Who’s controlling your corner of the Internet?

PRINCETON (US)—It’s a question most of us have asked, with growing annoyance, while we wait for a Web page to load: Why is my Internet connection so slow today?

A new global network of computers known as Measurement Lab will allow computer scientists to help answer that question by investigating how data moves across the Internet. The project could bring transparency to the debates over who should regulate that traffic.

“Many debates about the Internet and policy would benefit from more and better facts,” says Edward Felten, director of Princeton University’s Center for Information Technology Policy, during the lab’s launch event in January. “Measurement Lab provides that. It provides a facility for aggregating information and fostering a fact-based public debate about what’s going on and what government should do.”

Felten and Larry Peterson, chair of Princeton’s Department of Computer Science, spoke at the launch event, held at the Washington, D.C., headquarters of the New America Foundation, a public policy think tank that is sponsoring the initiative along with Google.

The project launched with three servers located in Google’s Mountain View, Calif., headquarters and will soon grow to 36 servers in 12 locations around the world. Under Peterson’s leadership, Princeton will serve as a key hub.

In addition to Princeton, Google and the New America Foundation, the consortium of organizations involved in the project includes a number of other universities and computer science research centers. Project organizers hope the network will grow far larger with time, as more organizations join the consortium.

The lab will allow researchers to test Internet connections for speed and other functionality by sending data between network computers. Among other things, the computers can measure precisely when data leaves one computer and when it arrives at another, providing a sensitive gauge for Internet speed.

The project Web site offers several tools, two of which were built for use by members of the public to test their Internet connections.

The Network Diagnostic Tool measures a computer’s upload and download speeds and determines whether problems lie with the individual computer’s configuration or somewhere beyond in the networking infrastructure. Another tool, Glasnost, attempts to determine if an Internet service provider, such as Comcast or Verizon, is specifically controlling the data-transfer speed of programs such as BitTorrent, a popular file-sharing program.

In 2007, customers complained that their download speeds when using BitTorrent slowed dramatically, and Comcast later acknowledged that it was specifically limiting the speed with which BitTorrent could transfer files.

The incident fueled an ongoing debate about whether Comcast and other service providers should be allowed to control the access certain computers or programs have to the Internet. The service providers have argued that they have to manage the traffic effectively to keep data flowing smoothly, and that putting limits on traffic from one source is necessary.

Others have advocated for what they call “net neutrality” and charge that service providers currently have too much control over Internet traffic. They argue that allowing the providers to limit access to the network as they see fit could have a chilling effect on innovation and free-market competition.

“If we’ve learned anything at Google, and maybe we can say even anything about the Internet in general, it’s that an open and accessible network induces and invites creativity and innovation,” says Vint Cerf, a vice president of Google.

Cerf, who helped build the computer network that became the Internet and later developed the first commercial e-mail software, says the new network will make it possible to accurately determine whether information is flowing freely through the Internet.

Whatever side of the debate you’re on, says Felten, Measurement Lab should provide much needed transparency. He notes that the policy discussions over Comcast’s decision to curb BitTorrent traffic were often based on fuzzy details.

“It really was a process that was starved for strong factual evidence,” he explains. “There were assertions about what was happening, but there were not a lot of really well established consensus facts about what was happening.”

He predicts that the project would have a global reach. “There are perhaps even greater opportunities and potential benefits from a project like Measurement Lab in the international setting,” he adds, “where . . . different national telecommunications and Internet authorities are doing different things. It would certainly be helpful for people around the world to know what’s happening in their corner of the Internet.”

For the latest Princeton news, visit www.princeton.edu/main/news.