If you ask me, one of the most curious happenings of recent months, as highlighted to me by a fellow Silicon Valley resident, is, as a friend said, just how quickly “banning TikTok” went from a “fringe movement” to “quite acceptable,” not just nationally but also here in the Bay Area, where so many tech companies and personalities are so openly hostile to government bureaucracy and regulation. The few conversations I had on this topic tended to go something like this: usually it would start with a genuine concern about election interference or some other national security risk to the US … but inevitably, it would somehow end with the conclusion that either “this was fair, because China banned US platforms,” or the more generous “when US companies are allowed to compete, then we can allow them back in.” In other words, a shifting of the argument from one of national security to one of defending commercial fairness and free market mechanisms (or something like it). Yet some others would pause and linger on the “algorithm,” and state that if it weren’t for lack of oversight and unknown consequences of opaque algorithms deployed at scale, then TikTok would only pose a fraction of the danger it does, or perhaps none at all. I found this curious, because these are very different arguments in turn. National security concerns aside, all seemed to miss what I thought was surely obviously the biggest headline of all, which is that a Chinese internet company — not even the biggest, not by a long shot — was dominating US app stores for such a long time and having such an outsized influence on the popular zeitgeist. Would a commercial quid pro quo (letting US apps back in) even help with that at this point? Chinese apps, whose virtues have thus far mostly been how much utility they manage to pack into one screen (think super-app!), actually winning over fickle Gen Z subculture kids and proper A-list celebrities both, with sleek designs and culturally relevant campaigns? How did that happen? Why’s that not the “big story”?
WHY BAN TIKTOK AND WECHAT?
Now that the Executive Orders are out, you can find the government’s reasons outlined in the actual document. But this list here includes not just the government’s concerns, but also those commonly raised by pundits.
1) Fear that Chinese government will access US user data and use it for adversarial efforts.
1a) Fear that the Chinese government will access the data of Chinese nationals traveling in the US use it for adversarial efforts.
2) Fear that Chinese government will (or have influence over) content in the form of censorship and / or disinformation, again for adversarial reasons.
2a) There is specific concern around TikTok as the most downloaded app.
3) Fear around AI algorithms, because of their lack of transparency.
4) The unfairness of unfettered US market access when competing products are not allowed in China.
Let’s look at each of these concerns in turn:
1) Fear of data access / breach / compromise. I do not have too much to say on this topic, because this is, without a doubt, a national security issue, and not my field of expertise. I do not know how clandestine government operations work, nor do I have any idea how they can be either buried or uncovered, and I don’t think anyone is going to tell me either, even if they knew. So color me a big fat zero on ability to assess this risk. What I do think is worthwhile to point out is that there is a debate amongst more professional experts than myself about whether or not the risk can be mitigated, or is even sufficiently large to warrant a total ban.
1a) I only added this point in because it was mentioned in the EO on WeChat but I am a little at a loss as to how to interpret this. Is the fear that Chinese international students or tourists capture something of … national security importance in their stay here? Is that mitigated by banning WeChat?
2) Fear of influence using censorship / disinformation. This is also a risk of national security, but one I feel much more comfortable weighing in on because of its potential to actually have a technical solution. That’s because we are talking about the distribution of the content, which would be public, in a sense, versus its access, which is ultimately only known to the company itself (or perhaps not even to itself, as the Snowden leaks alleged). Distribution of information, especially on public-facing platforms, is theoretically verifiable and auditable, right? If there could be clear standards on what constitutes disinformation and outright censorship (versus suppression), and they can be reduced to binary outcomes — yes there is, or no there isn’t — then it is possible that results could be verified and audited. However, therein lies the dilemma. Does the US have a clear and legal definition of what “censorship” and “disinformation” means? No, I don’t think so. And I’m not sure that clarity is coming any time soon. The rights of the platforms when it comes to moderating their own content is itself a very complex issue with potentially unexpected second-order consequences. There is good reason all of this has been and continues to be hotly debated domestically, even without the fear of foreign interference mixed in. Clearly, this problem is not unique to TikTok, and it also plagues US internet companies when operating outside the US. Still, in theory, it seems like it would be easier to discover (and therefore to regulate / audit for compliance) than point #1 because of its more public nature. Like the scandals earlier this year of TikTok censoring “ugly” users or suppressing black creators — these were discovered and the company confronted because such actions can be much more easily noticed and reported, even by a typical user with no special privileges. I don’t have a solution, only the observation that this is a separate and different problem from one of access for these reasons and that solving one, even if doable, does not in fact solve the other.
2a) Increased fear due to TikTok’s reach. In the case of TikTok, who is sitting at 80 million monthly active users, the consequences of a widespread censorship and / or disinformation campaign would be utterly disastrous. In an ideal world, any solution the company proposes would need to be effectively error-free. That is, disinformation cannot just be a mea culpa that is removed after it’s identified … it cannot happen in the first place. Of course, this is extremely difficult to do, as we have seen just last week with the covid misinformation video that went very viral before it was banned by all of the major social media platforms. (I was on YouTube later that evening and multiple copies of the video were still accessible, if you need any further demonstration of the tenacity of those intent on sowing falsehoods and how easily they can do so.)
