We're sunsetting PodQuest on 2025-07-28. Thank you for your support!
Export Podcast Subscriptions
cover of episode (Preview) Meta’s Moderation Changes, Tech’s Evolving Political Calculus, The Importance and Difficulty of Maintaining Principles on the Internet

(Preview) Meta’s Moderation Changes, Tech’s Evolving Political Calculus, The Importance and Difficulty of Maintaining Principles on the Internet

2025/1/14
logo of podcast Sharp Tech with Ben Thompson

Sharp Tech with Ben Thompson

AI Deep Dive AI Insights AI Chapters Transcript
People
A
Andrew Sharpe
技术政策分析师和播客主持人,专注于大科技公司和政府政策的交叉点。
B
Ben Thompson
创立并运营订阅式新闻稿《Stratechery》,专注于技术行业的商业和策略分析。
Topics
Ben Thompson: 我赞同Meta调整内容审核策略,我认为这并非向右倾斜,而是找到了更有效的方法。模仿Snapchat的故事功能就是一个成功的例子,证明了有时候模仿是正确的策略。在软件领域,虽然有无限的可能性,但最终的限制是用户的注意力,产品必须满足用户的需求。Meta的策略调整类似于微软将Office引入iPad,以及马斯克的社区评论方法,都体现了优先满足客户需求的重要性。 Andrew Sharpe: 我认为Meta的调整是必要的,因为之前的审核机制存在重大缺陷。2016年特朗普当选后,以及COVID-19疫情期间,之前的审核机制在应对错误信息和有害言论方面都失败了。对有害言论和错误信息的定义被过度扩展,导致了疫苗怀疑论盛行,以及对儿童健康的损害。容易找到不良言论并不代表整体讨论的价值,中央审查机制的建立也表明了人们对言论审查的渴望,而非对言论自由的坚持。我们需要根据新的信息更新之前的观点,认识到言论审查的负面影响,以及可扩展性解决方案的重要性。社区评论系统虽然不完美,但它是目前较好的选择。 Ben Thompson: 我认为Meta调整内容审核策略是正确的,这并非向右倾斜,而是找到了更有效的方法。模仿Snapchat的故事功能就是一个成功的例子,证明了有时候模仿是正确的策略。在软件领域,虽然有无限的可能性,但最终的限制是用户的注意力,产品必须满足用户的需求。Meta的策略调整类似于微软将Office引入iPad,以及马斯克的社区评论方法,都体现了优先满足客户需求的重要性。可扩展性是解决大规模内容审核问题的关键。 Andrew Sharpe: Meta之前的审核机制存在重大缺陷,在COVID-19疫情期间表现尤为明显。对有害言论和错误信息的定义被过度扩展,导致了疫苗怀疑论盛行,以及对儿童健康的损害。容易找到不良言论并不代表整体讨论的价值,中央审查机制的建立也表明了人们对言论审查的渴望,而非对言论自由的坚持。我们需要根据新的信息更新之前的观点,认识到言论审查的负面影响,以及可扩展性解决方案的重要性。社区评论系统虽然不完美,但它是目前较好的选择。

Deep Dive

Key Insights

What is the significance of Meta's shift in moderation policies?

Meta's shift in moderation policies reflects a move towards more scalable solutions for handling speech on large platforms. The company is adopting approaches like community notes, which rely on user-driven consensus rather than centralized fact-checking. This change is seen as a response to the failures of previous moderation systems, particularly during events like the COVID-19 pandemic, where suppressing dissenting voices led to unintended consequences like increased vaccine skepticism.

Why is the community notes system considered a better approach for moderation?

The community notes system is considered a better approach because it is scalable and relies on user-driven consensus rather than centralized fact-checking. It allows for a more representative and self-policing mechanism where different viewpoints must agree on a note for it to be approved. This system addresses issues like brigading and the challenges of defining truth in a global context with diverse value systems.

What were the consequences of the moderation policies during the COVID-19 pandemic?

The moderation policies during the COVID-19 pandemic led to the suppression of dissenting voices on issues like vaccine efficacy, school closures, and COVID origins. This suppression contributed to increased skepticism of vaccines in general, including those for diseases like measles. The policies failed to achieve their intended goals and instead caused broader societal mistrust in public health measures.

How did the centralized fact-checking system fail in social media moderation?

The centralized fact-checking system failed because it imposed a particular point of view, often aligning with traditional media consensus, which did not resonate with diverse global audiences. This approach led to the suppression of dissenting voices and created a disconnect between the moderation apparatus and the users it was meant to serve. The system also struggled with scalability and the complexity of managing truth in a global context.

What lessons can be drawn from the moderation challenges of the past decade?

The past decade's moderation challenges highlight the importance of scalable, user-driven solutions like community notes. Centralized moderation systems, particularly those that suppress dissent, often fail to achieve their goals and can lead to broader societal issues like increased skepticism of public health measures. The evidence suggests that allowing free speech, even if it includes some harmful content, is crucial for maintaining trust and addressing systemic harms.

