from the that's-not-how-any-of-this-works dept
It remains somewhat surprising to me how many people who have ideas for Section 230 reforms clearly do not understand the law and how it works. Perhaps much more surprising is that, when experts try to highlight where their analysis has gone wrong, these “reformers” double down rather than correct their previous faulty assumptions. Dean Baker is a fairly well-known economist whose views on copyright we’ve highlighted in the past for being quite insightful. Unfortunately, Baker seems to feel that his insight in these other areas allows him to skip the basics on Section 230, defamation law, internet business models and the like. A year ago he wrote two separate very wrong and very confused blog posts advocating for the full repeal of Section 230. Both of them misunderstand how 230 works, its interplay with the 1st Amendment, and how defamation law works.
I had planned to write a response to them last year, but never got around to it. However, Baker is still at it, and after Jeff Kosseff and I spent some time trying to explain some fairly basic principles that you need to understand in order to explore the trade-offs in any Section 230 reform proposal, Baker wrote a long thread ignoring the points we raised, and insisting that his plan for 230 reform wouldn’t run into any issues. He’s wrong, and despite my going back and forth with him over a dozen times, it’s become clear that he has no interest in exploring or correcting the mistakes in his analysis. That said, I do think that he makes so many fundamental errors, that it might be useful to go through his thread to explain to other, more open-minded folks, the very significant challenges in these plans to reform Section 230.
Baker’s latest proposal is apparently no longer the full “repeal” of Section 230 he wanted a year ago, but now just that only subscription supported sites (with no advertising) get the benefits of Section 230. As laid out in his thread, the underlying theory is that Facebook is too big, and by removing Section 230, this would force Facebook to downsize. This is wrong for a bunch of reasons, some of which we’ve explained before, but we’ll get there. He seems to no longer support a full repeal of Section 230 because people highlighted how it harms other sites. So his new version is that Section 230 is only removed for sites that have advertising as their main business model under the (incorrect) theory that this will magically create a world where every site other than Facebook moves away from ads to subscription only, and that somehow makes Facebook smaller. Substacks for everyone!
So, again, the keys to Baker’s plan seem to be that by removing Section 230 for ad supported sites, it somehow (1) forces Facebook to shrink and (2) forces paywalls all over the internet. And this is somehow good. Both assumptions are fundamentally wrong — but it’s important to understand why, because this mistake is made by too many people who haven’t bothered to take the time to understand Section 230.
Section 230 does not provide an outsized benefit to Facebook — instead, it protects everyone else significantly more than it protects Facebook.
This is one thing that many, many people fundamentally misunderstand about Section 230. They think that because Section 230 “protects” Facebook and Facebook is so big, that Section 230 protects Facebook more than it protects others, and therefore any removal of 230 protections will have a greater impact on Facebook than other sites.
The problem with this is that the real benefit from Section 230 is not the underlying protection from liability, rather it’s the procedural benefits that 230 provides that help companies get out of frivolous lawsuits at an earlier stage. We’ve discussed this before a few times, but many people seem to miss it. There are two important issues as it relates to liability for websites in cases that try to drag them in: (1) what is the likelihood of any underlying cause of action actually leading to liability (outside of Section 230) and (2) how expensive is it to find out whether or not that liability sticks.
In the vast majority of cases, there is no underlying cause of action that will create liability. We’ve actually seen this in action in the few cases that get past the Section 230 hurdle. One of the most famous cases that chipped away at Section 230 protections was Fair Housing v. Roommates, in which the court determined that, while Section 230 protected Roommates.com from content that other users created, it did not protect the company from liability for the pull down menus that it created itself.
Many people think this means that Roommates.com lost the case, and very, very few people realize that years later Roommates.com still won, when the courts determined that even though 230 didn’t kick the case out early, Roommates.com still didn’t actually violate the law. The same is true of the other big 230 exception case, the more recent Enigma Software v. Malwarebytes case, in which the court (somewhat bizarrely) argued that Malwarebytes doesn’t get Section 230 protections in cases where a malware designation might be deemed anti-competitive. But, in the end, many years later, Malwarebytes still won.
Again, the key benefit to Section 230 is not that it removes all liability, but rather that it gets cases dismissed very early on, cases that would have almost no chance if they went through the full litigation process. In other words, it’s a form of protection against frivolous lawsuits, and the main mechanism involved is getting cases dismissed earlier, rather than years (and millions of dollars later). That helps smaller companies way more than it helps Facebook. Facebook has all the money in the world and it can afford to litigate these cases all the way through. It would cost the company pocket change, but the company would likely still win in the end.
Smaller companies, on the other hand, cannot afford the costs. Getting a case dismissed on 230 grounds might cost six figures. Having to go all the way through the full litigation is more like 7 or 8 figures (depending on circumstances). Facebook can find that money in the seat cushions of their office couches. Smaller companies cannot.
