by Jared Cohen
We have often described the Internet as a “lawless” space, ungoverned and ungovernable by design. Its decentralized makeup and constantly mutating interlinking structure make government attempts to “control” it futile. But states have an enormous amount of power over the mechanics of the Internet in their own countries. Because states have power over the physical infrastructure connectivity requires—the transmission towers, the routers, the switches—they control the entry, exit and waypoints for Internet data. They can limit content, control what hardware people are allowed to use and even create separate Internets. States and citizens both gain power from connectivity, but not in the same manner. Empowerment for people comes from what they have access to, while states can derive power from their position as gatekeeper.
So far we have focused mostly on what will happen when billions more people come online—How will they use the Internet? What kinds of devices will they use? How will their lives change?—but we haven’t yet said what their Internet will look like, or how states will make the most of it in their own physical and virtual dealings with other states and with their own people. This will increasingly matter, as populations with different alphabets, interests and sets of norms become connected, and as their governments bring their own interests, grudges and resources to the table. Perhaps the most important question in ten years’ time won’t be if a society uses the Internet, but which version of it they use.
As more states adapt to having large portions of their populations online, they’ll strive to maintain control, both internally and on the world stage. Some states will emerge stronger—more secure and with greater influence—from this transition into the virtual age, benefiting from strong alliances and smart uses of digital power, while others will struggle just to keep up with and adapt to technological changes both domestically and internationally. Friendships, alliances and enmities between states will extend into the virtual world, adding a new and intriguing dimension to traditional statecraft. In many ways, the Internet could ultimately be seen as the realization of the classic international-relations theory of an anarchic, leaderless world. Here’s how we think states will respond to each other and to their citizens.
The
Balkanization of the Internet
As we said, every state and society in the world has its own laws, cultural norms and accepted behaviors. As billions of people come online in the next decade, many will discover a newfound independence—in ideas, speech and conversation—that will test these boundaries. Their governments, by contrast, would largely prefer that these users encounter a virtual world that allows the powers that be to mirror their physical control, an understandable if fundamentally naïve notion. Each state will attempt to regulate the Internet, and shape it in its own image. The impulse to project laws from the physical world into the virtual one is universal among states, from the most democratic to the most authoritarian. What states can’t build in reality they’ll try to fashion in virtual space, excluding those elements of society that they dislike, the content that contravenes laws and any potential threats they see.
The majority of the world’s Internet users encounter some form of censorship—also known by the euphemism “filtering”—but what that actually looks like depends on a country’s policies and its technological infrastructure. Not all or even most of that filtering is political censorship; progressive countries routinely block a modest number of sites, such as those featuring child pornography.
In some countries, there are several entry points for Internet connectivity, and a handful of private telecommunications companies control them (with some regulation). In others, there is only one entry point, a nationalized Internet service provider (ISP), through which all traffic flows. Filtering is relatively easy in the latter case, and more difficult in the former. Differences in infrastructure like these, combined with cultural particularities and objectives of filtering, account for the patchwork of systems around the world today.
In most countries, filtering is conducted at the ISP level. Typically, governments put restrictions on the gateway routers that connect the country and on DNS (domain name system) servers. This allows them to either block a website altogether (e.g., YouTube in Iran) or process web content through “deep-packet inspection.” With deep-packet inspection, special software allows the router to look inside the packets of data that pass through it and check for forbidden words, among other things (the use of sentiment-analysis software to screen out negative statements about politicians, for example), which it can then block. Neither technique is foolproof; users can access blocked sites with circumvention technologies like proxy servers (which trick the routers) or by using secure https encryption protocols (which enable private Internet communication that, at least in theory, cannot be read by anyone other than your computer and the website you are accessing), and deep-packet inspection rarely catches every instance of banned content. The most sophisticated censorship states invest a great deal of resources to build these systems, and then heavily penalize anyone who tries to get around them.
When technologists began to notice states regulating and projecting influence online, some warned against a “balkanization of the Internet,” whereby national filtering and other restrictions would transform what was once the global Internet into a connected series of nation-state networks.1 The World Wide Web would fracture and fragment, and soon there would be a “Russian Internet” and an “American Internet” and so on, all coexisting and sometimes overlapping but, in important ways, separate. Each state’s Internet would take on its national characteristics. Information would largely flow within countries but not across them, due to filtering, language or even just user preference. (Evidence shows that most users tend to stay within their own cultural spheres when online, less for reasons of censorship than because of shared language, common interest and convenience. The online experience can also be faster, as network caching, or temporarily storing content in a local data center, can greatly increase the access speed for users.) The process would at first be barely perceptible to users, but it would fossilize over time and ultimately remake the Internet.2
The first stage of this process, aggressive and distinctive filtering, is under way. It’s very likely that some version of the above scenario will occur, but the degree to which it does will greatly be determined by what happens in the next decade with newly connected states—which path they choose, whom they emulate and work together with, and what their guiding principles turn out to be. To expand on these variations, let’s look at a few different approaches to filtering in today’s world. We’ve identified at least three models: the blatant, the sheepish, and the politically and culturally acceptable.
