TikTok
-
They haven’t released the details yet of the deal but somehow I suspect the outline of the deal is China still keeps data and/or algorithm and Trump and/or his buddies get paid.
-
I struggle to get too worked up about this, and Fat Matt's contention about Chinese controlling all the code without oversight, seems ill-substantiated by the details he shared.
ChatGPT's description of the current state of things:
Here’s a breakdown of the recent deal around TikTok, what led to it, and why it’s happening:
What the deal is (as of mid‐September 2025)
- The U.S. and China have reached a framework agreement that would allow TikTok to continue operating in the U.S. under new ownership arrangements. (Reuters)
- Under the proposed structure, U.S. investors (including a consortium led by Oracle, Silver Lake, and Andreessen Horowitz) would hold about 80% of TikTok’s U.S. operations. ByteDance (TikTok’s Chinese parent) would retain a minority stake (around 19.9%). (Reuters)
- All U.S. user data would be hosted on U.S. servers (reportedly via Oracle in Texas) under this plan. (Reuters)
- There’s a contentious issue about the algorithm: China has signaled the U.S. version of the app might still use ByteDance’s algorithm under a licensing agreement. That raises questions about how much control or influence ByteDance/China might retain. (Financial Times)
- The deadline for this restructuring/divestment has been extended (via executive order) to December 16, 2025 so that there’s time to hammer out the details. (AP News)
Why it happened
Several interrelated pressures and triggers led to this deal:
-
National security concerns
The U.S. government has for years expressed worry that ByteDance (which is based in China) could be compelled under Chinese law to share data with the Chinese government or otherwise that the Chinese government might influence the content or algorithm of TikTok. (PBS) -
Legislation: PAFACA
Congress passed the Protecting Americans from Foreign Adversary Controlled Applications Act (PAFACA) in April 2024. It requires TikTok to be sold or divested from Chinese control (or be banned) because it’s considered a “foreign adversary controlled application.” (Wikipedia)TikTok and ByteDance challenged its constitutionality; the Supreme Court ultimately upheld the law. That meant the legal pressure was real. (Wikipedia)
-
Deadlines looming for enforcement / ban
The law set a deadline (initially January 19, 2025) by which ByteDance needed to comply (i.e., divest) or face the app being banned. (PBS)Trump’s administration has repeatedly delayed enforcement via executive orders, while negotiations continued to try to get a deal in place. (PBS)
-
Domestic political and electoral considerations
TikTok is extremely popular in the U.S., especially among younger people. A ban would have substantial backlash. Trump has publicly credited TikTok with helping in his reelection, and losing access to it would be politically costly. (AP News) -
Global/geopolitical tensions with China
TikTok has become part of the broader U.S.-China relationship involving tech, data, influence operations, trade, and national security. Negotiating a sale or restructuring is part of managing those tensions. (PBS)
Key uncertainties and criticisms
- Whether licensing the algorithm from ByteDance still gives China too much control or influence. Some lawmakers are skeptical that this deal satisfies the law’s intent. (Reuters)
- Whether the structure will fully disconnect ByteDance from decision‐making, especially on content moderation, data security, or how the U.S. app works. (The Guardian)
- Whether the transitions (data hosting, infrastructure, algorithm) will happen cleanly, or whether there will be loopholes.
If you like, I can pull up the latest version of the deal (since things are still under negotiation) so you have the most up‐to‐date details.
-
ByteDance is writing the US version of the app now.
But the app isn’t the concerning bit so much as the algorithm. So yeah, Matt’s right. And I was too. Unfortunately.
ByteDance is writing the US version of the app now.
But the app isn’t the concerning bit so much as the algorithm. So yeah, Matt’s right. And I was too. Unfortunately.
