Content moderation is hard because it is adversarial - there is someone on the other side trying to evade it. Even in China, if you block one specific phrase, it's relatively easy to make a more oblique reference to whatever you're discussing.
Detecting every person in America who has any actual influence on policy or the media to make sure they never see this behavior would be doable, I think.
Check whether the phone ever goes near a university, or travel to the downtown area during work hours. Estimate their education level and income. Don't do it with new users. Default to NOT adding them to your influence operation. Start slowly, and then ramp it up as your confidence that it is a genuine user increases. How would this be detected?
It's generally acknowledged that TikTok has excellent ML / AI teams. People often marvel over how incredibly well TikTok can learn your interests just by watching a few videos - they are also gleaning a lot of other information in the process.
It's also generally acknowledged that the Chinese government is already going to great lengths to e.g. hack American companies to steal their intellectual property. We literally banned 5G base station hardware from Chinese companies due to these concerns. I'm not sure I see how you can think that Chinese base stations are such a threat (with hardware we can actually inspect, and run tests on) vs. software running directly on hundreds of millions of American phones sending data to Chinese servers, and displaying personalized content decided by algorithms that are literally outside the reach of the federal government.
I generally agree with Techdirt on tech, privacy, and free speech issues like this over almost any other commentator. However I'm not yet convinced on this topic due to some issues I haven't seen talked about much.
1. Propaganda is one thing, but personally targeted propaganda is another. I have seen journalists say "TikTok is just silly dance videos" and I'm sure for them it is. But we really don't know how things are being targeted due to the user-specific nature of the feed. I would not be surprised if there are "journalist", "government employee", and "college professor" filtering criteria that will keep you from ever experiencing these influence operations (if they exist now or in the future).
If your countrymen are being targeted by an adversary's propaganda in a newspaper, you can read it and respond to it. But if you don't even know what's being said, and to whom? What's the free society's mechanism for countering propaganda then?
We do not know, and fundamentally cannot know this because the only source of data about TikTok is TikTok. If TikTok were sold to a company in the US, or at least in reach of our subpoenas, I would have more confidence that their executives would act responsibly. The fact that CCP spokespeople are indicating they will shut TikTok down rather than allow it to be sold (at a cost of tens of billions of dollars) at least hints that there is something to hide already.
2. This is secondary, but I think there's a good argument for blocking it on trade grounds. Counterfactually, if China hadn't de-facto banned American tech and social media companies from their market, I suspect the Chinese tech industry would never have developed. That's probably trillions of dollars of market cap that would have instead of accrued to American tech firms and investors.
When framed as a simple fairness issue and I think most of your complaints seem less relevant. We don't consider other tariffs or trade restrictions to be first amendment issues.
Thoughts?
Techdirt has not posted any stories submitted by Matthew Bishop.
Content moderation is hard because it is adversarial - there is someone on the other side trying to evade it. Even in China, if you block one specific phrase, it's relatively easy to make a more oblique reference to whatever you're discussing. Detecting every person in America who has any actual influence on policy or the media to make sure they never see this behavior would be doable, I think. Check whether the phone ever goes near a university, or travel to the downtown area during work hours. Estimate their education level and income. Don't do it with new users. Default to NOT adding them to your influence operation. Start slowly, and then ramp it up as your confidence that it is a genuine user increases. How would this be detected? It's generally acknowledged that TikTok has excellent ML / AI teams. People often marvel over how incredibly well TikTok can learn your interests just by watching a few videos - they are also gleaning a lot of other information in the process. It's also generally acknowledged that the Chinese government is already going to great lengths to e.g. hack American companies to steal their intellectual property. We literally banned 5G base station hardware from Chinese companies due to these concerns. I'm not sure I see how you can think that Chinese base stations are such a threat (with hardware we can actually inspect, and run tests on) vs. software running directly on hundreds of millions of American phones sending data to Chinese servers, and displaying personalized content decided by algorithms that are literally outside the reach of the federal government.
Under-discussed parts of this debate
I generally agree with Techdirt on tech, privacy, and free speech issues like this over almost any other commentator. However I'm not yet convinced on this topic due to some issues I haven't seen talked about much. 1. Propaganda is one thing, but personally targeted propaganda is another. I have seen journalists say "TikTok is just silly dance videos" and I'm sure for them it is. But we really don't know how things are being targeted due to the user-specific nature of the feed. I would not be surprised if there are "journalist", "government employee", and "college professor" filtering criteria that will keep you from ever experiencing these influence operations (if they exist now or in the future). If your countrymen are being targeted by an adversary's propaganda in a newspaper, you can read it and respond to it. But if you don't even know what's being said, and to whom? What's the free society's mechanism for countering propaganda then? We do not know, and fundamentally cannot know this because the only source of data about TikTok is TikTok. If TikTok were sold to a company in the US, or at least in reach of our subpoenas, I would have more confidence that their executives would act responsibly. The fact that CCP spokespeople are indicating they will shut TikTok down rather than allow it to be sold (at a cost of tens of billions of dollars) at least hints that there is something to hide already. 2. This is secondary, but I think there's a good argument for blocking it on trade grounds. Counterfactually, if China hadn't de-facto banned American tech and social media companies from their market, I suspect the Chinese tech industry would never have developed. That's probably trillions of dollars of market cap that would have instead of accrued to American tech firms and investors. When framed as a simple fairness issue and I think most of your complaints seem less relevant. We don't consider other tariffs or trade restrictions to be first amendment issues. Thoughts?