Chapters
This chapter explores the strategic brilliance of Meta's adoption of features from other platforms, using the example of Instagram Stories. It questions whether copying can be a successful strategy, particularly in the context of software's malleability versus hardware's constraints.
  • Meta's strategic copying of features from other platforms is discussed.
  • The contrast between hardware's and software's design constraints is examined.
  • The importance of user needs and product utility is highlighted.

Shownotes Transcript

Translations:
中文

Hello and welcome to a free preview of Sharp Tech. Well, and it is unfortunate since Snapchat came up with the idea to have big bad Facebook come through and just take what it wants and implement it at a much bigger scale. But from a strategy standpoint, elegant and effective. So well done, Zuck. So can I explain? Yes, please do. Why did you consider this analogy poignant? Okay.

Okay, so I reread the piece. You start that piece by comparing or by complimenting Samsung for how well they aped the iPhone design and that in that execution. Oh, that's always been one of my all-time zags because everyone always complains about Samsung copying the iPhone. I'm like, look at every other phone company pre-iPhone. They're all dead. It's like sometimes aping is the right thing to do.

Right. And so you write, to be sure, the physical constraints of hardware lead much more quickly to one ideal solution, that ideal solution being the iPhone form factor. The infinite malleability of software seems to give a much more expansive canvas for doing something original. And yet, even if it is possible to build just about anything, the ultimate constraint is the attention of the end user.

What do they actually want to do? And does your product help them do it? And then you go on to compliment Meta and the Instagram Stories strategy. But the idea there of having an ideal solution or an ideal framework, even in the software space, that reminds me of the speech questions here. Because a bunch of people in the media have framed the Meta changes as Zuckerberg bending the knee to the right.

But number one, it's not Meta's fault that having a bias toward more speech and not less speech is now conservative coded. And two...

It's possible that this is Meta just moving closer to a better, smarter answer for how platforms of that size should approach speech questions. And Musk came up with the community notes approach, and it just works. And it's how a company like Meta should approach all these dilemmas. And yeah, and so that's ultimately like...

the parallel there is it's just a better idea and meta is recognizing it and deploying it. Yeah. Does that make sense? No, it does. I think community notes did proceed Musk, but it's one of those things like where I would, whatever I pray Satya Nadella for the way he changed Microsoft and like his first big move was to put Microsoft office on iPad, even though it wasn't ready for windows eight sort of back then sort of touch interface and,

And I would always get pushback from Microsoft people like, hey, that was developed under Steve Ballmer. I'm like, I don't care who it was developed with. There is a matter of the resonance and the way that you're actually changing the priority and making the focus that, no, this is actually what we're going to do. We're going to be cross-platform. We're going to meet our customers where they are. We're not going to handicap –

some products to favor the other one, you know, this sort of tail weighing the dogs situation that Microsoft got in. And I think that's fair for community notes. It was a thing before must took over. He made it into sort of a much larger thing. And in conjunction with wiping out their sort of fact checking sort of apparatus, uh,

And yeah, I mean, the rep, like nothing is perfect. Let's start out. Number one, this is the overarching point to land and we'll get into lots of people's pushback and objections along those lines. But the reality is we're dealing with stuff that's super messy and there is no right answer. And in that context, you have to sort of have our priority stack. Like what is sort of actually at the top? What works?

And for social media, when you have this scale challenge of anyone can post, anyone can put anything, the solutions that work have to be similarly scalable. And I think that the – and you did say one thing earlier, as long as I'm on the correct Andrew train here, about they got rid of their army of fact-checkers earlier.

The way Facebook tried to work around this problem of, look, we don't want to be in the truth-defining business is they outsource the fact-checking. Right. Fact-checking contractors, correct? That's right. And so you had all these traditional media organizations that spun up entire orgs that were focused on fact-checking. You have people whose career was to be official fact-checkers. And the problem is that –

I mean, there's a broader discussion about the way social media tends to form sort of consensus thinking. And I think it's been most pervasive in the traditional media. And you had this sort of imposition of a particular point of view.

And this isn't just – it's most pertinent in the U.S. because the U.S., our culture wars are so visible and basically the spectacle for the entire world. One of the greatest entertainment products of all time is U.S. culture wars. The ultimate reality TV show. It kept people very busy over Christmas break. I can tell you that much and sort of it still is.

But the reality, but it's also a worldwide issue. And different places and different people have different value systems. And it's a really hard problem on how you're going to manage that and organize that. And I think the thing with the community note system is you can critique it by saying there is an abdication of accountability and that who knows who's

Who are these actual community note checkers? But actually, it's also a better representation of those being served. It's sort of a self-policing sort of mechanism that I think works well. And there's lots of interesting things about how X does it. Different people can propose community notes, but it has to come to some sort of consensus.

And there's like a tracking of the note takers and what they approve and disprove. And there has to be some aspect of people that regularly disagree, have to agree in a particular note for it to work. So it's trying to get against like one of the challenges of social media is you get sort of this brigading effect where people will get groups together to

We've seen this with like to downvote a particular post or to report. This was a challenge when you got into the reporting system, like reporting people for objectionable speech or DMCA. Or people do bomb people's Yelp pages and whatnot. That's right. It's been a problem on the internet since the internet came into existence. That's right. So community notes is not perfect. It can cause problems in this regard, but DMCA,

There is something, as you were sort of feeling in the dark to solve these various issues that come with scale. I do think as a general principle, scalable solutions are going to be the better ones in the long run. And community notes is scalable in a positive way, in a way that empowering a Mandarin class to rule from on high what's true or what's not turned out to not be, I think, a very good fit for a network environment.

of billions of people.

