Study finds fatal flaw in T-Mobile’s Binge On scheme that can affect non-video content
T-Mobile’s zero-rated video streaming scheme, Binge On, has been a controversial one. It recently hasn’t been, but the debate over its conformity to net neutrality may get renewed with research from Northeastern University.
Scholars reverse-engineered how the program worked and published their findings in a report called “BingeOn Under the Microscope: Understanding T-Mobile’s Zero-Rating Implementation”. The paper splayed the program’s nuanced policies out, many of which were not publicized through the study period.
“They are available now, but much remains largely hidden to the average content provider and subscriber,” said David Choffnes, an assistant professor at Northeastern and co-author of the report. “Both can be misled.”
The text and numbers reiterate what we already know about the program — that all content labeled as a video stream gets throttled to 1.5Mbps with Binge On active and participating providers’ content is free to stream while all other content is charged for. That’s a discussion that could be had at the FCC and the commission did talk with T-Mobile about the scheme a couple months ago.
But there’s another wrinkle to how T-Mobile reads content as “video” and decides whether or not to throttle and zero-rate the traffic, throttle and charge for the traffic or not throttle it at all. Instead of tracing hosting IPs, the program looks for host names to match against the Binge On participants list and then reads any strings in metadata that signifies what type of content it’s about to deliver.
“Once T-Mobile successfully matches a BingeOn-specific string in its classifier, it ignores all other fields that might support or contradict the classification,” the report stated.
The revelation opens up possibilities for scofflaws to spoof host and content type labels to help or hurt T-Mobile customers. And with a simple detection system, the content type parameters might incorrectly categorize and act upon non-video content as well. The detection mechanism and exploit are also found in use for the Music Freedom program, too, though speed-limiting doesn’t take effect in this case.
The study’s co-authors have reached out to T-Mobile about its findings. It’ll be interesting to see if and how John Legere responds.