(预告)Meta 对青少年的新限制、Nvidia 的开源投资、Netflix 问答、Druckenmiller、参数与性能
- Ben Thompson 将 Meta 与29个州达成的创纪录171亿美元和解——包括儿童每天在 Instagram 和 Facebook 上限时2小时、15/60/90分钟强制暂停、午夜至6点的禁用时段,以及上课期间不推送通知——视为典型的监管俘获,而非公司败北。 “大型公司本质上是政治动物……如果终点注定要到达,最好按自己的条件到达”:Meta 谈妥了结果,将赔付期限锁定为10年,并消除了长期压制股价的悬而未决因素。
- 据 Ben 判断,和解协议最精巧的一条是:只有 YouTube 和 TikTok 也加入,Meta 才需要付款——“太好笑了”——而在“昨天可能已提交约1,000起诉讼”的背景下,法院很可能把这份协议视为事实上的先例。 Meta 随后在《华盛顿邮报》、《洛杉矶时报》和《纽约时报》刊登整版广告,呼吁竞争对手采用同一套规则;类似9小时内90%响应率这样的重合规要求,天然偏向资源更充足的在位者。
- Ben 的原则性反对是:这份协议“很大程度上是在绕开第一修正案”——对进食障碍的讨论和美颜滤镜的限制“显然属于第一修正案的保护范围”,却在没有经过立法的情况下获得了法律效力;同时,它还固化了“社会认可地将养育责任转交给大型、去人格化组织”的转变。 Andrew 反驳称,在真正的立法中,儿童安全这一重大公共利益可能足以压过部分第一修正案顾虑;Ben 则认为国会不可能将这套框架写入法律。
- Andrew 对这些限制能否改变行为持观望态度:澳大利亚禁止16岁以下使用社交媒体后,调查中使用年龄限制平台的儿童比例只从禁令前的86%降至3个月后的81%以上。 Ben 补充称,许多危害研究受到“定义变化的严重混杂”,而且2012年前后,为某些状况开具处方的回报也在变化——“智能手机是2007年出现的,为什么是2012年?”——美国数据与世界其他地区“差异极大”。
- 在年龄验证问题上,Ben 最终倾向于由 Apple 和 Google 提供操作系统层面的 API——“既然要做,就把它做好,即使我保留反对这件事本身的权利”——而不是让每个应用各自拼凑身份核验方案。 他承认自由意志主义者的反对意见:这套 API 可能被滥用来确认用户身份。平台如今实际上“被要求越来越深入地探查你是谁”,这恰恰是大家过去10年一直在抱怨的监控模式。
- 更深层的框架是:这是一场保守主义与自由意志主义之争,而不是左派与右派之争;Ben 明确表示自己可以被说服,并发出“提前征求反馈”的邀请,以一种“小写 c”意义上的保守主义立场提出:“有时,灰色反而更好”(“Sometimes gray is good”),法律的作用部分在于表达社会不赞成,从而让父母更容易履行管教责任。
1. 创纪录的和解,以及 Ben 公开承认的矛盾判断
- 消息是:Meta 就社交媒体成瘾索赔与29个州达成171亿美元和解,内容包括儿童每天使用 Instagram 和 Facebook 不超过2小时、在15、60、90分钟时强制暂停、0点至6点的夜间禁用时段、学年期间上课日8点至15点不推送通知、年龄核验,以及限制美颜滤镜和公开显示点赞数。执行情况和实际效果将由独立审计机构及参与和解的各州评估。
- Ben 罕见地以“提前征求反馈”开场:“我对此有不少相互冲突的看法,而且确实有机会说服我改变立场。”他的默认观点是“父母首先得尽到自己的职责”,而这份和解意味着“社会认可地将养育责任转交给大型、去人格化组织”。
- Andrew 反驳称,设置护栏可以降低育儿难度。两人谈到学校使用能在学生入校时锁定手机的手机袋:这会让孩子无法借助 WhatsApp 或 Instagram Reels 退回原有朋友圈,迫使他们在新学校与更多同学互动。
- 第二个原则性反对是:“这样的和解在很大程度上是在绕开第一修正案。”对进食障碍讨论或美化滤镜的限制“相当明确地属于第一修正案保护范围”——这样的法律不可能通过,但与一众州总检察长达成的私下协议“基本会取得法律效力”。
- Andrew 表示,在真正的立法中,儿童安全可能构成足以压过部分第一修正案顾虑的重大公共利益,同时承认自己不知道具体会如何裁定。Ben 的回应是,国会不可能把这套框架写入法律。
2. 这事到底有没有用?澳大利亚与混杂不清的研究
- 在宣布这是分水岭之前,Andrew 主张先观望:澳大利亚禁止16岁以下使用社交媒体后,禁令前调查中有86%的儿童表示使用过年龄限制平台;3个月后,这一比例仍高于81%。
