Comments on Twitter’s Country-by-Country Tweet Removal Announcement

By Venkat Balasubramani, with comments from Eric.

Twitter recently announced its decision to censor tweets on a country-by-country basis. People were up in arms and planned a #twitterblackout. It was a big story last week. (Needless to say, I didn’t participate in the blackout.)

As an initial note, Twitter’s decision is entirely defensible, and I thought Twitter (and its General Counsel Alex Macgillivray) handled it with poise. I also don’t know that its decision can easily be placed in the ‘censorship’ category since it’s implemented by a private entity, which has tremendous discretion in blocking content. (Some of this depends on the actual policy, which we don’t know the contours of.) Anyway, this is neither here nor there.

What was striking about this story was how it played out in the media–in particular, the muddled nature of the media narrative that followed this story.

What Types of Takedown Requests Will Twitter Honor?: I would have thought the key question here would be the contours of Twitter’s policy–did it remove content in response to a court order? An administrative request? A takedown from a private party? Did it matter whether the request was premised on IP infringements? (no) Could it make certain topics totally off-limits in response to a government request? Would it block accounts? (yes) Hashtags? Would it make Twitter totally unavailable in a country? Here’s a blurb from a NYT article titled “Censoring of Tweets Sets Off #Outrage” (italics added):

Twitter, like other Internet companies, has always had to remove content that is illegal in one country or another, whether it is a copyright violation, child pornography or something else. What is different about Twitter’s announcement is that it plans to redact messages only in those countries where they are illegal, and only if the authorities there make a valid request.

Huh? What’s a “valid request”? An Associated Press story (“Twitter’s new censorship plan rouses global furor“) was similarly vague about what types of takedown requests Twitter would respond to:

Twitter said it has no plans to remove tweets unless it receives a request from government officials, companies or another outside party that believes the message is illegal. No message will be removed until an internal review determines there is a legal problem, according to Macgilliviray.

There’s a big distinction between a takedown notice from a government, one from an individual (including one sent under a takedown regime such as the DMCA) or a corporation. Another story from the Times of India adds some detail and hints at this specific question (“Twitter’s censor move with eye on China?“):

some experts wonder if Twitter’s position was really different from that of Google or Facebook. “Google and Facebook have said that they would remove content if ordered by the courts, and Twitter too is saying that it can block tweets if required by the law,” said an expert. “Where laws are codified, as in Germany and France about pro-Nazi propaganda, Twitter can block pro-Nazi tweets proactively. But in countries like India, where the laws are not that specific, this will be done reactively on the basis of court orders. That’s all Twitter is saying.”

(??) It’s strange that the stories all described the key standards for what type of request will trigger a takedown in totally vague terms. Obviously it wouldn’t make sense for the stories to describe in painful detail the innumerable types of requests an entity such as Twitter receives and how it deals with each of these types of requests, but it was clear after reading these stories that the media didn’t have a firm grasp on the contours of Twitter’s ‘policy’. This was somewhat strange because this was the crux of the story, right? There’s one larger aspect of the story which was clear which is that Twitter decided that whatever its policy is regarding takedowns, its response can be limited by country or region–i.e., if one particular country or region decides to send a takedown this may not affect all Twitter users. (The content will be available elsewhere. Also, as others quickly pointed out, as a user who is trying to access content on Twitter, there are probably ways to get around Twitter’s country-specific block of content.)

Not surprisingly, many press reports cited to EFF’s statement regarding Twitter’s policy but even EFF’s statement was fairly vague on the particular point of what takedown requests Twitter will honor (“What Does Twitter’s Country-by-Country Takedown System Mean for Freedom of Expression?“):

Twitter already takes down some tweets and has done so for years. All of the other commercial platforms that we’re aware of remove content, at a minimum, in response to valid court orders. Twitter removes some tweets because they are deemed to be abuse or spam, while others are removed in compliance with court orders or DMCA notifications. Until now, when Twitter has taken down content, it has had to do so globally. So for example, if Twitter had received a court order to take down a tweet that is defamatory to Ataturk–which is illegal under Turkish law–the only way it could comply would be to take it down for everybody. Now Twitter has the capability to take down the tweet for people with IP addresses that indicate that they are in Turkey and leave it up everywhere else. Right now, we can expect Twitter to comply with court orders from countries where they have offices and employees, a list that includes the United Kingdom, Ireland, Japan, and soon Germany.

From what I gather, Twitter’s blocking policy will be implemented on a case-by-case basis and it didn’t announce any sort of policy for what types of takedown requests Twitter will automatically honor. But to me this is a key point that none of the stories really dug into.

