Archive for the 'control & power' Category

Prior Authority on Prioritization


In anticipation of the FCC hearing on Monday, I want to draw everybody’s attention to some of the prior authority on ISP prioritization. In In the Matter of Madison River Communications (20 F.C.C. Rcd 4295 (2005)), the FCC’s Enforcement Bureau investigated Madison River Communications for allegedly blocking ports used for VOIP applications. The practice hampered subscribers’ ability to use certain VOIP services, much like Comcast’s de-prioritization of p2p services has limited subscribers’ ability to use those services. The FCC believed that Madison River’s practice violated section 201(B) of the Communications Act of 1934. Madison River ultimately agreed to a Consent Decree, which prohibited the blocking of ports used for VOIP applications. It also barred Madison River from “otherwise prevent[ing] customers from using VOIP applications.” Also in 2005, the FCC issued a Policy Statement that most people interpret to endorse Network Neutrality…). It included four Internet Principles, all of which were intended to “preserve and promote the open and interconnected nature of the Internet.” Madison River and the FCC Policy Statement seem to prevent filtering based on the type of transmission and is consequently encouraging for anyone who thinks p2p services offer legitimate legal functions and should be widely available to Internet users.

Wikileaks litigation


nbsp;, a site that allows whistleblowers to anonymously post confidential documents, has been taken offline by court order in California. The BBC article states that a case brought by Swiss bank Julius Baer resulted in the judgment along the following lines:

… the main site was taken offline after the court ordered that Dynadot, which controls the site’s domain name, should remove all traces of wikileaks from its servers.

The court also ordered that Dynadot should “prevent the domain name from resolving to the website or any other website or server other than a blank park page, until further order of this Court.”

Other orders included that the domain name be locked “to prevent transfer of the domain name to a different domain registrar” to prevent changes being made to the site.

The site is apparently still available in other countries such as Belgium and India, though I think these must be localised versions. The site claims they couldn’t defend themselves at the hearing because they were only given hours notice.

BBC news story

EDIT: See Kparker’s comment below explaining that the site has just been removed from the DNS so it is still available if you know the IP (it is hosted in Sweden).

Are other ISPs doing what Comcast does?


In light of the scandal regarding Comcast’s blocking (or “deprioritization”) of certain p2p applications, I thought it might be interesting to investigate the disclosures of other ISPs regarding prioritization. Here is what I found. (It is worth noting that some of this may change very soon in light of Comcast’s troubles.)

Most telecomm and cable companies do not publicly provide much information regarding their prioritization techniques. Time Warner Cable is one of the most forthright companies in this respect. In its Operator Acceptable Use Policy…), the company gives some insight into its approach to prioritization. The policy explains that Time Warner Cable “may use various tools and techniques in order to efficiently manage its networks and to ensure compliance with its Acceptable Use Policy.” Such tools may include “limiting the number of peer-to-peer sessions a user may conduct at one time” and “limiting the aggregate bandwidth available for certain usage protocols such as peer-to-peer and newsgroups.” These statements indicate that Time Warner gives p2p applications low priority relative to other types of data packets.

Other ISPs offer less information. Verizon promises prospective high-speed internet service subscribers “a dedicated connection to the Verizon central office so that you don’t have to share your local access connection with other users”…) Nonetheless, Verizon acknowledges that upstream congestion may hinder connection speed. Verizon also gives the vague explanation that “other factors” may influence connection speed. An analogous set of representations appears in relation to FIOS, Verizon’s fiber optic broadband internet service…). AT&T does not admit to any degree of prioritization. The AT&T website cites only “heavy Internet traffic, the condition of your telephone lines, and the distance of your home to the telephone company’s central switching station” as factors that may affect download speed…)

Three strikes and you’re out for illegal downloading in the UK


I have to confess, I’m not a fan of baseball, nor laws based upon its principles. A green paper in the UK (released by the government showing policy direction) has suggested that ISPs are going to be required to cancel connections of persistent copyright infringers. Your first offence gets you an email, your second a suspension and the third gets you kicked off your ISP. The details are as yet unclear, such as how disputes are to be settled and whether ISPs will share information about infringers, but it sounds to me like one of the most extensive and intrusive government efforts to regulate online behaviour I’ve come across. Lots of questions remain unanswered, such as the length of suspensions and the outright bans. One thought that occurs to me is that if this applies to University ISPs the effect could be huge, with no alternative provider in the University accommodation I stayed in for 4 years it really would have been 3 strikes and you’re out. The problems of erroneous reporting and detection would also be huge, if not insurmountable. All in all, its just not cricket.

Times Online article

Beijing Olympics Video


An addendum to yesterday’s class notes – this is the link to the video that JP mentioned regarding the Beijing Olympics (to which I have tickets!):….

Class 3 Blog Post


Despite the fact that I volunteered to be the first class blogger, Professor Palfrey began the class by maligning for my NY Giants shirt. Evidently reluctant to linger on a sore subject, he was eager to move on to Internet filtering.

