The problem with the twitter follow model

When the web first started people had static pages. These were typically personal web pages hosted on an ISP web server or on free hosting sites such as Geocities. These pages didn’t change all that much over time, but the number of pages kept growing.

Then with the popularization of programming languages like PHP and simple database products like MySQL people started to make dynamic websites. These websites changed their content dynamically depending on user-input or automatically based on attributes like time of day etc. However most websites even during this time were the same from user to user. As long as Jane and Jil were on the same page or used the same input parameters, they saw the same content with the obvious exception being web-based email.

The dynamic website era began to change the state of the web with the advent of what came to be known as Web 2.0. The Web 2.0 era was defined by a dramatic change in the relationship between website-user and website-owner. Whereas during static or dynamic web eras, most users were largely only consumers of content, Web 2.0 began to encourage users to contribute content. This came in the form of blogs, social networks etc. On these user content dominated websites, curation became very important because not every user generates good content or content that is of interest to a large number of other users. Moreover the volume of content grows rapidly with a low signal to noise ratio for most users.

Curation has taken two major forms — the active follow model like Twitter and the passive follow model like Facebook. The active follow model is easy enough to describe — you actively and explicitly express which people/accounts you are interested in following and content from those entities shows up in your feed. The passive follow model of Facebook is slightly more indirect. In the passive follow model you hint your interest by interacting with the site. The root of your interest start with adding or accepting friend requests or by liking pages. Then you gradually refine your interests by liking/commenting/viewing. The model is passive because the expectation that gets set is that you don’t explicitly follow anything, you express interest by actually interacting with the content produced.

At the end of the day, both models are trying to capture what content interests you. Since we humans are not very good at precisely describing what we are actually interested in, both models use proxies. The proxy in the active follow model is the entity that you follow. For e.g. I follow Paul Graham because I am interested in the startup economy. The proxy in the passive follow model is your interaction with the content. The more you interact with a certain type of content, the stronger interest you express in it implicitly.

In my opinion the active follow model is problematic for the following reasons

• Need for Active Curation — The active follow model puts the burden of curation squarely on the shoulder of the users. The users must continue to refine their follow lists as their interests evolve. This creates friction that can discourage a user from revisiting. The passive follow model can get richer and better as you add more roots, but it doesnt depend on it to bring you content of interest.
• Aspirational v/s actual interest — When people have to decide to explicitly follow an entity, they tend to express as much aspirational interest as they express actual interest. Aspirational interest is a person’s desire to develop an interest in something they “think” is cool or something which allows them to project an image. An example of this would be my interest in Ancient Rome. While its cool to express that interest through an active follow, actually reading and learning more about a new topic is tough and hence people’s follow lists become like New Year resolutions. The passive follow model allows you to express aspirational interest by say liking a page, but since it doesnt necessarily show you content from that page if you stop interacting with it, it evolves with you
• Proxies — While users/accounts can be a good a proxy, they aren’t always. For e.g. while I follow Paul Graham for his views on the startup economy, I have very little interest in his other interests. Herein lies the biggest advantage of the passive follow model. The passive follow model can directly understand and reflect the content you interact with rather than a proxy for that content.

I think eventually everyone including twitter will abandon the active follow model to allow users to discover great content without having to constantly tune their follow lists. Moreover as machine learning becomes more and more accessible, more companies will start with a passive follow rather than starting with active follow and moving to passive follow as they grow.