TikTok & the New Internet
A curious analysis as to why TikTok became pop culture's one of the most important sensation.
Number 5, 1948 by Jackson Pollock
I want to make a confession first. I have always been skeptical of TikTok and all that it stands for. When I see my niece spending a good amount of her time on TikTok it makes me cringe. Naturally, I breathed a sigh of relief when it was banned in India. But over time, I’ve come to appreciate what TikTok or probably what the technology behind it stands for. Also, when I tried it out of curiosity, I realized that my behavior belied my thinking towards it. I always ended up spending more time than I had planned. For some of you the same might hold true as well. Even though I always had this curiosity about TikTok in the back of my mind, I never dug deeper and always questioned and ridiculed the futility of the whole thing. This essay is one way to get rid of my biases and understand as well as explain the core reasons behind the company’s success.
Carving a niche
An analogy for TikTok could be something like - “YouTube on steroids”. Well, in some aspects. YouTube first popularized the concept of creators who could actually create high quality content that could go viral and help YouTube in return. With time, the creators on YouTube and other platforms came to be known as influencers, a more platform agnostic term for anyone who is able to use his online presence to exert influence. So when I say, TikTok is a pumped up version of YouTube, I mean to say, it has the two most viral aspects of YouTube; short videos and users in the age group of 15-24. The mix of these two, certainly provides a lot of benefits; some of it must have been obvious to the team from the start and not just in hindsight.
As a medium in itself, short videos are an excellent middle ground between photos and the longer video format. Short videos are probably the most optimal path for teenagers to achieve social status in society and that too on a large scale. As explained in this excellent piece on the importance of status -
Thirst for status is potential energy. It is the lifeblood of a Status as a Service business. To succeed at carving out unique space in the market, social networks offer their own unique form of status token, earned through some distinctive proof of work.
It goes on to explain the differences between adults and young people in terms of building their social capital.
Young people look at so many of the status games of older folks—what brand of car is parked in your garage, what neighborhood can you afford to live in, how many levels below CEO are you in your org—and then look at apps like Vine and Musical.ly, and they choose the only real viable and thus optimal path before them. Remember the second tenet: people maximize their social capital the most efficient way possible. Both the young and old pursue optimal strategies.
Short videos are probably the most optimal strategy for teens to build their social status. Instagram has more than 60% of its users in the age group 25-54 and YouTube besides the fact that its uniformly distributed between different age groups, lends itself better to more professionalized content. The weirdness in the life of teenagers can probably be best expressed through short videos. To compensate for my lack of vocabulary in how TikTok is able to do just that, here is a quote from this article -
TikTok is home to groups of contortionists, furries, cosplayers, fitness gurus, new parents, nurses, cops, Marines, adult babies, goth kids, and more.
There are innumerable subcultures on TikTok which also means that there is always scope for something new and accommodate the multitude of interests that teens have.
Short videos as a medium is also probably a good application of machine learning. The sheer scale of the number of short videos that are created and watched on TikTok means that algorithms always have something to learn or in in technical terms, there is never dearth of training data. The emphasis on a continuous stream of videos increases the average watch time and the incentive to produce more videos. This is unlike YouTube which requires you to make a choice or Instagram/Twitter which has a definite end to the feed. Also, the fact that its designed in such a way which allows people to skip and forget a bad recommendation, the algorithm gets more chances to learn and improve itself.
Not your usual social network
Short video format as discussed in the previous section is probably an excellent use case for applying machine learning models effectively. The fact that TikTok has taken an AI first principle with curation of videos is a key differentiation from other platforms like YouTube, Instagram or Twitter which rely on a mix of AI and user’s social network (channel subscriptions in case of YouTube) to recommend content. The primary problem with this approach is that it puts too much responsibility on the user and is not a perfect proxy for what user wants to watch or consume. My Twitter feed, I would say, has at any point 50% noise and 50% useful content even though I have tried my best to manage and organize it. As explained in this excellent piece by Eugene Wei on TikTok -
The idea of using a social graph to build out an interest-based network has always been a sort of approximation, a hack. You follow some people in an app, and it serves you some subset of the content from those people under the assumption that you’ll find much of what they post of interest to you.