First, the blatant: China is the world’s most active and enthusiastic filterer of information. Entire platforms that are hugely popular elsewhere in the world—Facebook, Tumblr, Twitter—are blocked by the Chinese government. Particular terms like “Falun Gong”—the name of the banned spiritual group in China associated with one flank of the opposition—are simply absent from the country’s virtual public space, victims of official censorship or widespread self-censorship. On the Chinese Internet, you would be unable to find information about politically sensitive topics like the Tiananmen Square protests, embarrassing information about the Chinese political leadership, the Tibetan rights movement and the Dalai Lama, or content related to human rights, political reform or sovereignty issues. When it comes to these topics, even some of the best-known Western media outlets fall victim to censorship. Bloomberg News was blocked in both English and Chinese following its June 2012 exposé on the vast family fortune of the then vice-president (and now president), Xi Jinping. Four months later, The New York Times experienced a similar fate after publishing a similar story about the then premier, Wen Jiabao. Unsurprisingly, information about censorship circumvention tools is also blocked. We learned how comprehensive and particular Chinese censorship authorities could be when, following a contentious trip by Google’
s executive chairman, Eric, to Beijing in 2011, all traces of his visit were wiped from the Chinese Internet, while media coverage of his trip remained accessible everywhere else.
To the average Chinese user, this censorship is seamless—without prior knowledge of events or ideas, it would appear that they never existed. Further complicating matters, the Chinese government is not above taking a more proactive approach to online content: one estimate in 2010 suggested that Chinese officials had hired nearly three hundred thousand “online commenters” to write posts praising their bosses, the government and the Communist Party. (This kind of activity is often called Astroturfing—i.e., fake grassroots participation—and is a popular tactic with public-relations firms, advertising agencies and election campaigns around the world.)
China’s leadership doesn’t hesitate to defend its strict censorship policies. In a white paper released in 2010, the government calls the Internet “a crystallization of human wisdom” but states that China’s “laws and regulations clearly prohibit the spread of information that contains contents subverting state power, undermining national unity [or] infringing upon national honor and interests.” The Great Firewall of China, as the collection of state blocking tools is known, is nothing less than the guardian of Chinese statehood: “Within Chinese territory the Internet is under the jurisdiction of Chinese sovereignty. The Internet sovereignty of China should be respected and protected.” This type of unabashed and unapologetic approach to censorship would naturally appeal to states with strong authoritarian streaks, as well as states with particularly impressionable or very homogenous populations (who would fear the incursion of outside information on an emotional level).
Next, there are the sheepish Internet filterers: Turkey has taken a much more subtle approach than China, and has even shown responsiveness to public demands for Internet freedom, but nevertheless its online censorship policies continue with considerable obfuscation. The Turkish government has had an uneasy relationship with an open Internet, being far more tolerant than some of its regional neighbors but much more restrictive than its European allies. It is impossible to get a completely unfiltered connection to the Internet in Turkey—an important distinction between Turkey and Western countries. YouTube was blocked by Turkish authorities for more than two years after the company refused to take down videos that officials claimed denigrated the country’s founder, Mustafa Kemal Atatürk. (In keeping with a 1951 law that criminalizes public insults to Atatürk, YouTube agreed to block the videos for the Turkish audience, but the government wanted them removed globally from the platform worldwide.) This ban was highly visible, but subsequent censorship has been more covert: Some eight thousand websites have been blocked in Turkey without public notice or official government confirmation.
The sheepish model is popular with governments that struggle to strike a balance between divergent beliefs, attitudes and concerns within their population. But by pursuing this path, the government itself can become the enemy if it goes too far, or if its machinations are exposed. To give a recent example from Turkey: In 2011, the government announced a new nationwide Internet filtering policy featuring a four-tier system of censorship, in which citizens would have to choose the level of filtering they wanted (from the most to least restrictive: “child,” “family,” “domestic” and “standard” levels). The Information and Communications Technologies Authority (known by its initials in Turkish as BTK) said the scheme was intended to protect minors and promised that people who chose the “standard” level would encounter no censorship. Many people skeptical of BTK’s record on transparency balked. In fact, the plan generated such an outcry among the population that thousands of people in more than thirty cities around Turkey took to the streets to protest the proposed changes.
