The Chinese government invested a lot for decades in R&D around population-scale behavioral manipulation, including running a lot of experiments on their own population. It was an impressive research effort; other countries invest in this too but the Chinese commitment to mastery of it was next level. Not an issue.

These capabilities and techniques can make populations wired into it dance like predictable puppets in aggregate but they don’t work that effectively over generic undifferentiated communication channels because humans are too chaotic. It requires tight real-time feedback, control, and instrumentation of the information channels with sufficient critical mass population-wise to matter. Those kinds of tight feedback and control loops under direct control of government systems for constructive manipulation aren’t really a thing at most social media companies. You can spam propaganda but that is qualitatively inferior.

Divestiture of TikTok removes the access and control the Chinese government needs to effect outcomes with TikTok beyond typical propaganda and influence operations.

Most countries desire this capability but the technical implementation and requirement of sufficiently tight control of the channel has been a formidable barrier. China outright banned any vehicle that had the potential to allow foreign governments to do the same in their own country.

All of this has been known and discussed in national security settings for decades. The difficulty of implementation in the real world made it mostly a hypothetical risk at any non-trivial scale until TikTok.

The most insidious aspect is that sophisticated operational analytics has made it such that the manipulation may seem completely unrelated to the desired population-scale effect, it is not propaganda in a conventional sense. Done well, the individual never perceives it but the aggregate effect reliably emerges. The extent to which humans can be analytically manipulated in very indirect ways at scale is both fascinating and scary.

(Many years ago I used to work on problems related to population-scale operational behavioral analysis. China was on the cutting edge of this research even back then. None of the experimental theory is new, but apparently the tech finally caught up.)