The one-way follow graph structure is well-suited to interest graph construction, but the problem is that you’re rarely interested in everything from any single person you follow. You may enjoy Gruber’s thoughts on Apple but not his Yankees tweets.
TikTok takes this responsibility of building a feed away from the user and into the hands of an algorithm. It keeps the fixed assumptions to a minimum in terms of user’s likes or dislikes. It also takes away the “winner takes all mentality” that is so prevalent in social networks. A Twitter user for instance who has a million followers, would definitely see his tweet getting more likes/comments than an average person. TikTok removes this hurdle or negative incentive. Quoting from the same piece referenced above -
In essence, some networks reward those who gain a lot of followers early on with so much added exposure that they continue to gain more followers than other users, regardless of whether they've earned it through the quality of their posts. One hypothesis on why social networks tend to lose heat at scale is that this type of old money can't be cleared out, and new money loses the incentive to play the game.
Old money used as a metaphor to refer to incumbent influencers is a real concern and probably the number one reason that a social network fails to attract new users or loses its steam. The fact that TikTok has been able to figure out a solution for this problem, is in itself a big value add over other networks.
The Dawn of AI
I previously wanted to name this article - “TikTok and the Dawn of AI”, but I realized as I was writing it, that to make such a strong conjecture, I should probably gain more technical expertise and knowledge into how AI works (though I did take up a class in my post grad under an excellent professor to whom I partly give credit to for this piece).
With TikTok, we finally see evidence of AI making decisions on behalf of users and knowing humans better than humans themselves. There are two important elements of this. First, the approach of TikTok’s algorithms is to religiously learn about its users. As this analysis in NY times writes about it.
TikTok favors whatever will hold people’s eyeballs, and it provides the incentives and the tools for people to copy that content with ease. The platform then adjusts its predilections based on the closed loop of data that it has created. This pattern seems relatively trivial when the underlying material concerns shaving cream and Crocs, but it could determine much of our cultural future. The algorithm gives us whatever pleases us, and we, in turn, give the algorithm whatever pleases it. As the circle tightens, we become less and less able to separate algorithmic interests from our own.
The more the algorithm learns about us, the better it becomes at suggesting us content to our liking. Second, what we are seeing here is probably a sneak peek into how AI algorithms gain consciousness. The thought itself is a little scary and still in its infancy. With TikTok, although every clip is different and weird, what if the algorithm is providing you content that is not just based on your likes/dislikes but is more opinionated which keeps things interesting. Can an AI do this ? This is where I see a thin line between what it means to be a human vs a machine. Pardon me for the cliched statement, but from time to time, I see this point taking more importance than ever before. I have appreciated the movie Ex Machina many times before, but a particular scene in the movie is similar to our case in point. Quoting this particular dialogue -
He (Jackson Pollock) let his mind go blank, and his hand go where it wanted. Not deliberate, not random. Some place in between. They called it automatic art.
The challenge is not to act automatically. It's to find an action that is not automatic. From painting, to breathing, to talking.
We digressed a little bit, but the essence with respect to TikTok is that maybe the algorithms knows us better than ourselves and at the same time provide a sense of serendipity that makes the platform all the more interesting and the reason we are not able to leave the app once opened.
The answer to my original question as to why TikTok worked is probably an obvious one. It worked because it had great technology, an amazing marketing team and the necessary product market fit. The fact that Vine or other short video platforms didn’t work is unfortunate in hindsight. But, I also think that what worked is this new way of interacting with the internet which was maybe missing and is a hint of things to come, along the lines of crypto currency. These technologies complement to the potential of Internet in a way that has never happened before and have the ability to make sense out of the mind boggling amount of information exchange and communication happening everyday and that too at scale.
For now, I rest my case and take comfort in the fact that I now understand TikTok a little better even though I would probably never use it.
If you enjoyed this article, please consider subscribing to the newsletter as well. You can also read about me and the blog here or follow me on twitter.