image: roberta pezas — 123rf

Twitter and the potential of recommendation technologies

Enrique Dans
Sep 16, 2013 · 5 min read

Right from the start, Twitter has maintained a close relationship with recommendation technologies. A network that began as a way for users to keep in touch with friends developed very quickly from something purely for personal use—such as knowing what was somebody up to by asking, “what are you doing?” to be able to follow what well-known people, companies, or news media.

The first phase in Twitter’s evolution ended when just about anything of any importance happening in the world was near-simultaneously reported on Twitter. There came a time when anybody who kept a regular eye on their Twitter account and who had applied the right criteria in choosing which accounts to follow, had the impression of a permanent déjà vu when they read the newspapers or watched the news on television. They may not have been given the whole picture, but they certainly had glimpsed part of it already on their Twitter account.

Now the showcase for what is going on in a world divided up into as many levels as users, the second phase of Twitter is seeing it become a service that its users value because it tells them what is going on, while it is going on. Twitter is now something that brings you all the news you are interested in, redefining the concept of news as something continuously developing and diffuse, that includes what a friend is doing, major world events, news about a specific topic or industry, and just about everything in between.

Which is where recommendation technologies come into the picture. The first such appearances on Twitter were for simple things like recommending accounts for users to follow. Initially, Twitter limited itself to proposing a short list of users that it recommended to all new users of the network. Later on, it began to segment them by country and language, and it was only much later, relatively speaking, that it began to introduce more complex algorithms, such as a collaborative filter based on the most-followed accounts by those that you were already following, etc.

Up to that point, this was limited to building up the network, based on the logic that Twitter is neither a good nor bad network, but simply as good or bad as the quality of the network of people that you choose to follow. The value perception that Twitter provides to a user depends almost exclusively on the group of accounts that the user decides to follow. After that, the emphasis was on trying to avoid a percentage increase in the number of users who, after opening an account on Twitter due in many cases to its constant mentions in the media, ended up dropping their account because they found no real value in it.

That said, the applications and possibilities of recommendation technologies have far greater potential, something that Twitter understands perfectly. Having bought recommendation applications such as We Are Hunted for music, Spindle for local content, Trendrr for television or Fluther for a variety of content, the company has now built a team of specialists able to pinpoint the next step in recommendation, something that should crystallize into a service able to put before your eyes anything that happens of importance anywhere in the world, and that knows that it could be of interest to you. This will likely lead to the media being used by the addition of other types of recommendations, either in the form of advertising or shopping referrals, something the company has been pondering for some time now.

Twitter yesterday unveiled its Magic Recs experimental account, which will supposedly send instant, personalized recommendations and content to followers through direct messages (DM). The experiment, which for the moment will be carried out with the collaboration of around 13,000 subscribers, will allow Twitter to gauge interest in the service, as well as testing the use of a channel, DM, that will offer a different degree of sensitivity: at some point in the not-too-distant future, we can imagine a user deciding to use a recommendation system to make sure that certain subjects or people are highlighted in their timeline, while others will arrive in their mailbox in the form of DMs, thus allowing them to establish different notification levels. Twitter would in effect have become a type of personal radar tuned into what is going on in the world: into a tool that lets you find out about whatever is important to you, with different levels of criticality.

Doubtless there are many other uses to which recommendation technologies could be put to. Aside from the interests a user lists when deciding which accounts to follow—interests chosen from a list or based on location—social questions can be introduced into the analysis, for example, how to avoid missing out on an event that is being discussed between the group of people you follow: a way of avoiding the dreaded FoMO, Fear of Missing Out. It is also possible to work with reactive functions that measure the reply level to tweets with certain names or subject matters, something similar to the semantic algorithms used by Klout to determine the topics upon each user has supposedly influence, as well as combining several of these elements using multi-variable algorithms or machine learning.

Obviously, a Twitter that is simply a network along which everything that happens in the world passes would be a very different animal to one that could organize and send to you specific subject matter about what is going on in the world. Such a service would surely end up being seen as essential for a sizable range of user profiles. Recommendation technology can be applied to a significant number of possibilities, and most are still very much in the early stages.

If one thing is clear about Twitter, it is that it has put together, through acquisitions and its own research and development, a team of experienced people to address these issues, and on which very probably will depend to a significant degree its ability to generate value in all senses.

Enrique Dans

On the effects of technology innovation on people, companies and society (writing in Spanish at since 2003)

    Enrique Dans

    Written by

    Professor of Innovation at IE Business School and blogger at

    Enrique Dans

    On the effects of technology innovation on people, companies and society (writing in Spanish at since 2003)

    Welcome to a place where words matter. On Medium, smart voices and original ideas take center stage - with no ads in sight. Watch
    Follow all the topics you care about, and we’ll deliver the best stories for you to your homepage and inbox. Explore
    Get unlimited access to the best stories on Medium — and support writers while you’re at it. Just $5/month. Upgrade