Twitter and the algorithm nobody wanted

Twitter Support says it has made an update that allows users to choose between viewing their timeline in algorithmic mode, with the supposedly most relevant updates first, or using the chronological approach Twitter used before 2016.

In early 2016, the company decided to increase use by developing an algorithm that would show users updates that could have been lost but that were likely to interest them. In doing so, they broke what had been a fundamental principle of Twitter until now: the timeline ran in chronological order, with the most recent updates at the top. After many versions and attempts to improve the algorithm in terms of relevance, the company now seems to have accepted that the timeline is central to Twitter: it’s what users expect, and interfering with it probably wasn’t a good idea.

To return to the good old days, users simply have to go to account settings, enter their content preferences and disable the show best tweets first option to retrieve updates in chronological order, without algorithmic interference along the lines of “In case you missed it” or “Such and such liked this tweet.” I don’t use the relevance option anyway and have always stuck with chronological order; that said, despite my ruthless rejection of suggestions to use the relevance option, the company keeps bombarding me with them.

Although Twitter says it has spent time working on ways to allow users to control their timelines, the reason for this change may also have been prompted by a thread created by a user speculating on the possibility that by excluding certain words she was able to create a purely chronological timeline. The thread went viral, obtaining more than 15,000 retweets and 40,000 likes. This presumably led the company to realize there was a large segment of users who preferred a timeline to an algorithm that clearly hadn’t been fully developed, which is why they brought the change forward.

Twitter’s obsession with controlling what users see or don’t see ran into a very simple reality: the timeline is perfectly clear and easy to use, while adding an algorithm to it added an annoying feeling of lack of control. It doesn’t matter that the algorithm was designed to prevent us from missing things that might interest us, to make us feel better informed or improve the product’s value proposition: users want their timeline in chronological order because that’s what initially attracted them to Twitter: they didn’t want a third party making decisions for them. The algorithm worked hard, came up with some really interesting personalization features and did a good job guessing user preferences… but quite simply, no one wanted it: the average user preferred the simplicity of the timeline.

Twitter will continue to offer features such as the search page where trending topics or moments can be immediately visualized, giving an idea of ​​what is happening regardless of which accounts we follow, but now we’ll have to select it. Now we can return to the purely chronological mode, leave the algorithm aside and recover the feeling of control over our timeline without any interference.

This is a good example of a company listening to its users, understanding their preferences and allowing them to enjoy the product the way they want, without being told how to. However smart they might be, algorithms are not always right or good at solving problems. In fact, some create them. As the saying goes, if it ain’t broke, why fix it?

(En español, aquí)