When talking about the extraordinary success of TikTok, one magic word is always evoked: algorithm. According to the most popular interpretation, the merit of the success of the Bytedance social network – which has exceeded one billion users in a very short time and keeps them glued to their smartphones for more than double the time compared to Instagram (52 minutes a day against 28) – it would be above all of the video recommendation system, capable of identifying with astonishing precision the most suitable content for each individual user.
TikTok is also going down the drain (the enshitting phenomenon)
by Riccardo Luna
What if this were some kind of tech legend instead? What if the merit were not of an extremely advanced secret algorithm, but of design and planning choices that are there for all to see? This is the thesis of Arvind Narayanan, professor of Computer Science at Princeton, who explains in an analysis: “There is no reason to think that the TikTok algorithm is more advanced than its rivals. Based on everything we know – the description of TikTok itself, the searches it has performed, the confidential documents that have been leaked, and the reverse engineering – it is a standard recommendation system, the same type used by any other major platform of social media”.
What is corecore, the TikTok trend that you only understand if you’re twenty years old
by Emanuele Capone
But if this is the case, how did TikTok manage to create such a successful product that it even sent Meta into a crisis, which today is trying in every way to clone the dreaded competitor? Taken by themselves, the various elements highlighted by Narayanan seem far from surprising: if you put the pieces together, however, a puzzle emerges that can show what are the different factors that together made the rise of a platform possible capable of overthrowing giants, from Facebook to YouTube, who seemed immovable.
First of all, TikTok understood the crucial importance of introducing ultra-fast and effortless scrolling: “On YouTube, every time you select a video but then decide you don’t want to watch it, you have to go through the process of scrolling until it’s gone. find one of interest,” explains Narayanan. Inside TikTok, things are very different: moving quickly to the next video (ie “swiping”) is so immediate that it is performed almost without realizing it: a sort of television zapping to the nth degree. This is one of the reasons why, while TikTok’s and YouTube’s algorithms are almost certainly equally accurate, the former gives the impression of being much more accurate (yet, on average, only 5% of videos viewed on TikTok receive a like).
Almost half of Italians on the internet use TikTok (which doubles users in one year)
by Pier Luca Santoro
Since both YouTube (with the Shorts) and Instagram (with the Reels) have tried to clone TikTok, however, a spontaneous question arises: why haven’t they been as successful up to now? “The scrolling paradigm, the vertical format and short videos are three intimately connected elements: each ingredient works best in the presence of the other”, Narayanan always explains. For example, it makes no sense to ask the user to take the time to select, choosing from various options, a video that lasts only 15 seconds. It is for this reason that short videos work well with instant swipe, which in turn works much better when the phone is held vertically, consequently affecting the aspect ratio of the video (horizontal videos are more suitable for content than require prolonged viewing).
TikTok has united these three elements from the beginning, while YouTube’s attempt to clone them has seemed artificial and unnatural: often the Shorts are cropped vertically from originally horizontal videos, while the creators seem to produce these short contents only to bring users on their traditional channels, making the Shorts ineffective.
What about Instagram instead? The problem, in this case, is partially different. TikTok treats each single video independently, evaluating its viral potential without paying too much attention to the amount of followers possessed by the creator or the success that his previous videos have had. “Every video seems to get a guaranteed minimum of views,” Narayanan continues. “If the video performs well, then it gets submitted to more viewers.” Consequently, the amount of followers becomes a secondary element.
It is true that the algorithm performs this selection, but TikTok’s competitors would have no difficulty in making changes in this sense. So why doesn’t Instagram follow the same path (for now)? “The only thing preventing them from doing so is the fact that the best-known creators, who together determine the success of the platform, would rebel, because they would risk losing the fruit of years of work”. It is also for this reason, as Ben Thompson of Stratechery also confirmed, that Instagram influencers have protested against the attempted “tiktokisation” of the social network attempted by Meta a few months ago.
Finally there is perhaps the most interesting element of all. If social network algorithms have always given absolute priority to content that met users’ tastes (thereby providing a continuous flow of similar videos), TikTok instead behaves differently, favoring the so-called “passive exploration”, i.e. “risky recommendations that are not very likely to be successful, but which, when they do, give users great satisfaction”. Furthermore, it is precisely the ease with which users can switch from one video to another that allows TikTok to act in a way that, until recently, was considered taboo in the social world. “From the user’s perspective, the algorithm always seems to want to provide new topics, even after it has identified a range of topics in which he is interested.”
Also in this case, Instagram would theoretically have no difficulty modifying the algorithm and giving more weight to exploration: only the priority given to the relationship between followers and creators prevents it from doing so. “Reels are a major departure from all of that,” Narayanan explains. “So it’s no surprise that a report showed Reels has a very low participation rate.”
The secret of TikTok, therefore, is not the algorithm. If anything, Bytedance’s social network offered an experience different from that provided by other platforms: an experience that users evidently appreciated. If Instagram, YouTube and Facebook are now unable to successfully clone TikTok elements, it is because they were designed, from the outset, to offer a very different user experience.