Defining exactly where a particular vendor sits in the infrastructure stack is important because this determines which legal rules apply around liability for content. If they are a publisher, they have much more responsibility for the content on their platform than if they are just a distributor.
As nicely summarised on Stratechery, any form of editorial control (automated or otherwise) is what defines a publisher (as per Cubby v CompuServe (1991) and the subsequent decision in Stratton v Prodigy (1995)). To adjust this liability and allow firms to moderate content online, the US passed the Communications Decency Act in 1996, section 230 of which allows for:
At the time this was passed, the reasoning was that the small but growing tech industry needed the protection offered to network providers like telecoms. Phone companies do not listen into all phone calls and major internet backbone providers offer a neutral service to distribute packets on the internet without filtering. As systems that purely distribute content, should they really be held liable for how the network is used?
Things become more challenging as the network vendor becomes a content platform.
Moving from network to content enabler
Cloudflare is an internet security service that sits in front of your website, distributing the content faster via their global network and protecting it from threats such as denial of service attacks.
Since launching in 2009, it has grown quickly, now providing services to 19 million sites — a significant portion of the internet. In 2018 it was processing over 6 million requests per second. This places it in the group of companies providing crucial internet infrastructure like AWS and Google. It has grown from the original DNS, CDN + DDoS protection services to offering a much wider portfolio of infrastructure products. Does that mean it has graduated from being merely a neutral network or does it have some greater responsibilities for the activities it enables?
Cloudflare has historically positioned itself as neutral when it comes to the content it serves. This has caused controversy because of the types of website it has allowed to use its services, although it is by no means alone. Gizmodo ran some research in July 2019 which showed almost 400 sites considered to promote “hate” that were being hosted by major tech infrastructure companies:
We reached out to a handful of non-profit organizations that work to monitor and counter hate-the Southern Poverty Law Center, the Anti-Defamation League, Hope Not Hate, the Canadian Anti-Hate Network, and the Counter Extremism Project. They each provided us a list of groups they see as being involved in the propagation of hate.
The organizations we looked at run the gamut from white supremacists, neo-Nazis, and chapters of the Ku Klux Klan to groups dedicated to stripping the rights of immigrants and LGBT people. There are also some neo-Confederates, black nationalists and racist Odinists in the mix as well. The list includes one website for a white nationalist beer company and a whites-only dating site.
GoDaddy and its subsidiary Wild West Domains provided services to the largest number of sites on our list, 130. Cloudflare, which provides protection against distributed denial-of-service attacks, works with the second largest number of sites, 56…Google provided services to 27 sites, Amazon works with nine, and Microsoft works with five.
Cloudflare’s position has always been that they follow the law and nothing more. They have specifically stated they do not wish to be making political judgements about what content they should or should not be serving just because they may disagree with it.
However, there are grey areas such as when they are providing services to organisations considered to be “terrorist” groups by the US Government:
Among Cloudflare’s millions of customers are several groups that are on the State Department’s list of foreign terrorist organizations, including al-Shabab, the Popular Front for the Liberation of Palestine, al-Quds Brigades, the Kurdistan Workers’ Party (PKK), al-Aqsa Martyrs Brigade and Hamas — as well as the Taliban, which, like the other groups, is sanctioned by the Treasury Department’s Office of Foreign Assets Control (OFAC). These organizations own and operate active websites that are protected by Cloudflare, according to fournational security and counterextremism experts who reviewed the sites at HuffPost’s request.
Here, the law prohibits anyone from offering “material support” but what constitutes “material” is vague:
…the Electronic Frontier Foundation, argue that “material support” can and has been abused to silence speech. Cloudflare’s general counsel, Doug Kramer, told Gizmodo over the phone that the company works closely with the U.S. government to ensure that it meets all of its legal obligations. He said that it is “proactive to screen for sanctioned groups and reactive to respond when its made aware of a sanctioned group” to which it may be providing services.
Whilst the groups that Cloudflare have been serving are clearly producing content that many find horrible (including myself), the material is not illegal in the US.
The First Amendment of the US Constitution prevents the US government from placing restrictions on freedom of speech. This does not apply to commercial entities and Cloudflare’s own terms of service allow them to terminate accounts for whatever reason they like. But they have decided to use the legal minimum.
I don’t like some of the outcomes of what this allows, but it does mean certainty and transparency. The alternative is worse.