Dean also appears to not understand how defamation law works at all. In his thread, he seems to think that without Section 230, if someone posted something defamatory that would automatically make Facebook liable for the defamation:
If a woman says in a Facebook post that Donald Trump sexually assaulted her, and he in fact did not, is there any doubt that he would sue both the woman and Facebook?
— Dean Baker (@DeanBaker13) December 28, 2021
Except, that’s wrong. First off, the actual bar for defamation is quite high, especially for public figures. Baker, incorrectly, seems to think that merely saying something false about a public figure is defamatory. That’s not how it works. It has to meet the standard of defamation, including the actual malice standard (which is not just that you were really mad when you said it). Second, and much more important for this situation, is that if the speaker was liable, that does not automatically mean that the intermediary would be liable. Under the two key cases prior to Section 230 becoming law, Cubby v. Compuserve and Stratton Oakmont v. Prodigy, the courts had to wrestle with what makes 3rd party intermediary liability consistent with the 1st Amendment.
And the ruling in both cases would go directly against Baker’s ideas here. In Cubby, the court determined that the website needed to know about the content before it became liable for it. So, in Baker’s hypothetical above, Facebook wouldn’t face automatic liability unless it was first provided notice that the content was potentially defamatory. And, in Stratton Oakmont, the court said that any moderation meant you were admitting to knowledge of what you left up — thereby encouraging no moderation at all (if you don’t look, you wouldn’t be liable). So Baker’s insistence that he’d want more moderation to happen under his plan isn’t necessarily supported by the historical precedents under the 1st Amendment.
If the courts (as would be most likely) followed a Cubby-like precedent, then all it would do is create a very expensive process for any website to handle moderation, because each “notice” would likely require adjudication. As we’ve seen in the copyright space, this greatly favors big companies who can hire teams to review takedown notices, meaning… it favors Facebook and would destroy smaller competitors. On the other hand, if courts actually moved to a Stratton Oakmont style standard, then it would create massive incentives for every website to do no moderation at all, filling websites with garbage (which goes against Baker’s desired situation in which Facebook took down more content). But in such a world, there would be so much garbage on any site that inevitably users would rely much more on recommendation algorithms to find the content they actually want to see — and that again favors the largest companies who can invest in such tools. Like Facebook.
I asked Baker privately why he ignored these historical precedents, and he insisted that since he’s talking about a new law, it doesn’t matter. But any law has to comply with the basic principles of the 1st Amendment, and if you’re going further than this in automatically placing liability (rather than simply removing the procedural benefits of 230) as Baker seems to want, well, that’s not going to survive any level of 1st Amendment scrutiny. The NY Times v. Sullivan sets the bar for defamation very, very high. You can’t just wipe that out and claim that your new lower standard (“someone said something false!”) automagically applies to third party distributors.
And, in fact, we know how this plays out in practice because we’ve seen it in the copyright context. Given Baker’s earlier research and writings into copyright, you’d hope he’d understand this, but he seems to have not bothered. In the early days of user-generated video content online, there were a large number of companies that rushed into the space. YouTube took an early lead, but there were many others, including Veoh. Both YouTube and Veoh were sued for copyright infringement. In copyright, there’s no Section 230. Rather there’s the DMCA 512 safe harbors, but (unlike Section 230) you have to fight out in court as to whether or not you comply with the factors to get the 512 safe harbors.
In YouTube’s case, with the help of Google money, it had to fight Viacom in court for seven whole years, and it would have gone on longer, but after Viacom kept losing every single legal argument it made, the company finally agreed to settle without any money changing hands. Veoh, on the other hand, a site founded by Hollywood insiders like former Disney boss Michael Eisner, fought for many years in its similar lawsuit which it eventually won, but only three years after the company had shut down, citing the expensive litigation as driving the company out of business.
So, to recap: Section 230 gets you out of frivolous litigation much sooner, saving smaller companies millions of dollars. Without 230, the internet websites would likely still win most such cases (as they have basically every time a plaintiff gets around 230), but it would be a lot more expensive. That’s a nuisance to Facebook, but it’s deadly to many, many other websites.
So, Baker’s initial assumption that more lawsuits from the removal of Section 230 would somehow force Facebook to shrink and aid smaller competitors is simply wrong. Beyond explaining all the reasons why above, we have practical examples from the copyright realm with YouTube and Veoh. And if that doesn’t convince you, we even have examples from Section 230 itself. In 2018, as you’ll recall, FOSTA became law — the first major attempt to chip away at Section 230. We’ve spent tons of time explaining the societal damage that FOSTA has created, but it also didn’t “shrink” Facebook. Instead, it did the opposite.