Will Twitter Implement its Policy Only Where it has People and Offices?: This is another question that I was curious about. Will Twitter honor requests from countries where it doesn’t have offices or does this work on a case-by-case basis also? If Twitter’s assets, offices, or people are at stake then this obviously changes the calculus, but what about far-flung jurisdictions where Twitter has no presence and no expected or future relationships? EFF’s post also hints at this but doesn’t really offer specifics:

Twitter’s increasing need to remove content comes as a byproduct of its growth into new countries, with different laws that they must follow or risk that their local employees will be arrested or held in contempt, or similar sanctions. By opening offices and moving employees into other countries, Twitter increases the risks to its commitment to freedom of expression. Like all companies (and all people) Twitter is bound by the laws of the countries in which it operates, which results both in more laws to comply with and also laws that inevitably contradict one another. Twitter could have reduced its need to be the instrument of government censorship by keeping its assets and personnel within the borders of the United States, where legal protections exist like CDA 230 and the DMCA safe harbors (which do require takedowns but also give a path, albeit a lousy one, for republication).

For what it’s worth, the tradeoff between keeping a local presence and complying with a foreign court order is not anything new. Google has dealt with it, among other countries in Italy. (For all I know @amac could have been one of the lawyers who dealt with this while at Google.) Yahoo! dealt with it in France when it was ordered to take down Nazi memorabilia. In evaluating Twitter’s policy, I would guess what people would want to know most (apart from what types of takedowns Twitter intends to honor) would be what types of jurisdictions Twitter intends on screening content in.


Maybe Twitter’s decision isn’t really a policy decision to screen content at the request of governments or entities but to make available the capability to screen content by geographic regions. There’s a fundamental difference between the two. I certainly got a clear sense that there was a policy change afoot from the stories announcing Twitter’s decision. Either way, none of the stories bothered to get into the details on what I thought were the two core issues. The other tangentially related issue that did not get much attention is how Twitter would respond to requests for user information from governments. We’re not much wiser in terms of Twitter policy than we were when we started. On the one hand, this is somewhat strange, given that most reporters live and breathe Twitter, regardless of whether this is their reporting beat. On the other hand, maybe it’s an example of how social media can infect journalism? Reporters are friendly with Twitter (as an entity, or a product) so maybe they were reluctant to ask the hard questions? Maybe everyone was in a rush to get their stories out so they didn’t dig deep?

I think we’ll have to wait and see to see how the policy actually plays out, but Twitter’s actions demonstrate a commitment to free speech and openness so it should have gotten the benefit of the doubt. For whatever reason, the story just spiraled and took on a life of its own.

[For my money, one of the best stories on this was from Al Jazeera, which raises the fundamental question of what Twitter’s policy is exactly: “Making sense of Twitter’s censorship.”]

Other posts worth checking out:

* Twitter’s initial blog post which wasn’t crystal clear on the issue: “Tweets still must flow.”

* Lauren Weinstein: “Twitter’s censorship muddle.”

* Inforrm’s Blog: “Legal questions about Twitter ‘censorship’ and country-specific content control – Judith Townend


Eric’s Addendum

To see if we could get our own answers to these unresolved points, Venkat and I took our questions to Alex Macgillivray, and he generously responded to us. Our exchange:

Our Q: What types of takedown notices will be sufficient to get Twitter to take down a post? Court order? Government demand without a court order? Private demand without a court order? (I believe 512(c)(3) takedown notices already work). Others? Does it vary by country?

Alex’s answer: “We do analysis of each complaint. For example, even a 512(c)(3) request does not necessarily lead to a removal.”

Comment: I infer this means takedown demands are evaluated on a case-by-case basis. If Twitter does not have hard-and-fast rules about takedown demands that clearly work or clearly don’t, that would explain why other media outlets weren’t precise on this point.

Our Q: Will Twitter take down posts only in countries where it has a physical presence, or will it remove content from countries even where it doesn’t have a physical presence?

Alex’s answer: “Again, would depend on the requests. For example, a child pornography complaint, even from a user in China, might result in a global removal even though we are not responding in general to requests from China

and are still blocked there.”

Comment: child porn is an extreme “test case” because of its toxicity, so I’m still not clear on what happens with less toxic content. I similarly infer everything is done on a case-by-case basis, which would also explain the muddled media coverage.


Eric’s Comments

It appears Twitter thought its announcement was good news. Instead of having to remove a tweet from its database entirely, Twitter will now remove tweets only from one country’s database. This leaves the tweet up for the rest of the world, and it makes it trivially easy for people in the affected country to get the tweet if they care. Furthermore, the tweet won’t vanish; instead, it will be a “noisy withdrawal” by leaving a note that says the tweet was removed. Plus, Twitter will turn over the takedown demand to ChillingEffects, allowing interested folks to monitor the activity and find out what happened. In a world filled with irrepressible censorious impulses, Twitter’s policies were designed to make the best of a bad situation.