The US government is offering no alternative
The technical design of the internet is decentralised. A large number of networks run by many separate entities all come together to provide a neutral service that allows everyone to communicate without rules. This has allowed amazing innovation but has also created huge societal problems as we deal with the fact that no single state or organisation has oversight.
One of the biggest challenges is that whilst the underlying architecture of the internet is decentralised, most of the services built on top of it are centralised. For example, there are only 2 major mobile platforms — iOS and Android; only two major desktop operating systems — macOS and Windows; only one major search engine — Google; only a few major social network operators — Facebook (owning Instagram and WhatsApp), Twitter, Snapchat.
The network is decentralised and neutral, so the distinction of distributor vs publisher makes sense. But when content is available on only a few platforms, decentralised neutrality disappears. Cloudflare is more than a neutral distributor because of the protection and CDN services it offers, but it isn’t a publisher because it is not hosting or editing the content. It enables content to exist but does nothing to edit, review or promote that content. In contrast, Facebook and Google (especially YouTube) actively promote the content they host to drive engagement with their ads.
In this middle zone, Cloudflare has decided to take the approach of a distributor network. The rules they apply are whatever US law says. They have co-opted the machinery of the US state — the government, legislature and judiciary — so they don’t need to make a decision themselves. If a the legislature or courts system decide to tighten up the definition of “material support”, we would expect Cloudflare’s position to change to apply those new rules too.
Whilst this has resulted in some horrific organisations having an easier job remaining accessible, the alternative is the arbitrary and opaque rulings of the content moderation operations of the major platforms like Facebook. Instead, Cloudflare are outsourcing their decisions to the democratically accountable US rule of law. If the constitution prevents the government from providing more clarity, then so be it. This contrasts with Facebook and Google who are defining their own rules in private, with no debate and no accountability.
Policing content is a problem that has been solved in the past through the development of Western legal systems and the rule of law. The separate powers of the state — government, judiciary and legislature — counter-balance each other with various checks and stages to allow for change. It’s not perfect, but it has had hundreds of years of production deployment and new version releases!
What has changed is the scale. And the fact that governments are delegating the responsibility of the implementation to a small number of massive, private firms.
We are seeing two approaches:
- Cloudflare: Applying a legal minimum, but in doing so providing consistency, clarity and transparency. If you don’t like it then you have a route through the legal system and/or legislature to get the law changed, assuming Cloudflare adapt their approach to any changes in the law.
- Facebook/Google: Adding their own layer on top of the legal minimum. We are only aware of what they are doing through leaks of operating documentation and there is no route to appeal or even review of what the current rules are.
Indeed, this is how Cloudflare have described their approach:
We get labeled as being free speech absolutists, but I think that has absolutely nothing to do with this case. There is a different area of the law that matters: in the U.S. it is the idea of due process, the Aristotelian idea is that of the rule of law. Those principles are set down in order to give governments legitimacy: transparency, consistency, accountability…if you go to Germany and say “The First Amendment” everyone rolls their eyes, but if you talk about the rule of law, everyone agrees with you…
It felt like people were acknowledging that the deeper you were in the stack the more problematic it was [to take down content], because you couldn’t be transparent, because you couldn’t be judged as to whether you’re consistent or not, because you weren’t fundamentally accountable. It became really difficult to make that determination.
Moving away from the rule of law?
Things changed following the recent mass shooting incidents in the US when Cloudflare decided to terminate the service of one of its customers, 8Chan:
The mass shootings in El Paso, Texas and Dayton, Ohio are horrific tragedies. In the case of the El Paso shooting, the suspected terrorist gunman appears to have been inspired by the forum website known as 8chan. Based on evidence we’ve seen, it appears that he posted a screed to the site immediately before beginning his terrifying attack on the El Paso Walmart killing 20 people.
8chan is among the more than 19 million Internet properties that use Cloudflare’s service. We just sent notice that we are terminating 8chan as a customer effective at midnight tonight Pacific Time. The rationale is simple: they have proven themselves to be lawless and that lawlessness has caused multiple tragic deaths. Even if 8chan may not have violated the letter of the law in refusing to moderate their hate-filled community, they have created an environment that revels in violating its spirit.
Terminating Service for 8Chan, Cloudflare
This is only the second time Cloudflare has terminated service for a customer, having done so with The Daily Stormer in 2017:
Our terms of service reserve the right for us to terminate users of our network at our sole discretion. The tipping point for us making this decision was that the team behind Daily Stormer made the claim that we were secretly supporters of their ideology.