Right after FOSTA passed, a bunch of dating websites closed down, including Craigslist shutting down its dating vertical — explicitly stating that the threat of liability from FOSTA made it too expensive to run. A year later, Facebook jumped in to launch its own dating app. So, smaller competitors get out of the market, and Facebook gets to jump in.
The same would happen under Baker’s “reform” plan here. Smaller ad supported providers would shut down, because the cost of litigation risk would be too high. Facebook would corner the market from the few competitors out there.
Subscriptions don’t magically solve any of this, they mostly make it worse
But Baker insists none of this is an issue, because his plan would encourage an internet of paywalls and subscriptions, and by incentivizing that business model, mighty Facebook would be brought to its knees. Except there’s no evidence to support that either. His entire basis for arguing that this would happen is that historically people subscribed to newspapers.
That seems rather dubious to me. It used to be very common for people, including working class people and even poor people, to have subscriptions to newspapers.
— Dean Baker (@DeanBaker13) December 28, 2021
In that context, it seems hard to believe that there would not be millions of households prepared to spend $5 or $10 a month to subscribe to a website that lets them (and their friends) say what they want.
— Dean Baker (@DeanBaker13) December 28, 2021
But that fundamentally misunderstands the nature of the internet, social media, what makes them work, and what makes this all so important for speech.
He later suggests that if sites couldn’t get enough subscription support, then that fundamentally shows they’re not valuable and it’s okay for them to go away.
If there were not large numbers of people prepared to make modest payments to get access to websites that did not limit their posting, then we can assume that most people do not care and that this is not a major issue.
— Dean Baker (@DeanBaker13) December 28, 2021
But all of that is mixing up some very, very basic concepts. First of all, social media only works well thanks to network effects. Metcalfe’s Law matters. If I subscribed to the NY Times, it makes no difference to me if you subscribe to the NY Post. There is no fundamental value difference in that case.
But for the vast majority of incredibly valuable online services — especially those involving user-generated content that need Section 230 — a huge part of the value is proportional to how many other people are also using that service. And that doesn’t work in a subscription world. Right now, I already use a half a dozen different messaging services because different friends and family tend to rely on different ones. If I had to pay $5 to $10 a month for each one, that would not work at all. Instead, the most likely scenario is that everyone would standardize on the one that everyone uses the most. And, horrifically for me (as someone who minimizes my use of Facebook), that would mean Facebook.
Also, I really wish that Dean spoke to someone who actually had some experience in the ad business and in the subscription business. Going back a decade or so it was common for people with experience in neither to insist that the two were somehow interchangeable, and you could easily convert ad-based businesses to subscription models. Except that’s not true at all, in part because of the network effects point above. For the vast, vast majority of users of any particular online service, the value they get out of the site is way below $5/month. Relatedly, the value they provide to those sites by themselves is way less than $5/month in terms of ad revenue. But the collective aspect of their usage of the site, combined with the aggregate advertising model, is what makes it work.
If you had to go subscription, you’d wipe out tons of useful services that today rely on Section 230: I can’t see how Reddit, Craigslist, Nextdoor, Glassdoor, Pinterest, Yelp, Travelocity, Eventbrite, Stack Overflow, Ravelry, DuoLingo, Fandom, WikiHow, Glitch and tons of other websites would survive. Baker seems to think that you’d just need one “social media” website, and so people would subscribe to the one they like best. But so much of the internet relies on Section 230 and ads to survive, and changing them all into subscription services would be untenable for the vast majority.
End result? Again, you wipe out most of the more innovative competition, and especially clear the field for new entrants. Under this model, new entrants would need to start with a subscription model, and it’s difficult to get people to subscribe to a service that has no track record… and has no users.
And, of course, Baker brushes off the idea that another impact of this is that it would only serve to help the wealthy at the expense of the poor, but it’s absolutely true. There’s so much value on the internet that is readily available to everyone, and Baker’s plan would lock that all up… because he’s mad at the amount of control Mark Zuckerberg has. But locking up all those useful services, and leaving it so that Facebook can clean up and provide all those services for free (while paying the pesky litigation costs to show that it’s not liable) seems like a great long term deal for Facebook.
Just as it helped get competing dating services, like Craigslist’s, shut down before opening its own service, this is why Facebook is running ads and telling Congress that it’s time to “update Section 230.” Facebook knows that the long-term impact of such things may raise some direct costs (litigation) in the short run, but over the long run, it wipes out the thing that Mark Zuckerberg has always feared the most: disruptive innovation that competes with Facebook and takes away their userbase.
Baker’s reform plan misunderstands how and why Section 230 works and where it provides the most benefit. And he misunderstands how and why network effects work online, and what his plan would do to much of the open internet. His plan would, clearly, provide Facebook significantly more power, while wiping out a stunning amount of competition. It’s a dangerous plan.