Second thoughts about that Wall St. Journal article on Iran and DPI

The censorship of Internet communications by the Iranian theocracy has been known for years. Months ago, a Freedom House study singled out Iran as one of the four worst enemies of Internet freedom. Yet a 22 June Wall Street Journal article got about 100 times more publicity than the Freedom House report, by making what turns out to be a spurious claim. Nokia-Siemens Networks is alleged to have sold the theocrats deep packet inspection (DPI) equipment that made it possible for them to, in the reporter’s words, “not only block communication but to monitor it to gather information about individuals, as well as alter it for disinformation purposes.”

The story was eaten up because it pushes hot buttons on both sides of the American political spectrum. For liberals and the left, the article fingered DPI technology, which many fear will be used to undermine if not destroy net neutrality. And to many in that camp, nothing could be more ideologically simpatico than to place some of the blame for the Iranian debacle on greedy capitalists. For conservative nationalists, on the other hand, the story hit an equally strong nerve. They tend to favor a hard-line foreign policy toward Iran, a charter member of the “axis of evil.” Their agenda is to isolate and demonize the Iranian government and, in a replay of the Cold War, push to cut off all trade and dialogue – if not to invade it outright. DPI becomes a proxy for nuclear weapons and a new kind of nonproliferation is advocated.

But even as the story was rippling through numerous email lists and blogs, I smelled something fishy about it. And so did others.

First, a reality check: anyone watching the Iranian crisis can see that Iran’s censorship is notable less for its use of powerful, airtight control technologies than for its leaks and gaps. Spilling out of Iran are videos taken from mobile phones, twitters, emails, broadcast television, and so on. Expat Iranians, such as a student here in Delft I spoke with today, are aware of information being circulated about servers and contact points that dissident Iranians can use to circumvent the censorship. If some diabolical new techniques are being used, they aren’t working very well. (What does seem to be working, alas, is killing, beating and jailing people on the street.)

Second, a more careful reading of the piece reveals that the reporter has fudged the issue of exactly what kind of “monitoring capability” Nokia-Siemens supplied and what it is used for. Like all major equipment manufacturers, NSN does sell network infrastructure with DPI capabilities, which it advertises as useful for “customer relations management” and “congestion management.” But it sells a lot of other equipment that doesn’t have this capability, or has similar capabilities not based on DPI. From Nokia itself, we have a flat, public, unambiguous denial: “Nokia Siemens Networks has not provided any deep packet inspection, web censorship or Internet filtering capability to Iran.”

Third, even if a DPI capability exists somewhere in Iran, the article provides nothing but vague insinuations about its role in Internet censorship. There is no proof that it forms the backbone of an all-encompassing surveillance and censorship program in Iran. Indeed there are numerous indications that Iran’s Internet censorship is relying on other, more basic methods: e.g., a national telecom monopoly that gives it the ability to slow or cut off bandwidth, regulations that require ISPs to use that monopoly, and more standard DNS or IP address-based blocking techniques. If one understands what DPI does the idea that Iranian bad guys sitting at consoles in a control room can “alter” packets in midstream “for disinformation purposes” sounds ridiculous. DPI works on the basis of pre-programmed decision rules and it is difficult to imagine how that capability would permit packets to be carriers of disinformation that would be meaningful in a specific situation.

But there’s something deeper amiss here. The overall effect of this article is actually to divert our attention from the more pressing issue of technical measures supporting Internet censorship. Internet censorship exists in the US and Europe as well as in Iran. Framing this as a kind of “trading with the enemy” issue and making a big deal about one specific technical measure, as if the DPI itself is responsible for the problem of dictatorship and censorship, is an unwelcome distraction. It treats the problem as if it were something other countries face, not us. The real issue here is the problem of Internet censorship by national governments, and the institutions and norms that support it. It’s the policy that matters, not just the technology that implements it. This policy of Internet censorship and surveillance is shared by a growing number of states, both democratic and authoritarian.

It’s interesting that the WSJ article didn’t draw links to, or question the morality of, the collaboration between AT&T and the National Security Agency in the warrantless wiretap program. Now that was an interesting and systematic use of packet inspection! Insofar as they are using DPI, the Iranian government draws on a technology and an economic demand that is honed in the West. Many other states are using software and hardware to filter or block internet content. Deep Packet Inspection is but one of these technologies, and not necessarily the most effective or damaging in all situations.

Most recently, Germany and Australia have implemented Internet blocking systems. In Belgium, copyright owners mounted a lawsuit to force an Internet service provider to install DPI to check for file-sharing (fortunately, they lost).

On the other hand, the cases of Burma, North Korea, and Cuba show that effective Internet censorship can rely on extremely primitive, nontechnological methods, such as disconnecting the entire country, or restricting internet access to a privileged few from the get-go. Conversely, the leaks in the Iranian firewall show that even a highly authoritarian state cannot rely completely on technological means to repress expression when an educated, middle class population avails itself of the power of widely diffused information technology. What makes censorship more or less effective is not just the technology, but the institutions that back them up and their popular legitimacy (which is why Australian censorship is failing and Germany’s may, too).

Censorship is a social institution, not a piece of hardware. You can’t buy it in a box. And you won’t stop it by publicly shaming businesses who sell a specific box to an evil government. There is a long and good debate to be had about the morality of trade in multiple-use technologies with evil states. But the larger, more practical point is that no single technology can, by itself, give a state the ability to effectively repress electronic expression. Likewise, no moral or legal constraint on technology vendors can, by itself, ensure freedom of expression. Only a pluralistic political system and a competitive market economy, coupled with broad popular support for the legitimacy of diverse, dissenting opinion and a readiness to actively resist repression — can do that.

If the West wants to make a moral statement about Iranian censorship, let it lead by example. Let it sweep away its own DPI installations when they are used to censor and block.


  1. Anonymous

    This was a well written intelligent article on both sides of the coin of censorship. We personally are for the free market trade of technology, but not to countries we believe will misuse this technology. This by itself is a form of censorship, but we are on the side that there truly is an “axis of evil” alive and well in some of these countries including Iran and North Korea.
    Admittedly though, when it happens on our soil, censorship of the internet, then we are offended. What faction of our government was responsible for “the pull” is a question, but we personally could cite several videos and articles on the internet that vanished without a trace within a very short time of web circulation.