Our team has been thorough and have had thoughtful discussions for years about what the right policy was on censoring. Like a lot of people, we’ve felt angry at these hateful people for a long time but we have followed the law and remained content neutral as a network. We could not remain neutral after these claims of secret support by Cloudflare.
Why We Terminated Daily Stormer, Cloudflare
Cloudflare have clearly developed their thinking since terminating The Daily Stormer. In that instance, their reputation was at stake and they could not tolerate the claims of support. One could argue that this is libel and so protection under the First Amendment could be forfeited. This is debatable. The 8chan decision is much easier to understand:
There is a flaw in the system which is that this all works so long as the end platform is responsible. But if the platform itself has been designed from the beginning to be lawless, to not respect the rule of law, to not respect abuse complaints and, in the case of 8chan, actually take an abuse complaint as an opportunity to threaten and harass whoever was sending it, the question becomes, “Who else then has an obligation to deal with it”?
If you are on Twitter right now all of the 8chan supporters are saying “Why don’t you cut off Facebook or Twitter because there are horrible things that get posted to them?” That’s true, there are horrible things that get posted to them. The difference is that those aren’t lawless platforms. They have teams and they have procedures and they take these things seriously, and they’re trying to create a community that respects the rule of law. If you don’t have that, then you do fall into a different bucket. That was the rationale that we came to.
Put simply, Cloudflare are saying: if the platform we provide service to applies the rule of law, we will too. If not, we may take action.
I find this to be a well considered elaboration. It continues to be clear what is and is not allowed, and it removes some of the risk of Cloudflare making political or ideological decisions.
Instead of inventing their own pseudo-legal system with rules and guidelines based on the values of the company which happens to be founded in a Western democracy, Cloudflare are able to take advantage of the public system of law. If the values of that system change, we would expect Cloudflare to adapt their approach too. But it will be in the knowledge that such change has been brought about by proper democratic process. That is the rule of law.
Eroding free speech
Speech is different from action. This is why they are legislated differently. Talking about killing is not the same as the action of killing. Things start to blur as you get closer to the action e.g. conspiracy, but there is still a distinction.
In the UK, section 4 of the Public Order Act 1986 creates an offence if a person “uses threatening, abusive or insulting words or behaviour, or disorderly behaviour” with “intent to cause a person harassment, alarm or distress”. This causes problems of interpretation because what is “insulting” for one person could be completely benign for another.
The Human Rights Act incorporates a right to freedom of expression but that is limited in such a broad way that it essentially allows the government to legislate how it wishes. You can say whatever you like so long as it is not prohibited. This might seem broad but there are many restrictions.
Freedom of speech is defined as an absolute right in the US First Amendment because it is so tricky to determine what exceptions should be made. We can see this with how difficult a job Facebook are finding it to moderate content on their platforms.
Distinguishing between action and speech is important:
But this is not satisfactory when speech is the first step towards action. “Incitement” used to be a common law offence in England & Wales which used to make it illegal if you:
In many cases, the law has become more precise in how it limits free speech. This offence of incitement was abolished in Section 59 of the Serious Crime Act 2007, leaving such statutory offences as in the Misuse of Drugs Act 1971. However, a lot of things are still prohibited resulting in some absurd situations:
The US constitution is absolute in its protection of free speech so there is no middle ground. This why we have the different positions being adopted by Cloudflare on the one hand and Facebook, YouTube, etc on the other. Since the US government is unable to facilitate a debate using democratic systems, there is a wide gap without regulation. This leaves it open to political pressure, something that has become a major problem on university campuses:
Whether it takes the form of “no platforming” or political activist pressure on companies to deny service to customers they dislike, it’s an attack on free speech.
The legislative process is being bypassed because of the aspects that make it work so well — it is slow and deliberate. Or in the case of the US constitution, has very high barriers to change. This is by design — it doesn’t just allow a single viewpoint from a minority to force change. The downside is that you have to accept the existence of views that you do not agree with.
“If you think of the 10 companies that have the infrastructure to really survive on the crazy dangerous internet that is out there, a huge percentage of them are advertising-driven companies,” said Prince. “They will by their very nature have a much more filtered, cleaned up version of the internet because that is where their economic incentives are.”
The overwhelming dominance of cyberspace by those handful of companies means that the battle over speech that is protected by the first amendment, rather than corporate terms of service, will be relegated to the remaining, truly public spaces in the real world, such as the public university campuses that have become the forum of choice for rightwing provocateurs like internet troll Milo Yiannopoulos and white nationalist Richard Spencer.
