The internets largest mass migration event is here

As of today, January 14th, TikTok will be banned from the United States. In the week leading up to the ban, TikTok users are mouring the death of the platform, not just as a loss of a public sphere, but a loss of community. TikTok started to gain traction in 2019 and exploded in 2020, meaning the most avid users have spent almost 5 years on the platform, cultivating an audience and sharing a culture. Memes and trends that unified communities small and large now serve as a memoir of the platform that will soon be ripped from users hands. In many ways, TikTok and the landscape its algorithm shapes has molded into a society, a shared, yet differing, culture that is not all that different from America. A cluster of states, unified themselves through a shared sense of identity, further brought together by the platform that ultimately dictates the foundation and functioning of the states.

On January 19th, The United States government will forcefully dissolve this shared culture, creating internet refugees, searching for a new community to call home. Many have already embraced the identity of refugee on a new platform, Xiaohongshu or Rednote (or more directly, Little Red Book), where Chinese netizens are warmly welcoming the “TikTok refugees” with intro lessons of Mandarin and a rapid move towards English across the platform. Others have turned to Lemon8, ByteDance’s Pinterest clone, which ByteDance has been pushing through influencer campaigns and making it easier for users to transfer who they follow and their followers, not unlike Meta on Threads’ release. These platforms will likely see only a fraction of the 170 million TikTok users migrate to their platforms, each waxing and waning in popularity as people acclimate towards a platform’s proclivities. Nonetheless, there are 170 million users and their respective communities that will be in search of a new home.

Meta has proven to be an unsuitable alternative to TikTok. Meta has a long-standing history of election interference, tracking users across the web, and misinformation. They’ve recently added to the general disdain by riding the red wave, explicilty allowing hateful language towards basically anyone who is a minority and Mark Zuckerberg culturally aligning himself with the tech-right, alongside Elon Musk. On top of the more political objections, the Instagram and Facebook platforms are generally terrible. Facebook is almost exclusively older people and the content is all farming for engagement, culturally flattened beyond imagination. Instagram is Billboard: the Platform. Everyone is marketing something, even the people you know from high school. Either it’s companies selling their products, influencers selling their business or other businesses, or the people you know selling an image of their life, often a highlight reel, even in the age of the Gen Z photodump. It’s Linkedin for life, where you collect people you’ve met throughout your life to follow along in the highlight reel. The only thing that changes is what you’re posting about, your life or your career.

The algorithms that drive these platforms are also fundamentally different from TikTok’s. In 2022, ByteDance published a paper describing Monolith, believed to be the basis of the TikTok algorithm. Monolith was a real-time response algorithm, that was intensely responsive to the users current interactions with the platform. This means the videos served on your for you page were more heavily influenced by what you enjoyed in your current session that in previous sessions. All sessions would contribute to a less fidgety placement in the algorithm, but what you experienced here and now dictated what you saw. Instagram, on the other hand, appears to be a lot more sticky. There is no official documentation that I know of on how the Instagram algorithm works, but TikTok users generally have been dissatisfied with what they are able to find.

The algorithms shape the content, and the content shapes the platform. These suggestion algorithms will latch onto certain ideas of engagement depending on how they are trained and what they are trained to look at. Today, these algorithms are heavily gamed to find the features that extract the most watch time and over time, result in content shaped by the demands of the algorithm as everyone discovers these features. The high speed, more ephemeral nature of the TikTok algorithm lends itself to a sense of carelessness, that any video posted will be lost to the ether if not latched onto by the algorithm so the content is often more spontaneous, less developed, and lower stakes. In many ways, mindless experimentation. Like that one idea with infinite monkeys eventually writing Shakespeare, but the works of Shakespeare can be anything that makes enough people feel something. With the center focus of the app lying on the for you page, people are not worried about curating the perfect profiles but rather getting something out that resonates with people. One piece of advice passed onto aspiring TikTokers is volume, volume, volume. Making something every day, because eventually something will land with people and the more you do it, the better change you have at finding your people. Be yourself, they say, and the audience will find you. The ones who excel the most found their formula and stuck to it, often aligning with the wants of the algorithm. This experimentation is key to any culture, and has bred a particularly interesting culture on TikTok because of how fast and rampant the experimentation was. Ideas could spread like wildfire, inspiring microtrends that impact culture for hours, maybe days, and then peter out as quickly as they started. These ideas would rally communities together and tear them apart, depending on the hour of the day and the direction of the wind, but it all influenced collective placement in algorithm-space. As much as these communities are exclusively virtual, they live near each other in virtual space. Rather than addresses outlining your house number and street name, they outline the features in the content that makes engage with the content in some way. It might know you watch cat videos to the end, and like videos with breaking news, and scroll past videos of teenagers dancing, and place you near others like you, feeding you all similar content. We have no idea how many features they look at, but it’s likely far more than you’re thinking. If LLMs are trained on billions of parameters, or features, these algorithms could have tens of millions to hundreds of billions of features, depending on how truly complex they are made. As you continue to use the platform, and engage with and disengage with certain communities, you are moved constantly throughout this space, aligning you closer and closer to a mapped out version of your most engaged self.

The communities then are wholly unique, created both by chance and on purpose. People moving across algorithm-space, chasing their next hit, like walking through an ever-changing mall, each attractive in their own way and each with shoppers interested in their offers. Eventually, the algorithm finds you some stores you keep coming back to, and it does the same for many others. Slowly a communities around that store forms, but no one shops at just one place and is always getting exposed to something new. Communities rally around creators, and creators are placed into niches, and niches collect community as more creators and viewers are funneled towards their niches. It’s a living breathing organism dictated by an algorithm