A Brief History of Collections

In the beginning, every Medium story had to be published into a collection. That collection had to be selected before you started your story (“New Post”) only existed on collection pages. And a story could be in only one collection.

Part of the original reasoning for this was because we had multiple post “types.” Types had different templates that defined their structure and form. When you created a collection, you defined what template posts in that collection would use. I.e., you were defining “a collection of what?”

Two big things changed: We decided posts should be able to live in multiple collections. One reason was to maximize distribution. Another was to just make it a way easier decision where to put your post. Personally, I found it very difficult to decide where to publish a particular story when it could only be in one place. Was it the right place? Would people see it? What if I chose wrong? I also found it very hard to decide how to define collections — what was the right level of granularity: Food or Fruits and Vegetables? Now we could let a million collections bloom, and see what took off.

One thing that enabled a many-to-many posts/collections relationship was the decision to kill templates and post types. We still wanted content of all shapes and sizes to be published in Medium, but templates seemed like a limiting, brittle, and messy way to get there. Thus the emergence of the “flex template” (still not flexible enough).

Next, we changed the rule that you had to publish to a collection at all. This was to further reduce friction to publishing. Why put up a barrier of making people decide where to put something if they just wanted to write a little story and share it on Facebook? That didn’t preclude it from later being added to one or many collections, of course. But this was a controversial decision, because we worried if we allowed posts to exist sans-collection, than maybe nobody would bother putting their posts in collections, and collections would get the momentum they needed to drive more discovery at all. We’d have a bunch of orphaned, non-connected posts. That turned out not to be a big issue.

Next, we finally added the ability to explicitly follow collections. For the first time, this made it possible to have an idea what the distribution of a story would be if added to a collection.

Last, of course, we change the rules of collections, so there was no such thing as an open collection. They were all moderated. This, in combination with following, made it possible to promote and build and audience for a collection.

Enabling Publications

Collections, as originally defined, were somewhere between tags and publications. Tags (as they generally exist on other systems) are open organizing systems where anyone can define a label and anyone can throw things under that label. Tags are not owned or managed. Publications are owned and managed and generally not open. Collections could (and usually were) open but they were also owned and (optionally) managed.

With following and moderation, for the first time we enabled collection owners to build and curate content to an identifiable (and measurable) audience. Without these changes, it was fairly impossible to get most of the theoretical benefits we wanted to out of collections. With these changes, it’s possible, but let’s be clear: There’s a hell of a lot missing. The lack of branding, external distribution tools, measurability, monetization, and control make Medium a pretty poor platform for building an actual publication today.

The couple things it has going for it are: a) A good editing environment. b) A decent amount of content you can syndicate (for free). c) Some built-in audience you can potentially tap into (though it’s not clear how).

If we want to make our vision of meaningful publications building meaningful audiences and even meaningful revenue to pay for meaningful professional content — and I still think we do — we need to invest heavily in this area.