Why a TikTok Divestiture Never Happened
As this gets published we are about 18 hours away from TikTok going dark. From the standpoint of global Internet and data governance, the impending ban raises fascinating questions. If we look closely at how the US sought to ban a platform, and who pushed for it, and whether a change in ownership mitigates the alleged risks posed by TikTok, we find more questions than answers.
What it Means to Ban a Platform
The Protecting Americans from Foreign Adversary Controlled Applications Act (PAFACAA) defines the goal as a change in ownership of the U.S. part of TikTok (whatever that is, see below). Exclusion of TikTok from the U.S. market is the penalty for not effecting a divestiture. The law authorizes the following penalties to incentivize the objective:
TikTok will be removed from app stores, rendering new downloads and updates impossible for most Americans. The app stores would pay $5,000 multiplied by the number of users within U.S. borders who have accessed, maintained, or updated the app after the ban goes into effect.
It will be Illegal for American businesses to host services that support TikTok in the U.S. Hosting services would pay $500 times the number of US users “affected by such violation.” The meaning of “affected by” is not clear. If it means all of the TikTok users, then the fine for continued hosting would be $8.5 trillion).
These fines are applied not to TikTok itself, but to the app stores and hosting services who support TikTok. The law does not require ISPs to block anything.
It’s a clumsy, excessive approach. Aside from the app stores of Google and Apple, and Oracle’s data hosting, no one knows exactly who else is implicated in these penalties (e.g., domain name registrars?). We don’t know whether TikTok can continue to function by moving all its support services out of the U.S. These uncertainties arise, however, because Congress never intended to enforce a ban. They believed that the law would scare everyone into a sale (just as Trump tried to broker a deal back in 2020).
The hope for a divestiture explains why the Biden administration told ABC News that it won’t enforce the law one day before the president leaves office, and why several Congressional Democrats who voted for the law suddenly spoke out for “saving TikTok.” No one wants to claim responsibility for making the app go dark. But TikTok was not coerced into a sale, and “divesting” the U.S. part of TikTok is not really feasible, for reasons I explain below.
The Politics of PAFACAA
There are two ways to interpret PAFACAA and its attempt at a forced divestiture.
First, if forcing a sale was the real objective of the law, one can see it as a reflection of the Congress’s and Supreme Court’s ignorance of the digital ecosystem. Whoever devised this law believed that “the American part” of ByteDance can be sold without giving up The Whole Thing. The law assumed that digital platforms are territorial entities that can easily segregate their users and uses into specific jurisdictions. If the ownership is domestic, voila, we are secure, if it’s foreign we are not. If one has a national, territorial mindset and is ignorant of the operations of an AI-driven, globally extended digital platform, it would appear to be easy to broker some kind of deal with some US-favored ICT business. So PAFACAA was designed to scare Chinese Tiktok away, paving the way for a McCourt or a Musk or an Ellison to swoop in and save us.
Ignorance does explain much of the law but there is a second, more realistic interpretation. The Justice Department and intelligence agencies – what I call the ban faction – really did want to shut TikTok down, but couldn’t get it through Congress and the courts without holding out the false option of divestiture. Divestiture was the law’s best shield against a First Amendment challenge. The September 2020 court ruling against Trump’s Executive Orders trying to ban TikTok made it clear to the ban faction that free speech concerns would be in play, so while plotting their comeback they cloaked the law in ownership not speech. The ban faction knew that TikTok would not sell, but they needed that fig leaf to convince the Courts and Congress that they were not trying to suppress free expression. The ignorance explanation still has a place: the prospect of divestiture made it possible to sell the law to gullible Congressmen like Chuck Schumer, who voted for it but suddenly decided he wants to save TikTok now that the prospect of a shutdown looms. The divestiture strategy also disabled or weakened the Supreme Court Justices’ First Amendment antennae. The SCOTUS opinion used “ownership change,” coupled with “control of data” (a “content neutral” thing, you see) as a way of averting their eyes from a clear strike against the freedom of the media. This interpretation can also take on a darker and more conspiratorial flavor. If a sale actually happened, somehow, it would expropriate ByteDance’s intellectual property – its algorithm – which cannot be “split.” Either a ban or a an expropriation would shut off the possibility that Chinese access to American-generated data would advance China’s progress in artificial intelligence applications. The story about TikTok user data being a tool of espionage threatening Americans was a cover story floated by the FBI; it’s really just another attempt to cripple or retard China’s digital capabilities, like the semiconductor export controls.
