Understanding how the algorithm achieves its accuracy matters even if you’re not interested in TikTok or the short video space because more and more, companies in all industries will be running up against a competitor whose advantage centers around a machine learning algorithm.

most experts in the field doubt that TikTok has made some hitherto unknown advance in machine learning recommendations algorithms.

recall that the effectiveness of a machine learning algorithm isn’t a function of the algorithm alone but of the algorithm after trained on some dataset.

the magic of the design of TikTok: it is a closed loop of feedback which inspires and enables the creation and viewing of videos on which its algorithm can be trained.

when considering how to design an app, you have to consider how best to help an algorithm “see.” To serve your users best, first serve the algorithm.

algorithm-friendly design

Everything you do from the moment the video begins playing is signal as to your sentiment towards that video.

Before the video is even sent down to your phone by the FYP algorithm, some human on TikTok’s operations team has already watched the video and added lots of relevant tags or labels.

Some of TikTok’s camera filters are designed to track human faces or hands or gestures so vision AI is often invoked even earlier, at the point of creation.

The default UI of our largest social networks today is the infinite vertically scrolling feed (I could have easily used a screenshot of Facebook above, for example). Instead of serving you one story at a time, these apps display multiple items on screen at once. As you scroll up and past many stories, the algorithm can’t “see” which story your eyes rest on.

judging sentiment is a challenge.

By relying on a long scrolling feed with mostly explicit positive feedback mechanisms, social networks like Facebook, Twitter, and Instagram have made a tradeoff in favor of lower friction scanning for users at the expense of a more accurate read on negative signal.

content derived from a social graph can drift away from a user’s true interests because of the mismatch between your own interests and those of people you know.

if the algorithm isn’t "seeing" signals of a user’s growing disinterest, if only positive engagement is visible, some amount of divergence is unavoidable.

Algorithm-friendly design need not be user-hostile. It simply takes a different approach as to how to best serve the user’s interests.

The goal of any design is not to minimize friction, it’s to help the user achieve some end. Reducing friction is often consistent with that end, but not always.

In this software era, true competitive advantages, or moats, are increasingly illusory. Most software features or UI designs can be copied easily by an incumbent or competitor overnight. All you will have done is test the impact of the design for them.

the actual magic is how every element of TikTok's design and processes connect with each other to create a dataset with which the algorithm trains itself into peak performance.

All that’s needed is an understanding of how the flywheel works and a commitment to keep every element and process in it functioning.