T-Mobile’s “Binge On” service: beautiful or beastly?

There is a media and regulatory battle on between T-Mobile USA and ‘net neutrality’ campaigners over the new Binge On service for video. In the absence of a market price mechanism for resources, there must be rationing. Binge On is merely an improved form of rationing compared to the status quo.

The basic facts about the service

The Binge On service constrains all video (streamed and downloaded) to 1.5Mbps. This is enough for DVD-like quality. According to TechnoBuffalo, as a result “customers should consume up to three times as much content (not just video, but photos, website data) with data optimized video [than before].” Ars Technica has data showing a >50% volume reduction.

The benefit to retail users is that data from nearly 40 sources is zero-rated against data caps. The service is optional, but is opt-out. It is only available to users on at least a 6Gb data plan, which is the lowest one currently on offer.

For video content providers there is no charge to become a partner who is zero-rated, and anyone is apparently welcome to join the programme. There are technical integration requirements to meet to reliably identify the data to be zero-rated.

Google has publicly complained that YouTube videos are being “throttled” since they do not meet those technical requirements, and have not joined as a partner. The Electronic Frontier Foundation (EFF) is apoplectic at the multiple perceived “neutrality violations” (both technical and commercial).

Technical consequences of Binge On

T-Mobile is both limiting the load on its network, and retargeting its resources to different users and uses. All things being equal, the overall offered load will significantly decrease. As the zero rating means things aren’t equal, price elasticity means some of that decrease will be undone by increased video viewing. We’ve only 24 hours in a day, so that effect is likely to be moderate.

As a result of the lower load there will be less contention for resources, as the network will be have more idleness. If T-Mobile engineer their traffic management well, the network will also appear to offer a more consistent level of performance: video streams will see a “virtual network” held steadily at roughly the same quality.

The restriction to 1.5mbps means adaptive video codecs will move to low resolution. Non-adaptive higher-definition codecs will buffer or fail. (If they’re making rigid resource assumptions for a ‘best effort’ mobile service, what do they expect?) The difference in resolution is a small loss in value. For comparison, Amazon here in the UK charges £6.99 for a standard definition Harry Potter movie download, and £7.99 for high definition.

The effect of this decreased load will in the short term means a quality of experience (QoE) increase for all other users and applications. It is possible that the low-resolution video will also have more predictable performance than before. The mere act of “throttling” cannot be assumed to increase buffering events given the overall systemic changes.

In the long term T-Mobile may choose to run its network ‘hotter’, decreasing QoE. Given the market is reasonably competitive, this would presumably result in lower prices and/or allow more people to experience the benefits of their service.

Commercial consequences of Binge On

T-Mobile has looked at the nature of demand and how it is allocating supply to meet that demand. Their assessment is that the “resource trade” away from high-res video to everything else is a customer experience “profit”. The QoE improvements for non-video applications outweigh the disappointment of low-res video. (Remember in the 1990s when we all switched from VHS to DVDs and felt they were “wow!” quality. Me too…)

By restricting video quality, rather than delivering maximum possible performance to all users at all times, T-Mobile has set itself up for higher customer satisfaction. Why so? Humans suffer from a cognitive bias called the “endowment effect”. We value anything we already have too much, and therefore feel great pain when it is taken away. By not delivering high-resolution video in the first place, nobody will experience its loss as network loads rise. It is typically better to deliver consistent performance than swinging between “wow!” and “yuk!”.

They have also positioned themselves in an interesting position on the network interconnect side. They are now able to ask these partners for cash if they want to deliver high-resolution video. I suspect that the significant extra resources required (i.e. most of the overall network capacity) aren’t worth the benefit, so nobody would pay. That means T-Mobile has eliminated a fundamentally unsustainable cost structure of delivering “best of best effort” to everyone all the time on licensed spectrum.

So, is Binge On beautiful or beastly?

Networks are finite resources, and there are only two possible ways of allocating those resources:

  • Either there is a market for a “quantity of quality” and the price mechanism decides who really values it, or
  • There is rationing of resources by some decision process imposed on users.

There is no third option. Pick one of the above.

Historically the rationing has been done by allowing the emergent processes of packet networks to arbitrarily allocate the disappointment when demand exceeds supply. This can be both unfair and inefficient, and sets up perverse incentives. This is demonstrated by T-Mobile’s network being flooded with unneeded fidelity of whiskers on cat videos, at a QoE cost to every other application.

The T-Mobile user base as a whole is likely to be substantially better off due to this service. Whilst it is still a rationing scheme, Binge On is one that is far more responsive to demand. This should be a welcome development. Users are likely to vote with their feet in favour of T-Mobile, and that should be plenty enough to tell you this scheme is a legitimate one.

Google is not entitled to any performance level whatsoever. There is no quality floor for “best effort” broadband. They have no grounds for complaint, since they have not paid for any performance. Indeed, their greed is palpable: they assert that their data-hungry video ads business model should be sustained regardless of its QoE externality.

The EFF is attempting to perpetuate the myth of ‘non-discriminatory’ and ‘neutral’ networks, with the implicit belief in benevolent packet pixies making good resource allocation choices. T-Mobile’s actions are a rational response to the inevitable collapse of that belief system.

Binge On is far from beautiful: that attractiveness would require a proper wholesale interconnect market for performance, with rational pricing of quality. That said, it’s a lot less ugly than a futile attempt to sustain the historical status quo of the pre-video era. So overall I’d say Binge On is pretty good for everyone.

For the latest fresh thinking on telecommunications, please sign up for the free Geddes newsletter.