Under pressure, the government dialed back its plan, ultimately instituting just two content filters—“child” and “family”—which users could adopt voluntarily. But the controversy didn’t end there. Media-freedom groups reported that their own tests of the censorship system revealed a more aggressive filtering framework than BTK would admit. In addition to the expected banned terms having to do with pornography or violent content, they found that ordinary news websites, content that was culturally liberal or Western (e.g., anything including the word “gay,” or information about evolution) and keywords related to the Kurdish minority were all blocked under the new system. Some activists argued that blocking information about Kurdish separatist organizations with the “child” filter was evidence of the state’s nefarious intent; the international media watchdog group Reporters Without Borders called the Turkish policy “backdoor censorship.”
The Turkish government responded to some of the public concerns about the new system. When a Turkish newspaper reported that educational websites about scientific evolution were blocked while content from a prominent Turkish creationist were not, the authorities eliminated the block immediately. But there is little to no transparency around what content is censored under these policies, so the government is forced to react only when such discrepancies are brought to light by citizens. The sheepish model of Internet filtering, then, combines a government’s ability to evade accountability with its willingness to take constructive action when pressure mounts. This approach would appeal to countries with growing civil societies but strong state institutions, or for governments without reliable bases of support but enough concentrated power to make such unilateral decisions.
The third approach, politically and culturally acceptable filtering, is employed by states as diverse as South Korea, Germany and Malaysia. This is limited and selective filtering around very specific content, based in law, with no attempt to hide the censorship or the motivations behind it. Outliers within the population might grumble, but the majority of citizens often agree with the filtering policies for reasons of security or public well-being. In South Korea, for example, the National Security Law expressly criminalizes public expressions of support for North Korea in both physical and virtual space. The South Korean government regularly filters Internet content affiliated with its northern neighbor—in 2010 it was reported that the government blocked some forty websites associated with or supportive of the North Korean regime, took down a dozen accounts with potential ties to Pyongyang on social-networking sites like Facebook and Twitter, and forced website administrators to delete more than forty thousand pro–North Korea blog posts.
Germany has strong anti-hate-speech laws that make Holocaust denial and neo-Nazi rhetoric illegal, and consequently the government blocks websites within Germany that express those views. And Malaysia, despite promising its citizens that it would never censor the Internet—going so far as to codify it in its Bill of Guarantees—abruptly blocked access to file-sharing sites like Megaupload and the Pirate Bay in 2011, claiming that the sites were in violation of another law, the country’s Copyright Act of 1987. In a statement, the Malaysian Communications and Multimedia Commission defended the move, writing, “Compliance with the law is not to be construed as censorship.” Many Malaysians disagreed, but the block remained politically and legally acceptable.
Of the three models, activists will pray that the third approach becomes the norm for states around the world, but this seems unlikely; only countries with highly engaged and informed populations will need to be this transparent and restrained. Since most governments will make such decisions before their citizens become fully connected, they will feel little incentive to proactively promote the kind of free and open Internet exhibited by countries in the “politically acceptable model.”
The trends we see today will continue in ways that are, for the most part, fairly predictable. All governments will feel as if they’re fighting a losing battle against an endlessly replicating and changing Internet, and balkanization will emerge as a popular mechanism to address this challenge. The next stage in the process for many states will be collective editing, states forming communities of interest to edit the web together, based on shared valu
es or geopolitics. Collective action—be it in the physical or virtual world—will be a logical move for many states that find they lack the resources, the reach or the capability to influence vast territories. And even with balkanization, cyberspace is still a lot of ground to cover, so just as some states leverage each other’s military resources to secure more physical ground, so too will states form alliances to control more virtual territory. For larger states, collaborations will legitimize their filtering efforts and deflect some unwanted attention (the “look, others are doing it too” excuse). For smaller states, alliances along these lines will be a low-cost way to curry favor with bigger players and simultaneously gain some useful technical skills and capacity that they might lack at home.
Collective editing may start with basic cultural agreements and shared antipathies among states, such as what religious minorities they dislike, how they view other parts of the world or what their cultural perspective is on historical figures like Vladimir Lenin, Mao Zedong or Mustafa Kemal Atatürk. In the online world, shared cultural and normative sensibilities create a gravitational pull among states, including those who might not otherwise have reason to band together. Larger states are less likely to engage in this than smaller ones—they already have the technical capabilities—so it will be a fleet of smaller states, pooling their resources, that will find this method useful. If some member countries in the Commonwealth of Independent States (CIS), an association of former Soviet states, became fed up with Moscow’s insistence on standardizing the Russian language across the region, they could join together to censor all Russian-language content from their national Internets and thus limit their citizens’ exposure to Russia altogether.