- Ben 对数据持怀疑态度:相关危害研究“受到同一时期定义变化的严重混杂”。他认为,这个问题总体上与 Obamacare 有关;与此同时,随着智能手机开始普及,相关定义以及为某些状况开具处方的回报都在变化。“智能手机是2007年出现的,为什么是2012年?”美国数据与世界其他地区“差异极大”,在他看来,这进一步支持了美国数据主要源于定义变化的解释。
- Ben 表示,这种不确定性意味着人们最终还是得把亲身体验纳入判断。Andrew 指出,人们会把自己使用这些工具的经历投射到整个社会;Ben 反问,有多少用户在刷了2小时 Reels 后抬起头,觉得问题确实存在。Andrew 补充称,如果 Twitter 每天使用1小时后就自动关闭,他的状态会更好。
3. 有时灰色更好:保守主义与自由意志主义的裂缝
- Ben 用法律“表达社会不赞成”的功能,来平衡自己的自由意志主义本能。他举例说,那些在大学里抽大麻、始终隐约感到内疚的人,会支持大麻合法化,却没想过“工业化大麻”会带来什么后果——“走在纽约街头都能闻到臭味”。他的结论是:“有时,灰色反而更好”(“Sometimes gray is good”)。停留在灰色地带没关系,没必要把一切都做成非黑即白。
- Ben 预计,更重要的分歧将来自 Scott Alexander 关于“灰部落/蓝部落”的文章:右派内部保守主义与自由意志主义之间存在“一道相当巨大的鸿沟”,如今正在浮出水面——“这不是一场左右之争”。
- 两位主持人都认为,育儿层面的回报在于默认选项很重要。儿童社交媒体存在“集体行动难题”,当法律将默认状态设为限制接触时,父母就可以把法律拿出来作为管教依据。
4. 年龄验证应由 Apple 和 Google 负责
- 对于另一个“无法决定”的问题,Ben 的结论是:如果年龄验证必须存在,就应由 Apple 和 Google 负责——两家公司拥有系统底层权限、信任度高;相比未来让每个应用各自东拼西凑年龄验证流程,操作系统层面只返回“是/否”的 API 更具规模化能力,也更可取。
- 他同时提出自由意志主义者的反对意见:谁来核验一个人的身份?这套 API 也可能被滥用来确认用户是谁。两人指出其中的讽刺之处:平台实际上“被要求越来越深入地探查你是谁,以免触法”——“而这正是我们抱怨了10年的事情”。Ben 最终的表态是:“既然要做,就把它做好,即使我保留反对这件事本身的权利。”
5. 商业分析师 Ben:监管俘获的漂亮执行
- 贯穿始终的判断是:Meta“谈妥了自己的结果”——一次性埋掉所有历史包袱,把赔付安排在10年内,消除股价悬顶因素,而且按 Ben 的说法,只有 YouTube 和 TikTok 也加入,Meta 才需要付款。“太好笑了。”类似9小时内90%响应率这样的合规负担,正是典型的监管俘获:“我们应付得来。”
- Ben 指出,Meta 没有承认过错。即便公司认为条款不公或衡量方式有误,大型公司依然是政治动物:Meta 正在抢在政治趋势之前行动,让问题就此收口,而且是在自己谈妥的条件下完成。
- 没有正式立法并不妨碍其产生影响。Ben 认为,这是又一次绕过失灵立法程序的做法。Andrew 解释称,法院遇到新问题时会参考其他法院的处理方式,因此这份和解会成为“相当有说服力的支点”;Ben 预计大多数法院都会对其予以遵循。
- 竞争压力也随之而来:Meta 致 YouTube 和 TikTok 的公开信,以整版广告形式刊登在《华盛顿邮报》、《洛杉矶时报》和《纽约时报》上;Sagar Ejati 注意到,广告甚至没有提到 Snap,这颇有意思地说明了该平台所处的位置。Ben 一开始猜 Google 会先行动,随后又改口猜 TikTok:“我不知道。”
- Sagar Ejati 发推追问,为什么这些规则不应适用于成年人。Ben 回应:“我们确实需要年龄限制,只是年龄限制的方向完全反了……孩子没问题。”Andrew 则提名超过65岁的人群,认为他们应首先接受严格限制。
完整逐字稿
Hello, and welcome to a free preview of Sharp Tech. Hello, and welcome back to another episode of Sharp Tech. I'm Andrew Sharp, and on the other line, Ben Thompson. Ben, how are you doing?
