New blog post, this the first of maybe a 3 part series on the company du jour, TikTok.

This first piece isn't about the possible ban or divestment or the security or geopolitical issues but about what makes it so unique and fascinating to me. https://www.eugenewei.com/blog/2020/8/3/tiktok-and-the-sorting-hat

TikTok is remarkable as the first Chinese app I can think of to make real inroads in the U.S. market.

I had long thought the cultural barrier too significant to overcome, but it turns out that machine learning algorithms can now pierce the veil of cultural ignorance.

For a self-described cultural determinist like me, it's a miracle that an app launched by two Chinese guys out of Shanghai, http://Musical.ly , could evolve into a leading short video app in not just the West but other foreign cultures like India and the Middle East.

http://Musical.ly  eventually hit their asymptote in the US—there are only so many teen girls who want to do lip synch videos—but Bytedance bought them and did two things to jump start their growth again.

First, they poured an insane volume of ad dollars into filling the top of their funnel. TikTok ads were everywhere you looked.

At first this seemed crazy. The 30 day retention was widely rumored to be sub 10% on this acquisition campaign.

Second, and more importantly, they swapped out the feed algorithm for one built by Bytedance, widely thought of as the algorithm company in China for their first breakout app, an algorithmically powered news app called Toutiao.

The ad money pulled a more diverse audience into TikTok, and the algorithm acted like the Sorting Hat from Harry Potter, matching content with the viewers who'd appreciate that content. This enabled new subcultures beyond teen girl lip-synching to form.

In the two-sided entertainment marketplace that is TikTok, its algorithm acts as the market maker. Because it works so well, TikTok could stand up an entertainment network without an explicit follow graph.

Most social networks use a social graph to approximate an interest graph. This is mostly fine, but at some scale this approximation diverges. I don't share the same interests as all my friends and family.

More importantly, at scale, social graphs come with all sorts of negative network effects, many of which Facebook, Twitter, Instagram, and other companies are grappling with now.

TikTok skipped that step and just used its algorithm to directly derive an interest graph. Its algorithm is so effective that it doesn't feel like work for viewers. Just by watching stuff and reacting, the app learns your tastes quickly. It feels like passive personalization.

This is massively disruptive, because the interest graph is more valuable than the social graph. Interests are what allow for effective monetization like targeted advertising or e-commerce or subscriptions.

An efficient algorithm for intuiting user interest can be used to compete in so many adjacent markets. E-commerce. Media/News. Audio (music/podcasts). Dating. Job hunting. Education.

Don't get me wrong, TikTok is a long long way from realizing any of that potential. However, Douyin, the http://Musical.ly  clone that Bytedance launched in China, has realized some of that potential.

IMO, if TikTok is sold off along with its algorithm, every tech company should be bidding hard for it. I've heard $30B tossed around as a figure, I'd happily pay twice that without a second thought. This is a once in a generation purchase opportunity.

Would Microsoft, or some other tech giant, realize this full potential? That's far from guaranteed. I wouldn't put it past any of them to screw the company up.

But I'd be terrified of a competitor owning them instead of me.

I'm biased, of course. I think video is still massively underrated as a platform for all sorts of things from commerce to education.

TikTok is the closest to video as platform as I've seen so far and it's only scratched the surface of that potential. <END>