Anyway, that threat is greatly minimized, however, when we look at WeChat, which has also been accused of censorship and disinformation. There is one huge difference. It’s pretty much only used by Chinese diaspora (myself included) and its content is primarily in Chinese. Even those accusing WeChat of disinformation are talking about content distributed by the platform to its Chinese users, not foreign. Its US appstore rank: around 1500, users here: less than 3mm after nearly a decade in existence. The product is also fundamentally different, having started as a private messaging app focused on enabling transactions. There is no way within the app whereby anyone can easily target more than 5,000 people, since that’s the max of any single user’s friend limit, so that would make for a pretty dismal disinformation campaign (anyone beyond that number can only receive private messages). WeChat’s current Official Account system, its main publishing platform, which is subscription / commerce based and for businesses only, requires a business license and is basically only in Chinese language, as are the rest of WeChat’s products. In fact, that feature, one of its most popular, is marketed to foreign businesses as a marketing channel to market to Chinese customers, and indeed, requires a Chinese national as representative. While it is not impossible for WeChat to increase its users here in the US, that just seems extremely unlikely because well … it’s tried to do so globally several years ago, and failing miserably, hasn’t bothered renewing its efforts. Even its more recent attempt to grow Tencent Pay abroad was explicitly to capture the transactions from Chinese tourists versus acquiring foreign users. As many have noted, why would anyone abroad who has no personal connection to China or business need use WeChat? The famed super-app functionality that Chinese users rely on is almost entirely unavailable to foreign users, and for a while now, anti-spam and anti-fraud measures have made it very onerous to register new accounts, as I found out when taking investors to China last October. Its lack of traction and lack of interest in the US market makes it pretty difficult for me to understand its risk to national security. But of course, that hasn’t stopped governments from banning it, as the Indian government just did. As I wrote in the last Extra Buzz, WeChat also had no traction in India and was ranked well below 1000, because as is the case here, only those with personal and business ties to China would think of installing it. WeChat’s scale makes it far, far less risky in this respect than TikTok.
3) Fear of algorithms. We are in the first years where most of humanity is subject to information dissemination by algorithm. Before, it was not as persistent, large-scale, nor did it cover quite so large a portion of the population. I highlight the “algorithm” worry because it has been thrown around in conversation without specificity. Is it for censorship and disinformation (as discussed above) or for other reasons that cannot be measured in boolean terms? Something more insidious, subtle, and not clearly in violation of any particular legal or moral code, but is nonetheless detrimental to the public, perhaps not immediately but over a period of time? This seems possible, even probable. It is why many are clamoring for algorithmic transparency. But what algorithmic transparency even means and how it can even be enacted is being debated right now. Just Google for … Google’s algorithms and the arguments both for and against varying degrees of transparency. It’s not quite so simple as “show me the source code” either, since despite being their creators, humans cannot always explain how their own algorithms work, especially when it comes to deep learning. That doesn’t mean companies like TikTok shouldn’t try, however, even if it’s a lot of math, and veers into “black box” territory. However, at least at present, we might need to finetune our explanation of what a full explanation looks like. For example, is this blog post from TikTok on its algorithm satisfactory? The company claims its algorithmic source code is available at the Transparency Center — is that sufficient? (Deep learning “source code” may not be very easy to understand or explain at all.) The lack of a clear benchmark for “transparency” or even a shared understanding of the responsibilities of the algorithm-maker does mean that in the near term, I don’t see how TikTok will appease its critics.
As much as that is a worrying trait of TikTok, conversely, algorithm-phobia should mostly absolve WeChat on this count, given how very little of the product is algorithm-powered. (Everything within WeChat defaults to “friend recommended,” even the newest video Channels feature.) It is, after all, one of Allen Zhang’s most stubborn beliefs despite ByteDance’s AI-enabled success, that it is better to be “people-powered” than AI-powered. He has held onto the social graph long after everyone else has abandoned it, to his detriment, many think.
4) Reciprocity. I read through many posts and comments on the proposed ban, in both Chinese and English, and can confidently say that to my great surprise, a significant amount (what feels like might be a majority) of the time, the simple commercial reciprocity of you-banned-me, so-I-will-ban-you was one of the most oft-cited reason for supporting or explaining the TikTok ban … on both sides of the ocean. Yes, for some reason, both Chinese and American citizens seemed to think that this was the main justification for the ban, despite the US government’s clear statements to the contrary. The difference was that some Americans would start off with the national security argument, some meandering through some or all of the risks I highlight above, but the commonality was that both camps tended to belabor this point the most, usually accompanied by a partial list of US services banned in China (web only): Facebook, Google, WhatsApp, Instagram, Snapchat … and so on. Furthermore, a surprising number of people would reach the same conclusion — that the lack of full, open competition has already hurt Chinese / would now hurt US consumers the most. As a fellow consumer, I agree. I certainly didn’t enjoy having to use a VPN to access Google when I was residing in China, since it was far superior to the competition for my needs. I do not think I would enjoy using VPN or whatever workaround ends up being necessary to send a message on WeChat. Does it take a willing user away from their preferred service provider without her control or input? Yes. Are there good reasons for this happening?