Right. Well, and I think you said a couple of things there. One, community notes is not a perfect solution and there are no perfect solutions to what are just really, really difficult problems that, again, people have been trying to solve and grapple with for 30 years on the Internet. And now the scale is much larger. But a lot of these problems aren't necessarily new. And community notes may be the best of several bad options for moderating at scale, but

And the other factor here is that we just watched in the wake of the Trump election in 2016, like the moderation apparatuses that were developed in those years were

COVID happened and was like this massive test for the net impact of those policies. And I would say there were some pretty resounding failures there, which Zuckerberg was honest about. But like whether you're talking about vaccine information and whether it was essential to get the vaccine, the origins of COVID, the impact of school closures on kids and

the way COVID was transmitted. These were all very contested issues. And the voices that dissented from establishment media consensus were pretty frequently suppressed. And it turns out a lot of those voices may have been right. And so whether you can remove Biden and Trump from the equation, it just seems clear that over the past eight years,

expanding the definition of harmful speech and misinformation has not achieved its intended objectives because obviously vaccine skepticism is at an all-time high after the vaccine approach. This is the one that really bugs me because I was like, you know,

When the mandate stuff came out, I was like flipping out. This is going to be a disaster. And the reason it's going to be a disaster is not, has nothing to do with COVID. It has to do with, we, we have really important vaccines that work and they're,

The way this is going to translate broadly is into skepticism of all vaccines. And that's exactly what we see. There's numbers coming out about the relative penetration of like the measles vaccine and things like that way down just from even five years ago. And that's like... And these were...

You couldn't even argue. I, you can even argue this at the time, right? Like, like, like something went terribly wrong. And there was in all those moderation debates between 2016 and 2020, it was a hard position as someone that for 98% of the time, and we can maybe get to the 2%, I was super hardcore. What free speech matters. Like, like it actually is important. And you were always sort of arguing in theoreticals, like,

Yes, the harms are visible that you're trying to limit, but there's systemic harms you're introducing to the system if you don't allow free speech, if you don't allow dissent, even if 95% of the stuff you want to do is actually wrong, right? The 5% is really, really important. And I do have a degree of frustration with folks that don't look back in COVID and realize that.

It wasn't theoretical anymore. You had the vaccine issue. No longer an abstract argument. It became very literal. The school thing's crazy, right? Like I still talk to people today about,

That don't understand the age skew in COVID deaths and the fact that the majority of COVID deaths were people of over 80 years old. And it would, you know, the risk to kids was very, and you think if you just zoom out and had a theoretical discussion about what as a society do we want to prioritize and what is most important to,

I think everyone would say kids are the most important, right? Like that's our future. Like there's lots of cliches sort of about that. And yet at the time, that discussion could barely even be had. And now we're reaping the consequences. And the challenge with this is you could always find examples of bad

bad speech of stuff that you wanted to sort of censor and limit. This is goes back to our bit about stuff stays on the internet forever. And by the way, there's search. So you, it was the easiest sort of story to write in the world. You go to Twitter, you go to the search bar, you type in something objectionable. You find 20 tweets that say that. And you write a story about, yeah, there is this terrible trend on Twitter. People type on X, Y, Z.

And the problem is that number one, that's not necessarily representative. And number two, you created this, like, I just, this has to be, how can people not see this as a wake-up call where you have number one, once you have the means to

of censoring people centrally, people reached out for it sort of desperately. Like the commitment to free speech turned out to be a matter of capability, not necessarily a matter of principle. And then number two, the fact that you could find bad stuff trivially

was not dispositive about the overall value or overall sort of discussion and now we're reaping the consequences we're reaping the consequences for years and years when in 2017 you had to say look in theory this is why you need free speech so i'm going to support it now today you can point to evidence and i do think there's people that need to update their priors like like all

Everyone's the thing. Yeah. I mean, that's where I am just super sympathetic to the changes that Meta is putting forward now, because the key point is, in retrospect, even if you agree with all the intentions underlying the former moderation regime, the goals of those policies were not being achieved. They were policing covid vaccine skepticism and there's now an uptick in skepticism of all vaccines there.

Trump was kicked off major social media platforms and he just won the election and was more popular in basically every demographic in the United States. So whatever they were trying to do with those policies was not being done effectively and was causing more problems than it was solving. That's a really important point. None of this worked.

So what are we trying to do? Like, like why you have to update with information, even if you thought something different at one time, look at the evidence of the last few years, like all the, and a lot of the progress on social issues that people felt needed to be defended through sort of censorship. All that progress happened in a world without censorship, right? Like, like, like everything sort of backslid. Well, there was the, these sort of strict controls going on. And, and,

I don't know. It's like, it almost ties into like some of our parenting advice and like the way to sort of like, like, like, like think about the world.

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 at sharptech.fm.