Public Release: 

Northeastern researchers find T-Mobile's Binge On doesn't live up to the hype

Northeastern University

New research from Northeastern University shows that what T-Mobile promises regarding its Binge On service is not what subscribers, or con­tent providers, may actu­ally get. In many cases, subscribers were left with lower quality videos and unexpected charges.

"At the time of our study, impor­tant details about the Binge On policy were not in public doc­u­ments," says researcher David Choffnes. "They are avail­able now, but much remains largely hidden to the average con­tent provider and sub­scriber. Both can be misled."

Choffnes' team con­ducted the research in Feb­ruary and March, and the paper was accepted in May by the IGCOMM Internet-QoE work­shop. He has shared the paper with the Fed­eral Com­mu­ni­ca­tions Com­mis­sion to help inform its inves­ti­ga­tion of T-Mobile's com­pli­ance with the Open Internet Order, passed a year ago.

To ana­lyze Binge On, the researchers "reverse engi­neered" how T-Mobile imple­mented its poli­cies. "We set out to learn exactly how Binge On works, and we com­pared what we found with its stated poli­cies," says Choffnes. "There were sig­nif­i­cant dif­fer­ences between the two."

Con­sider: Binge On touts its use of "zero rating," which means the ser­vice doesn't count your streaming internet usage against your data plan as long as the videos you're viewing are from providers who par­tic­i­pate in Binge On.

That qual­i­fier, says Choffnes, presents a sig­nif­i­cant danger to the FCC's Open Internet Order. The order's cen­tral con­cept is "net neu­trality," which means that, with few excep­tions, ISPs must treat all internet traffic and appli­ca­tions the same. For an ISP to slow down one provider's video but not another's, says Choffnes, puts the former at a com­pet­i­tive disadvantage.

"The internet has been hugely suc­cessful because it enables inno­va­tion, where all new internet appli­ca­tions receive the same net­work ser­vice as incumbents--it's a level playing field," says Choffnes. "T-Mobile's policy gives spe­cial treat­ment to video providers that work with them. What if every ISP did this, but in a dif­ferent way? In such a world, the next Net­flix, Hulu, or Pied Piper might never get off the ground because keeping up with ISPs and their poli­cies would leave them chasing their tails."

Choffnes' find­ings also call into ques­tion T-Mobile's claims that it's up to sub­scribers to decide whether to enable Binge On on their plans. "Sub­scribers are opted in by default," he says, and he ques­tions whether that's what they expect. Being opted in by default means means that to stream videos at high res­o­lu­tion, sub­scribers have to be savvy enough to drill down and figure out how to dis­able Binge On.

And then there's the chal­lenge to con­tent providers. "Our research showed that if a video provider does nothing, that is, nei­ther opts into or out of par­tic­i­pating with Binge On, its video traffic to T-Mobile sub­scribers who use Binge On will be given reduced band­width, but the sub­scribers will still be charged for the streaming," he says.

This prac­tice of lim­iting the band­width avail­able to an application--which in this case leads to lower res­o­lu­tion video--is called "throt­tling." The Open Internet Order con­tains a clause that expressly for­bids throt­tling. In the case of T-Mobile, the lim­iting of band­width occurs without sub­scribers or video pro­ducers explic­itly agreeing to opt in to the ser­vice, Choffnes found.

After con­cerns regarding net neu­trality and free choice, the ques­tion for con­sumers becomes: Is the reduc­tion of video quality worth the free streaming?

"When Binge On is enabled, it can have a very large impact on the quality of the video that you're watching," says Choffnes.

T-Mobile says that the res­o­lu­tion for Binge On streaming is 480p (pro­gres­sive scan) or better, which is con­sid­ered stan­dard for DVD movies. How­ever, the researchers did not find evi­dence to back up these claims. In their trials using YouTube, the res­o­lu­tion was only 360p, notice­ably blurry on a modern smartphone.

On the other hand, when they turned off Binge On, they achieved a whop­ping 1080p, or full HD res­o­lu­tion. For its part, T-Mobile claims that 480p or better is pos­sible but cur­rently requires a spe­cific Android device that sup­ports it and works only for cer­tain video providers.

Finally, the researchers ques­tioned the ability of T-Mobile to track video streaming at all, given that con­tent providers are not required to iden­tify the kind of con­tent they send over the internet. In one case, the researchers iden­ti­fied a video provider that T-Mobile had incor­rectly labeled, and thus was enjoying high-quality streaming by default, unlike other providers.

The scary part, says Choffnes, is that the reverse is also true. "T-Mobile's detec­tion methods are very simple, so there's no way they can always be right," he says. "That means that Binge On is likely slowing down traffic that is not video. This raises serious con­cerns about com­pli­ance with the Open Internet Order."

Those simple methods open the door to exploita­tion as well, allowing sub­scribers to get free data even for non-video con­tent. The researchers devel­oped simple soft­ware that manip­u­lates internet traffic so that it looks like video. For example, it makes any web content--web pages, app down­loads, and photos--look like YouTube traffic. "We real­ized we could make any net­work traffic zero rated by just putting the right text in the right place," says Choffnes. "That is a secu­rity vulnerability--it's poten­tially an open cash reg­ister that people can take from."

The researchers told T-Mobile about the flaw and have helped the com­pany repro­duce it so it can try to address it. Choffnes, for one, is not opti­mistic that the free-riding vul­ner­a­bility can be fixed per­ma­nently. "We believe that it is fun­da­men­tally intractable," he says.


Disclaimer: AAAS and EurekAlert! are not responsible for the accuracy of news releases posted to EurekAlert! by contributing institutions or for the use of any information through the EurekAlert system.