I like to take issue with the tech industry. Too much news on Wednesdays. We've talked about this. The worst part is that I'm always vaguely nervous about Sharp Tech.
Mm-hmm.
Because, of course, it's timely. We're recording on Thursday, so we can talk about stuff that I haven't written about. On the other hand, it just reiterates—apropos of the conversations we've been having—the value of writing is that I feel fully confident I've thought through all my takes instead of going a little bit off the cuff.
Right.
Maybe I won't do a lot of editing for this one.
We're going off the cuff today.
We're going to see how it goes, yeah.
It is pretty funny because it didn't feel like summer until the past 2 weeks. The news cycle really started to slow down. Did you experience that while writing Stratechery over the past couple of weeks?
Oh, absolutely. I took my vacation earlier than usual in July.
Right.
Usually I take the vacation in the middle of August.
And it was busy back then. In July, it was very busy.
I'm hurting right now, coming up with topics. I'm staying up later than usual to write, mostly because I'm dilly-dallying, trying to figure out what I'm actually motivated to write about here.
Squeeze some meaning out of all sorts of nonsense, yeah.
But, basically, yesterday it started to feel like, okay, summer is drawing to a close. Maybe it's because my kid went to school this week. Who knows what it is, but yes, we are approaching the end—
As of Wednesday—
Of summer.
Summer officially ended. So yes, we have a number of news hits to work through this week. I will begin with news that came down Wednesday morning, and I'll read a tweet from Variety:
“Meta will enforce a 2-hour daily limit on Instagram and Facebook for children as part of a record $17.1 billion settlement with 29 states over social media addiction claims. Behavioral remedies agreed upon as part of that settlement include productive pauses for children, including mandatory pauses after 15 minutes of continuous use, and again at 60 and 90 minutes to interrupt endless scrolling.
“Nighttime blocks restricting children's access from 12:00 AM to 6:00 AM. Limited school-time access for children, eliminating push notifications on weekdays from 8:00 AM to 3:00 PM during the school year. Robust age-assurance measures to more effectively verify the age of young users.
“Safer, age-appropriate content controls, including stronger safeguards against bullying, content promoting eating disorders, and content related to suicide and self-harm. Stronger, more user-friendly parental controls and limits on social-comparison features, including beauty filters and visible like counts that have been linked to poor mental-health outcomes in kids and teens.
“Both the implementation and effectiveness of the features will be regularly assessed by an independent auditor and the settling states.”
So this seems like a pretty big deal. What comes to mind when you see this news?
Is it a big deal? Well, actually, what comes to mind, number one, is that this is the early call for feedback because I have a number of conflicting takes on this, and there's a real opportunity to sway me personally, I think, one side or the other around a number of these topics.
Okay.
1. Parents Must Do Their Jobs
First up, my default position is that parents need to do their jobs.
Mm-hmm.
There's a certain implication of a societally sanctioned shift of responsibility to large, faceless organizations for raising children that properly sits with the parents. Even if it's well-intentioned, arguably necessary, or whatever case you might make, if you zoom out, there's a very disconcerting structural shift embedded in this entire topic.
Mm-hmm.
That's number one. Number two is that a settlement like this is very much an end run around the First Amendment.
Okay.
There are things in here about, for example—and I'm going to use extreme examples because people care about them—limits on discussion about eating disorders, for example.
Mm-hmm.
Or there's a ban, or a limit, on some of the beautifying filters and stuff like that. All of those pretty clearly fall under First Amendment protections.
Mm-hmm.