A. Palfrey began the Internet filtering discussion with a chalkboard exercise using Professor Benkler’s layers of the Internet as a frame. The layers include:

1. Content (e.g. this blog post)
2. Applications (e.g. Microsoft Word)
3. Logic Layer (e.g. TCP/IP and other standards)
4. Infrastructure (hardware)

B. For more on Benkler’s layers, see Yochai Benkler, From Consumers to Users: Shifting the Structures of Regulation Toward Sustainable Commons and User Access, 52 FED. COMM. L.J. 561 (2000), available at….

C. Palfrey then moved on to explain the inspiration for the Access Denied project. A number of scholars, Palfrey, Zittrain, and Business School Professor Ben Edeleman among them, set out to determine whether Internet users could access the same Internet from anywhere in the world. They tested the notion through a number of case studies, most notably Saudi Arabia and China:

1. Saudi Arabia – the Saudi authorities were reluctant to bring the Internet to their populace. They maintain a website that contains information regarding the state’s filtering program. It includes a mechanism by which Saudi citizens may request that a blocked site be made available. The Saudis were initially cooperative with the Access Denied project and blocked primarily pornography.

2. China – much less forthcoming. There is no analogous site that explains the state’s filtering policies or any process for requesting that a given site has been blocked unnecessarily.

3. With a smooth segue back to the Internet layers, Palfrey explained that the while both the Saudis and the Chinese filtered at low down levels of the Infrastructure and Logic layers, the Chinese also engage in higher level filtering at the Application and Content layers. They enlisted the assistance of the ISPs, OSPs, and search engines to engage in filtering.

a) Palfrey then brought up the Open Net Initiative Google China Search Comparison site and ran a number of search comparisons, including “human rights” and “Tiananmen Square.”’s search results page states that “all results are not here.” The URL for the ONI Google Search is

B. Turkey: A New Case Study?

1. Professor Palfrey is on his way to Turkey this evening to attend a number of meetings on Internet filtering. When Palfrey asked the members of the class who they expected to be involved in the discussions, people suggested not only major ISPs and application providers from both Turkey and the U.S., but also religious leaders and academics.

2. As Palfrey explains, in many countries, there are disputes between the Minister of Economic Development, who is generally in favor of an open Internet, and the Minister of Telecommunications, who is generally in favor of filtering on the basis of protectionism.

C. Timed Censorship Strategies – Recently, Palfrey and others have observed strategically timed censorship efforts in the context of elections. The preferred method of timed censorship strategies is overloading. Overloading the censored sites is an ingenious methods of censorship because it is difficult to distinguish from high Internet traffic.

D. Wikipedia in China – Why might Wikipedia be blocked and unblocked three times? It’s clear why it got blocked initially. Wikipedia is potentially dangerous for the Chinese government. But why would they unblock it? They probably realized the beauty of Wikipedia – that anyone can edit it at any time. As a result, the Chinese government may have figured they could win the propaganda war against Wikipedia users. Why block it again? Other countries (ex. Japan) may be dedicating the same resources to the editing war. And so the cycle continues.

E. Internet and Gender – In the last few minutes, Professor Palfrey turned to the question of gender differences on the web, in the context of the Pew survey. Understandably, many students had issues with the survey methodology, most notably the potential inaccuracy of a self-reporting survey method.


A cost of an end-to-end network


Slashdot is discussing the discovery that the Snopes site — a popular and trusted urban-myth buster — installs adware on the machines of the unwary. Here’s how one commenter (patio11) explains it:

A quick primer in online advertising, for those of you who block it:

At one end of the chain, we have Content Provider A. At the other end of the chain, we have Service Provider Z. Z wants to place advertising on A’s site but, importantly, doesn’t know how to do it, doesn’t generally know specifically who A is, and needs this to scale to potentially thousands of As. This is where participants B, C, D, E, F, Google, H… etc come in. There are advertising aggregators, affiliate networks, affiliates, affiliates of affiliates, affiliates of affilates of networks of affiliates who subdivide the advertising market into smaller and smaller slices before it finally gets on A’s site.

Now, somewhere in the chain, let us inject one person who is less than scrupulous. He doesn’t work at Snopes — this would tarnish a brand for a week’s worth of income, not a smart play. He probably has a steady stream of relationships with each of the numerous advertising concerns on the Internet, picking up and moving from one after he has collected a check or three and then had the banstick for TOS violations catch up with him. He is the one working for, most probably, affiliate of an affiliate of an affiliate of Zango.

This is the way most malware makes its way onto ad networks and, from there, onto high-trust sites. Volokh Conspiracy, one of my favorite blogs, had a nasty browser hijacker which affected non-US users for months before their advertising network caught wind of it. A few popular MMORPG sites have ended up hosting keyloggers in the same fashion. It is an unintended consequence of a system without central control — much like the Internet itself, actually. (The system being split up this way does have its advantages, for both endpoints of the chain and for everybody between. Google’s business model is based on snapping the chain and replacing it with a big cloud labeled Gooooooogle, but they’re not yet the only game in town.)