The likes of the Southern Poverty Law Center, the Anti-Defamation League, Hope Not Hate, the Canadian Anti-Hate Network, and the Counter Extremism Project might seem to be doing some good work in helping create the list Gizmodo used for their research quoted above but they are still unaccountable organisations.
Regardless of which “side” the definitions come from, it is the unaccountable, un-appealable arbitrary decisions that place campus protestors in the same category as Facebook and Google. At least the definition of “insulting” in the Public Order Act offences above are determined through a legal process. It’s a problem when one world view turns into the basis for censorship.
Which rule of law?
It just so happens that all the major tech companies have been founded in a Western democracy with liberal values that I agree with (having been brought up in a similar society), so I have no problem with most of the rules they create. It might have been a different story if the major internet platforms had ended up deriving their values from more authoritarian or less-liberal societies. What would this have looked like if Facebook had been a Chinese organisation instead? We only need to look at the censorship within China to get an idea.
Although it has many problems, the US is still a modern, liberal democracy where the rule of law applies and freedom (of all types) is a core part of the constitution. Cloudflare are based in the US so they have a well developed framework from which to start with. The challenge comes when operating in countries with completely different approaches to the same issue.
Companies have to abide by the rules in the countries they operate otherwise they face retaliation. As noted in the Cloudflare interview:
I think it’s hard. I don’t think we have a perfect answer. We don’t have people in every country around the world, but we do have equipment in 80 countries I think today. We operate in a lot of places and we increasingly have people in a lot of places. We have a team that’s now based in China. We have offices all around the world. It’s one thing if your servers get seized. It’s totally another thing when one of your employees gets thrown in jail. You have to think about those consequences as you’re thinking about how you operate around the world.
Cloudflare are approaching this by minimising the amount of personal information their systems store, so there is nothing for them to turn over from a government request. But when they do get requests to take down content, they can simply apply the policy of complying only with due legal process:
We regularly get requests from governments that say, “This particular thing is causing harm and they’re not responding to us, take them down.” For example, there’s an LGBT support group that operates in the Middle East, and a lot of the Middle Eastern governments really hate them and say they are corrupting the children of their particular region.
If you start developing your own set of rules of acceptable content like Facebook and Google have done, that has to be localised in every jurisdiction you operate in. It is much easier to apply the rule of law of each country, because the law making process has already been defined. A court order is distinct from the government requesting a site be terminated:
Other companies either choose not to operate in those jurisdictions e.g. Google withdrawing from China ( although they are now trying to re-enter with filtered search results) or comply with the local requirements e.g. Apple hosting Chinese citizen’s iCloud content in Chinese controlled facilities.
Commercial entities always have a choice about which countries they operate in. If the people in those organisations do not agree with the laws of the land then they should not operate there in the first place. Some people will take an absolutist view of this, others will be more pragmatic if they think the benefits outweigh the disadvantages.
Sometimes there are even specific requests to keep such sites online so they can be tracked rather than have them disappear into the darker areas of the internet:
We have yet to find a better method of deciding what is acceptable in society than a deliberative legislative process backed by democratic accountability. It is dangerous to rely on definitions of “hate” or what is “acceptable” that are not set by this process because they cannot be subject to a sufficient level of scrutiny and debate.
The scale that internet companies now operate at means they are having to deal with problems that have mostly been solved by society over hundreds of years. Our democratic institutions are imperfect but do the job best. But instead of the approach developing in public, private companies are deciding policy, sometimes at the direction of a single person (the CEO).
Each country has its own approach. This worked when activity was contained within borders, but the internet changes all that. China has the most developed system with the Great Firewall — the internet inside China is essentially its own entity, entirely separate from the rest of the world. Will we see that happen in other countries too?
…if a country like Germany says that a particular site is a Nazi-promoting site, that content is illegal in Germany, and we can’t do anything that thwarts Germany’s ability to enforce that law. And increasingly, especially as more and more of the connection to websites becomes encrypted, more and more of those requirements are going to fall to us. I think that if Germany sets policy inside of Germany and if that policy doesn’t spill over beyond its borders, and if there’s transparency to it, so that if you try to access one of these sites and it says “You can’t get this because” and there is a copy of the court order, that’s not the Internet that some of us have all imagined, which is one network all over the world, but it’s probably the Internet that is able to survive and thrive over the long-term.
Originally published at https://davidmytton.blog on August 15, 2019.