You couldn't pass a law that bans those sorts of things, but in this case, you have this settlement between a private company and a bunch of AGs. It's going to basically have the force of law. It's going to be referred to, and it's going to be a basis for these sorts of things—
Yeah.
—without actually being a law.
So it makes you—
Just on principle.
—instinctively uncomfortable on principle.
It makes me uncomfortable. The counter to all of this is that this stuff is hard and difficult. I worry about my son being on Instagram Reels too much, right?
Well, I would say one counter on the parenting point is that it would make parents' jobs easier across the country.
Right.
Because it does seem very, very difficult to constantly monitor how your kids are using cell phones.
Right. He does go to a school where they have these pouches that every kid is assigned at the beginning of the year. When they get there, they have to put the phone in the pouch, and the pouch automatically locks when they go into school or something like that.
Right.
Same pouch you use when you go to a Dave Chappelle comedy show. Seems like a no-brainer as far as I'm concerned.
I think they're awesome, especially in the context of my son moving back to America and going to a new school. You could imagine those first few weeks in particular. It would be so easy—
Just living in WhatsApp.
—to just retreat to a corner.
Yeah.
—and talk to your friends back in Taiwan or just be on Instagram Reels. Whereas you're there, you kind of have no choice but to talk to others.
Mm-hmm.
I've referenced this in passing as a challenge that I think my daughter had going to college, where it's just so easy to stay in touch with your old friends that you don't have the “I have no choice but to make new friends” sort of pressure that we had in college.
Sure.
And this is why it's a hard thing to talk about, because I could definitely see both sides. I could definitely be a hypocrite. And then, overarching all of it, is business-analyst Ben, which is regulatory capture, Meta burying all the bodies at once, locking in how much they're going to have to pay, and basically trying to set the rules of the road going forward.
Right, and pressuring their competition to make all of these same changes themselves, which will be interesting and is sort of a topic unto itself. In terms of whether this is a big deal, I do think it's a fair question. One thing I'm keeping in mind before declaring this as some kind of watershed moment for our relationship to technology is that Australia banned social media for kids under 16, and the early returns on that experiment are not great because kids have found ways to work around the ban.
Right.
Eighty-six percent of children surveyed reported using at least one age-restricted platform before the ban. Three months later, that figure remained above 81%, so not a huge change down there in Australia. I'm taking a wait-and-see approach in terms of what this actually means and how teens will use Meta platforms, but—
2. The Evidence Remains Uncertain
Well, the other thing to note is that a lot of the research—
Uh-huh.
—about the harms—that come from smartphones and social media is heavily confounded by changing definitions in the exact same time period—
Okay.
Basically, I believe it was related to Obamacare generally, but the definitions and the returns on prescribing certain conditions were changing at the same time smartphones became big. This rise was in 2012, or around that time period. This is something people should look into before they start accepting a lot of these quantitative figures, where people point to a specific time period and there are very legitimate and real things that would significantly change the numbers, so a seemingly shocking one-time shift would be like, “Oh, there's the smartphone. You can see it.”
Yeah.
It's like, well, the smartphone was in 2007. Why is it 2012? That is something overarching here.
And I bring this up not because I want to litigate this point, other than to say there’s a lot more uncertainty around this, and there really is an aspect of all this that you ultimately have to go on, to some extent, which is lived experience.
I was going to say, yeah.
3. Society Needs Some Guardrails
And I think the optimistic way to frame this is—and I think you mentioned this before, and I just want to build on the point you said—to the extent it helps parents parent, it’s a positive thing.
Mm-hmm.
So, for example, if you have these limitations there by default, right? Or it’s just like everyone’s doing it.
There’s a collective-action problem with social media among children.
Well, not just that, but sometimes laws are useful to the extent they express societal disapproval.
Mm-hmm.
Right? This is where I’m not a libertarian around things like legalized drug use, for example, or sports gambling. There was, I think, a real push among a certain cohort of people who liked to smoke a joint on the weekend in college. They always felt vaguely guilty about that because they knew they were breaking the law, and they were motivated by the desire to relieve their guilt by making it legal—not thinking about the implication of getting industrial-grade weed that people can get and smoke every single day, and you walk down New York City and it stinks.
Mm-hmm.
Right? There’s some aspect here of, even if the reality is kind of gray and ambiguous—and maybe this is my living-in-Taiwan background. I’ve talked about this in the context of the U.S., China, and Taiwan. Sometimes gray is good. It’s okay to be there. We don’t have to make everything black and white.