Well, yes, unfortunately. While Google was not initially “banned” per se in China, it ended up shuttering most of its operations* because of a hack from the Chinese government discovered in early 2010. (It was later reported to be from two academic institutions, including a vocational school, a fact that has never stopped being the brunt of jokes.) Rolled into this was ongoing tensions around censorship, although to be very clear, Google had operated a censored search engine in China since January 2006, under the stewardship of Kai-fu Lee. (The entire Google China effort is covered in “In the Plex,” which is well-known in Chinese tech circles and often quoted when it comes to this topic, as is Kaifu’s own answer as to why he left.) We also covered it briefly in Tech Buzz Episode 18 but the TLDR is that the decision was widely lamented by Chinese people at the time.
*Even today, Google still operates a web portal (that redirects to Baidu) at 256.com (an odd remnant of internet 1.0 if you ask me), which it bought from a Chinese entrepreneur in 2003. It also continues to sell to Chinese businesses for its services abroad, obviously, and has employees there, so it does and can have legal entities in China.
But let’s not belabor this point, because regardless of whether or not you believe this is really an equal “trade” / comparison (many Chinese do not, including Kaifu himself, allegedly), I think fighting over that has completely and utterly missed the point. Frankly, I’m only including the details as a refresher and also to assure you, the full details of the tale is well-known in the Chinese tech ecosystem. Here is the point: Chinese companies were protected from competing with Silicon Valley in the past, but in the intervening decade, they’ve been competing with each other and improving quickly. TikTok is one example — but probably not the only one to come — of a Chinese product that is globally commercially competitive.
I LEAVE YOU WITH THIS
If you are not familiar with the Google China story that I link above — I urge you to read it. It’s fascinating. Google made every mistake in the book when it comes to management no-no’s and was still able to achieve a good amount of market share (36% at the very end). They didn’t hire a local, not even a returnee with local ties (Kaifu is Taiwanese-American). The founders, Sergey and Larry, never visited the country after operations commenced. They refused to let local engineers work on anything related to search. Obviously, that led to an exodus of talent (culminating in Kaifu’s departure) and made sure that the product wouldn’t reach its full potential. But this was ten years ago, and Silicon Valley technology was still strong enough and the Chinese competition weak enough that through sheer algorithmic strength and design and to some degree brand recognition, Google could still make significant inroads. I do not believe this is still the case.
Well, maybe in traditional search it is, given Baidu’s stagnation, which you can consider to be a +1 for “less walls and more competition.” But technology-wise on other fronts of consumer internet? China has definitely caught up, if only because they have to due to demanding consumers, intense competition, and immense scale. (Recall our episode on how many financial transactions a minute that one of the top red packet apps needs to be able to handle during Chinese New Year’s.) To treat the Chinese market as Google did 14 years ago? You’d get slaughtered. It wouldn’t be a competition. You’d have to have something truly superior — in real technology, in brand recognition, in product experience. And you’re certainly not going to do that by not letting your local talent work on your core product. (Uber, it must be said, empowered its employees and is actually seen as a pretty “successful” company in China for its efforts although you wouldn’t know that from reading English coverage after their pullout.)
So where exactly are we today?
Well, I’ll make some really broad statements, which won’t apply to every single instance, obviously. Let’s just face it, consumer internet isn’t “that hard.” There is no difference really between the US and China in this respect amongst many of the players, except maybe at the very top of the heap. Fundamentally, we are talking about the application of technology, scaling it, refining, operating and commercializing it, which are not dependent, for the most part, on some novel discovery. As many have noted, even though TikTok seems to have a damn good recommendation algorithm, the logic behind how to create it is widely known. The way it has been optimized and commercialized is the result of years of relentless hard work accumulated by thousands of engineers, designers, product managers, operations … it’s not some sort of black magic. But it also means that in all of these areas too, aside from pure engineering, ByteDance is also becoming globally competitive. And what’s more, to support all this, there’s an emerging new system of management that is enabling this kind of success. Finally, they’re not making the mistakes Google made … they’re trying completely different things from previous Chinese companies — hiring a truly global talent force, and even a foreigner (Mayer) as global COO. (I wrote about why I think that effort is in complete earnest, not for window dressing, in a previous Extra Buzz issue.) It’s not just a one-go “insight” — oooh, algorithmic distribution of video! — at this point anymore, it’s actually an “entire company” upgrade that has happened, because it’s been forced to happen from the competition in the markets. I’ll leave you with this Twitter thread of mine, which gives you the faintest outline of what ByteDance had to get right in order for TikTok to succeed.
The article was originally published on TechBuzz China.