I will always stipulate that you have a vague rhetorical framing in mind where you are right. But over here in a more objective reality, it is ill-supported to say that there will be no oversight of the American version of the code by the new owners. The other thing you're "right" about, something something Trump gets richer, here's a more benign description of that bidding process:
Yes — there was significant competition and multiple bidders interested in owning TikTok’s U.S. operations, not just Oracle / Andreessen Horowitz / Silver Lake. The deal that seems to be coalescing involves those firms, but there were others in the running or at least considering bids. Here are the details:
Other bidders / interested parties
According to multiple reports, the following entities either submitted proposals or expressed interest in acquiring TikTok’s U.S. assets:
- Amazon — submitted a last-minute offer for TikTok’s non-China assets. (Reuters)
- OnlyFans founder (Tim Stokely) in partnership with the Hedera cryptocurrency foundation. (Reuters)
- AppLovin — proposed a merger involving TikTok’s global assets (excluding China). (Reuters)
- Perplexity AI — submitted a plan to create a new entity combining Perplexity and TikTok U.S., with some proposals even involving U.S. government involvement post-IPO. (Reuters)
- Frank McCourt and Alexis Ohanian — under what has been called “The People’s Bid” initiative. (Reuters)
- Jesse Tinsley and MrBeast — made an all-cash $30 billion offer. (Reuters)
- Microsoft — had explored the idea earlier; but Microsoft’s efforts seem to have fallen through or become less central in recent rounds. (Bloomberg.com)
So yes — there was quite a lot of competition, from large tech companies, venture capital/PE groups, individual entrepreneurs, and other tech-oriented firms.
Why Oracle / a U.S. consortium got picked
Among all the bidders, a consortium led by Oracle with Andreessen Horowitz, Silver Lake, and some other investors (including existing U.S. shareholders in ByteDance like Susquehanna, General Atlantic, and KKR) appears to have become the lead-choice. Some reasons why:
- Existing relationship & infrastructure: Oracle is already hosting TikTok’s U.S. user data (Project Texas) and is involved in U.S. cloud infrastructure for TikTok. That gives Oracle credibility and infrastructure readiness to take on the data hosting / security responsibilities. (Reuters)
- Meeting legal / regulatory requirements: The U.S. law (PAFACA) requires that ByteDance’s Chinese ownership be reduced to under 20% for U.S. TikTok assets, and the new deal keeps ByteDance at about 19.9%. The consortium structure with U.S. majority ownership helps satisfy that requirement. (Reuters)
- National security / data safeguards: Because Oracle already handles some of the data, there is an established trust (or at least precedent) for U.S. operations to manage user data in a way more directly under U.S. jurisdiction. That helps with satisfying national security concerns. (Reuters)
- Political feasibility: The chosen group seems more likely to satisfy U.S. government evaluations, to negotiate with China, etc. Large tech/finance players with existing regulatory exposure are more predictable choices vs. some of the more dramatic bids (like those from individuals or newer companies).
If you like, I can pull up more details about the competing bids (e.g. what they proposed, how far they got) so you can compare them side-by-side. Do you want that?
-
We should have a thread dedicated to your JDS, so that when you want to turn the conversation to me instead of the topic under discussion, you can post it there so that people can choose to read that if they want or just stay with the topic at hand.
At any rate, I’ll repeat, the code isn’t all that important. The algorithm is.
-
We should have a thread dedicated to your JDS, so that when you want to turn the conversation to me instead of the topic under discussion, you can post it there so that people can choose to read that if they want or just stay with the topic at hand.
At any rate, I’ll repeat, the code isn’t all that important. The algorithm is.
We should have a thread dedicated to your JDS, so that when you want to turn the conversation to me instead of the topic under discussion, you can post it there so that people can choose to read that if they want or just stay with the topic at hand.
At any rate, I’ll repeat, the code isn’t all that important. The algorithm is.
Obviously my post was almost entirely about the discussion at hand.
I guess this is where I ask you what the difference is between algorithm and code, and then it's your turn to weasel out of the question (to which there is no good answer).
-
It’s true the algorithm is implemented in code, but what seems to be relevant here is that the consortium will be licensing the algorithm from ByteDance.
It’s true the algorithm is implemented in code, but what seems to be relevant here is that the consortium will be licensing the algorithm from ByteDance.
Licensing does not imply zero oversight, which is the point I originally made in response to Yglesias and you. That is his leap, it is ill-supported, and I suspect just wrong.