There’s a drive here like, “We had to make it clear this is legal because I felt bad—”
Right.
—kind of breaking the law. Actually, sometimes gray’s good, right? Is there an aspect of this where society broadly signaling that it’s okay to police your kids means parents can point to the law?
Mm-hmm.
They can realize that the default is limiting exposure, as opposed to the other way, where the default is whatever you may do. I mean, it’s almost a small-c conservative view of this.
Right.
You know, this is not a right-left debate. This is a conservative-libertarian debate. That’s what this question is coming down to in many respects.
Yeah, and I would add, in terms of what you were talking about with our own lived experience, I do think that’s why a lot of people identify with research that links social media use and phone use to poor mental-health outcomes. I think a lot of people look at their own experience using these tools and are not a fan of how it makes them feel. And they project that experience onto society.
The U.S. numbers vary wildly from the rest of the world, which adds credence to the—
Definitions.
—the U.S. definitional change that sort of happened. But to your point, how many of us have looked up from 2 hours on Reels and been like—
It feels true. If you’ve used a smartphone over the last 15 years, it feels true. But I also, from a social perspective, do think it’s a small-c conservative position to say regulating human behavior to optimize for productivity and a healthy society became taboo over the last 20 or 30 years, but there are healthy arguments in favor of laws that do that.
Again, if Twitter just shut off for me after 1 hour every day, I would definitely be better off. And I should make that change.
Yeah, there’s a real—I think this is going to be increasingly important—where Scott Alexander wrote a piece a long time ago about gray tribe versus blue tribe. It was really compelling and framed the reason why the gray tribe and blue tribe hate each other as being that no one even thinks about the red tribe.
Hmm.
That’s way out there. But that was about a distinction on the left. If you read Scott Alexander, you saw a lot of the rationalists and EA stuff, because he’s a brilliant writer and a very foundational figure with a lot of that stuff.
But on the right, the distinction between conservative and libertarian is actually a fairly massive—
Gulf.
—chasm.
Okay.
Chasm, whatever it is.
Sure.
Right? I can see that distinction actually coming to the fore more and more, and a lot of people being somewhat oblivious to it just because they’re not even aware of the distinction.
The age-old fight, sure.
And this is a great example, to your point. Another example of this is—maybe I don’t know how much this maps—another issue I have a really hard time talking about because I can’t decide—
Mm-hmm.
—the age-verification question.
Yeah.
On the one hand, if we must have age verification, I do think that should be Apple and Google’s responsibility.
Mm-hmm.
Already, they have so much trust. We’ve talked about this. You trust your smartphone provider the most. They have root access. They see everything. Apple’s entire privacy proposition is predicated on the fact that, “Trust us, and we’ll take care of everything else.”
In this context, them verifying who you are and giving you an API that says whether you’re above age or not feels much more scalable, usable, and preferable than Facebook having to—
Or each app in this future having to jerry-rig its own age-verification process. It just doesn’t seem very efficient.
Right. They’re being mandated.
Right.
They’re being—no, not just that. They’re being mandated to pry ever more deeply into understanding who you are so that they don’t break the law.
Which we all complained about for 10 years.
Right.
So, yeah.
Exactly. It’s funny how stuff changes, right? On the other hand, this is the libertarian side: Who are you to be verifying who I am?
Granular data. Yeah, sure.
And to have an API that can be abused—used in ways it wasn’t intended to—to ascertain who I am. Sure, it’s bad that Facebook can guess pretty well. We’re going to make it binary: yes or no, right? And sort of really understand deeply who you are.
I think I come down on Apple and Google providing this API just because the current of society is going in this direction.
Mm-hmm.
If we’re going to do it, let’s do it well, even while I reserve the right to object to the overall doing of it as we go.
4. Meta Turns Regulation Into Leverage
Fair enough. It will be interesting to the extent this could be a watershed moment for social media and how America uses social media. I think the way we get there is this is a framework that then is codified by Congress going forward, and—
That’s not going to happen, though. It can’t be, though. Number 1, there are First Amendment violations.
Hmm.
Number 2, this is just another end run around our broken legislative process. This is going to become the law, which, by the way, to the business point, is probably why Meta is not too upset about this.
