Published on [Permalink]
Reading time: 4 minutes

Initial Thoughts on the TikTok Hysteria

Is TikTok anything more sinister than a benign waste of time?

Why the outrage over TikTok? It seems that those who are outraged aren’t TikTok users, and to the extent they are, they’re sympathetic to a forced removal of their biggest time sink.

Their appearance before Congress was riddled with near-incomprehensible questions and legislators thinking they were clever.

The CEO was asked such ridiculous questions as, “have you directed them to change the source code,” and similarly absurd accusations that TikTok is spying through our phones' cameras to capture our facial expressions.1

The real threat TikTok poses, if any, is hardly articulated.

Yes, it’s probably bad for a foreign government, especially the Chinese, to have access to the sorts of data TikTok does – preferences, demographic and other inferential data.

Why? A tap into the obscene number of collective hours we collectively spend on TikTok could be invaluable.

Could the Chinese government use TikTok to influence Americans' thoughts or behaviors? Maybe, but probably not in an obvious way.

Facebook, Google, and other adtech are necessarily good: they wouldn’t survive otherwise. Yes, there are the predictable ads that follow you around seemingly forever after visiting e.g., Allbirds and viewing a single pair of shoes.2 But for every one of those instances, there’s a matching one of incredible prediction and statistical maneuvering.

My favorite illustrative example is this: Target knew a teenage girl was pregnant before she did. And this is obviously not because she searched for e.g., diapers. The short version is this: people who are pregnant, whether they know it or not, tend to act slightly differently than other people. This, apparently, extends to their shopping behavior at Target. They used an algorithm that could parse purchase habits and determine, with apparently great accuracy, whether a customer was pregnant.

It doesn’t take much imagination to believe companies, with considerably more tech prowess than Target, use techniques similar to these to predict demographic, health, interest, etc. information. People who are, eventually, known to be members of a certain group or have a certain attribute might act in certain, slightly different ways on Instagram or Google or… you get the idea.

This is why it might seem like Facebook is listening in through your phone’s microphone. But they aren’t. If they were, they’d (1) have to keep it very quiet, and (2) exploit a horrible bug in iOS (or Android), since Apple certainly wouldn’t be on board – they cost Facebook millions because of their App Tracking Transparency popups, which allow users to opt out of cross-app tracking. Unless this is some sort of 4D chess. ;)

So the influence TikTok might exert would be more sublime: it’s much more powerful when someone draws the conclusion you want them to – it’s ‘their idea’ – than to overtly try and convince them. Maybe there are certain types of videos their recommendation system could prefer that would subliminally promote e.g., feelings of resentment of or distrust in government. The usefulness of such an effect is clear.

Whatever the case, our legislators are sorely ill-equipped to handle such a matter of technology. Even the Supreme Court admits they “really don’t know about these things [… we’re] not like the nine greatest experts on the internet.”

I might think differently after reading some more about this. Noah Smith wrote a piece arguing that “of course we should ban TikTok,” and Platformer just published a critique of the CEO’s performance before the House committee. Maybe I’ll revisit this topic.

1I got this from a video on TikTok. An interesting meta argument is that videos ridiculing Congress seem to be doing well on TikTok. But is that because of their manipulation? Maybe it is, but maybe it’s what the median TikTok user agrees with.

2I can’t stand ads and have removed them (as much as reasonably possible) from my life. But that’s a topic for another post.