Why Divestiture Can’t Happen
Let’s talk in more detail why TikTok in the U.S. cannot be “divested.” Can a globalized social media platform be split into two pieces, one owned by an American and the other by the incumbent (which is in fact multinational, not just Chinese)? No. TikTok USA is not a detachable Lego stuck on the side of a Beijing company. It is a global service built upon the export of ByteDance’s algorithm. TikTok US is just a legal entity; operationally, TikTok is as globalized as Facebook or Google. It is not easy to pull apart the dense flows of content and data that connect the phones of 1 billion users in dozens of countries who are constantly sharing videos, making comments, watching each other in real time, especially when the recommendation algorithm governing the whole thing is centralized. While it is true that globalized American platforms can apply different policies to different users based on the applicable law, they can only do that because their management and ownership are integrated. No one is asking Google to give up ownership of its European operations in response to the Digital Services Act, they are just asking it to comply with territorial law. The same could have been asked of TikTok. Ergo, the purpose of the law was not to protect Americans’ data, it was to kill TikTok.
The IP protecting the algorithm and its operationalization into running code is not something that can be split into two. Either you have it, or you don’t. Either it is plugged into and governing recommendations for your entire user base, or it is not. If it is plugged into only part of it, the US part, it gradually becomes a different algorithm producing different results than the Global part. Hence, TikTok cannot “sell” its algorithm to an American firm without giving it up its core intellectual property to a competitor, a firm that could operate globally in competition with it. And if it isn’t selling the algorithm, it isn’t selling TikTok; it’s selling something else. No one is quite sure what it would be. Perhaps just a list of users that would have to re-subscribe to a new platform?
If the American-owned, “divested” version TikTok is to continue to function as TikTok, its users would have to be able to share videos with the existing TikTok global user base, and be able to follow and comment on their videos, otherwise it wouldn’t be TikTok, it would be something else. But if American TikTok and the global TikTok interconnect, the latter would still be owned by the same “untrusted” firms and shareholders and, oops, ByteDance is still getting US users data. All the things that supposedly make ByteDance’s subsidiary in the US a national security threat today are true of ByteDance outside of the US, and true of American TikTok if its operations are sharing data with old TikTok. So the national security rationale, to be consistently carried out, would require nota divestiture, but ongoing blocking of Internet access to TikTok Global as well as forcible separation of the two TikToks. The split entity shares all the “data espionage” threats of its predecessor. The law has not reduced any threat. The only thing that stops the “threat” of Americans generating data that might be accessed by the PRC is to shut out ALL Chinese applications from access to US users permanently, including e-commerce and games.
The U.S. intelligence agencies – what I call the ban faction – had to know this. During the CFIUS review, the interagency review processes run by the Biden and Trump administrations, Project Texas was never accepted because localizing the storage of U.S. user data does not alter the fact that a machine learning algorithm engineered in China needs access to the platform’s data stream, regardless of who generated it or where it is geographically. If Project Texas was not an acceptable response to the “risk” of Chinese access to the data, neither is a divestiture. Only blocking Americans’ access to all Chinese apps, including the post-divestiture “Chinese” part of TikTok global, would mitigate that risk. Apparently, no one on the Supreme court has the technical chops needed to understand that if Project Texas not a solution to the security agencies’ worries, neither is divestiture.
Even if divestiture were possible, the legislators had no realistic idea of how long such an unbundling process needs to take. It took 4 years (1981-1984) for AT&T to break itself up into RBOCs and a separate long distance company. Additional years of planning and negotiation preceded the actual process. Ten years of court-mediated supervision followed. TikTok was told that it had to break up an equally large scale, perhaps even more technically complex (but certainly less hardware-burdened) system in 6 months. And the timer started before anyone knew whether the law was constitutional and would ever go into effect. Now that we know the law is constitutional (and damn the Supreme Court for not taking advantage of the opportunity to slow things down by granting TikTok’s motion for an injunction) TikTok got 48 hours to figure out what to do.
Maximum Uncertainty
It is now evident that even the Biden administration doesn’t know what to do. Faced with a growing tide of public opposition to the ban, it said it wouldn’t enforce the law, publicly. But can the intermediaries and suppliers stake a trillion dollars on that statement? Trump has promised he has a solution, but he is not saying what it is and it’s likely his incoming administration will be too busy chasing immigrants out of Chicago to do anything quickly.
If I ran a hosting service or an app store, and the government told me it wouldn’t enforce the law, and I wanted to stand up for free markets and free expression, I would definitely ignore the law and keep TikTok going. If the government went after me for that, I’d fight them, but that’s just me. Not many Internet businesses are that principled. Big corporations like Google and Apple probably have the legal weapons and money needed to mount a strong challenge, but with millions of dollars at stake who would take the risk?
And that is why 18 hours before the deadline when all those intermediary activities become illegal, no one knows what will happen.
The post Why a TikTok Divestiture Never Happened appeared first on Internet Governance Project.
Source: Internet Governance Forum