Okay.
They got to negotiate their outcome.
And how this will work.
They’re still going to have lawsuits going on, but I would expect most courts to defer to this settlement going forward. There’s going to be a whole power game in suing everyone else in social media and pointing to this settlement and expecting the same thing.
Mm-hmm.
Meta gets to come in and say, “Oh…” There’s a requirement in there about responding within 9 hours at 90%, something that requires a lot of resources, right?
Right.
The classic regulatory capture: “We can handle this. Oh, and by the way, we can handle $18 billion,” or whatever it is, paid over 10 years. “And we only have to pay if YouTube and TikTok sign up as well.” Hilarious.
Not even a drop in the bucket—a molecule in the bucket—for Meta over the next 10 years.
And then they put this to bed. This has been an overhang on the stock for a while.
It’s very—like, even if overall they think it’s unfair, they didn’t admit fault, right? Even if they think it’s unfair, it’s not right, and the measurement’s all wrong, large corporations are political animals. This is where the politics are going, and from a business-management-analysis perspective, it’s very smart.
They’re getting ahead of it. They’re putting it entirely to bed. They got to negotiate it around their preferred parameters.
Mm-hmm.
They get to layer on this pretext: “Oh, sorry, we really have to profile you because the age requirement requires us to. It’s so sad.” If you’re going to get there, it’s best to get there on your terms, and that is kind of what Meta did.
Yeah, I agree. The only note that I would add on the potential for legislation is that you're right: betting on Congress getting its act together in this area and passing any legislation is usually a bad bet. However, I do think you could overcome some of the First Amendment concerns because you're talking about children and child safety as a compelling interest alongside the First Amendment concerns. I don't know exactly how that would be resolved, but I do think that there's political will to go in that direction. So it'll be interesting to track over the next several years here.
As far as the competition and Meta's competition, Mike Isaac on Twitter shared an open letter that Meta has written to fellow social media giants YouTube and TikTok, imploring them to implement these changes as well. That letter ran as a full-page print ad in The Washington Post, the Los Angeles Times, and The New York Times on Thursday. Isaac noted that Snap was not even mentioned, which is an amusing indication of where that platform stands. Many have compared that to Don Draper saying that he'll no longer represent tobacco companies. Do you like that comparison for Meta there?
It's one of those things: make lemonade out of lemons.
Right.
We are the leaders in this. We've done all these sorts of things.
Just like Don did.
Have you thought about whether YouTube and TikTok would ever comply with Meta's demands there and make these changes unilaterally themselves without prodding from litigants?
I think there's a good chance they will. Basically, you're the lawyer; you can tell me. This case is not precedent-setting like Supreme Court precedent, but there's a whole other arena of the political culture around court cases where this is de facto precedent-setting. And there were probably 1,000 lawsuits filed yesterday.
Courts dealing with these novel issues are going to look at how other courts handled this. How did it get resolved? So it's going to be a pretty persuasive tentpole that people look to.
Right. And so probably the question facing Google and TikTok is, do we want to wait for the inevitable lawsuit that is probably going to be settled on the exact same lines as this, or do we want to—
Do we get ahead of it? Yeah, and go there ourselves. Okay. Well, we'll see.
So we'll see. I could imagine—my guess, my instinct is that Google will move first—
Mm-hmm.
—and TikTok will try to skate by for a while. But then again, TikTok has been very eager to be seen as following the law. Yeah, I'm going to reverse that. I think maybe TikTok will move first.
Okay.
I don't know.
Well, time will tell. Final note, Sagar Ejati tweeted on Wednesday, “Is there any good argument for why the Meta changes for teens shouldn't be universal for all adults?” I wouldn't have a problem if they went in that direction for reasons stated previously. If we're nominating classes of people who need to be regulated, though, I would like to nominate boomers and have their social media use regulated before all of society.
Right. We do need age limits. The age limits are just all in the wrong direction.
Exactly. Anyone over—
Kids are fine.
—65: strict limits. All right, and that is the end of the free preview. If you'd like to hear more from Ben and I, there are links to subscribe in the show notes, or you can also go to sharptech.fm. Either option will get you access to a personalized feed that has all the shows we do every week, plus lots more great content from Stratechery and the Stratechery Plus bundle. Check it out, and if you've got feedback, please email us at email@sharptech.fm.