So how did the messaging, and the community response, go so far wrong? Twitter ran into a small but vocal minority that believe that catering to foreign governments’ censorious requests is wrong. I discussed this issue in some detail in connection with Google and China. As I wrote in connection with that situation:

what should a US service provider do when trying to expand internationally? It has a few options, none of them particularly attractive:

* It can skip unreasonably censorious markets altogether, like Google proposes to do in China.

* It can comply with local laws, even though that runs counter to US laws and norms.

* It can ignore local laws, which is typically not a successful plan. In extreme cases, it can lead to local company executives going to jail.

* It can try to change the local country’s laws to be more like ours, either through direct advocacy or by asking the US government to pressure the local government. We routinely use trade negotiations to do this; for example, we have successfully exported our copyright laws this way. But countries usually aren’t thrilled to have the US tell them what their laws should be.

Undoubtedly, the purists would prefer it if Twitter just stayed in the US “bubble” and engage in regulatory imperialism by getting foreign governments to see and do things “our way.” A lot of self-satisfied hubris underlies that stance; something we saw with Google’s situation where many people appeared to think that denying the Chinese people access to Google would bring the Chinese government to its knees (it hasn’t yet). Ironically, American regulators have been on a censorious rampage recently (see, e.g., SOPA, Wikileaks, Operation in our Sites, etc., etc., etc.), so we’re hardly on any moral high ground.

The reality is that iif Twitter chooses to expand globally, of course it will have to comply with local law, and of course other countries will require Twitter to take down posts. Twitter has built a technical architecture to reduce the collateral damage of such censorial demands. And I, for one, believe that American dot-coms do more to spread free speech by supplying the technology, even if hobbled through censorship, on an international basis than by not offering the technology at all.

In the end, though, Twitter’s move–combined with similar moves, like Google’s redirection of Blogspot on a country-by-country basis–remind us that geographic borders remain incredibly relevant to the Internet. This is a political reality, not a technical imperative. Technologically, the Internet is a borderless electronic network, but we continue to erect artificial geographic borders anyway. (See my post, Geolocation and A Bordered Cyberspace). Once again, with censorious proposals like SOPA/PIPA (and, for that matter, OPEN) that seek to create a Fortress USA, America is teaching the world how to embrace artificial geographic borders rather than teaching the world how to tear them down.

One thing I don’t understand: if Twitter can turn tweets on and off by country, will that mean countries can assert jurisdiction over it even if Twitter doesn’t have a physical presence there? Recall how these issues played out in the LICRA v. Yahoo case, where Yahoo’s ad geo-targeting was held against it. Because Twitter can customize views of its databases on a country-by-country basis, foreign governments have a good argument that Twitter can “control” what content goes into a country. Recall, for example, the kerfuffle about Britain’s “super-injunction” against Twitter. Even if Twitter didn’t have a Britain presence, could Britain now have more leverage to force Twitter to honor its super-injunction? Or, could a foreign country assert that Twitter needs to comply with its data privacy laws on the theory that Twitter could simply turn off tweets in that country if it doesn’t want to comply? I’m not sure how Twitter will now explain why it’s chosen not to comply with a foreign country’s laws, irrespective of its physical presence.

I also asked Alex about this issue:

Our Q: How do you think this policy will affect Twitter’s compliance with laws in countries where it doesn’t have a physical presence? In other words, because Twitter could simply choose to remove all tweets from showing in a country, Twitter might have a more difficult time arguing that it had no choice about whether or not to show tweets in the country. So, for example, a country may assert that Twitter should comply with its data privacy laws for users in its country even though Twitter has no physical presence there.

Alex’s response: “This doesn’t change our philosophy with respect to freedom of expression and I don’t think it changes the pressure we’ll get from countries (other than the transparency piece). Companies that have no way of doing local withholding still get plenty of pressure to do removals. Generally ‘I don’t have a way to just do this for your country’ is a positive from the requesting country’s perspective, not a negative or a viable excuse.”

Comment: I’m glad I don’t have Alex’s job. It sounds like Twitter gets a lot of heat from government officials who aren’t used to having people say no to them.

Some other discussions about this matter that I found interesting:

* Zeynep Tufekci, Why Twitter’s new policy is helpful for free-speech advocates

* Wired’s coverage. Check out Cindy Cohn’s quotes.

* WSJ’s coverage of Alex Macgillivray’s comments

*’s New Takedown Policy Thwarts Censorship