Ever since ISPs like Comcast, Time Warner, and Charter began imposing monthly data caps on their Internet customers, there’s been criticism that so-called “data hogs” — the folks who are the top consumers of data — aren’t actually responsible for much of the potential network congestion issues faced by Internet providers.
The idea goes like this: While there are certainly some broadband users who download far more data than others, their use tends to be relatively consistent, and may even peak during the ISP’s most-idle hours in the early morning. Conversely, ISPS tend to face their biggest traffic loads in the late afternoon through about 11PM, as home Internet users come home, check their email, romp on Facebook, and maybe watch or download some high-definition streaming video from the likes of Netflix, Amazon, YouTube, or Apple. In those times, the “data hogs” usually aren’t any more “piggish” than anyone else. ISPs’ network congestion, then, mostly stems from the swarms of everyday users jumping online to do ordinary, everyday things.
The idea that usage caps are a blunt, inaccurate tool to control network congestion is not new. However, the CEO of the mid-sized ISP Sonic.net has recently thrown a new notion into the mix: what if those data caps are about protecting ISP’s television businesses?
The idea behind data caps
From an ISP’s point of view, data caps are usually presented as a way to prevent over-zealous users from consuming an unfair share of network bandwidth. As the capacity of broadband connections to the home has grown, individual home users are now capable of subscribing to Internet services that would have been in the realm of commercial operations and even ISPs just a decade ago. ISPs don’t necessarily have the back-end network and peering capacity to deliver 100 percent of the bandwidth sold to their subscribers at any given time. While ISPs work to build out their infrastructure, they rely on the fact that not every customer is going to be online all the time, demanding their full allocation of bandwidth. ISPs might not be able to give all their customers all their allocated bandwidth all the time, but ISPs gamble that being able to give most of their customers most of their allocated bandwidth most of the time ought to be good enough. After all, most Internet users don’t have the technical background to reliably measure the bandwidth their receiving from their ISP at any given moment.
[Source: Google Measurement Lab]
The idea that bandwidth is a limited, precious commodity at ISPs is what leads to data caps: They’re a form of rationing. Imagine a community experiencing a famine. If each family was rationed one bag of rice a week, it wouldn’t tolerate a single individual walking off with five bags of rice every day. That’s how ISPs characterize so-called “data hogs:” consuming an unfair share of bandwidth, to the detriment of other users on the network. Take more than your fair share, you’ll be punished — either by being required to pay more for service, or by having service limited or terminated.
Who do data caps punish?
To challenge the notion that so-called “data hogs” are slow things down for everyone else, industry analyst Benoit Felten issued a challenge a couple years ago on his personal Fiberevolution blog (to separate it from his professional work as an analyst): If ISPs were willing to provide him with a particular dataset about their usage patterns, he would conduct a full analysis on it for free, and publish the results regardless of whether they backed up his personal assertion that heavy data users were major contributors to network congestion.
Felten carries some clout in the industry — he’s not some random blogger with a weird notion. Nonetheless, it surprised no one when ISPs failed to clamber to Felton’s doorstep and offer up exactly the data Felton wanted. After all, if he was right, there would be public data analysis refuting the fairness of ISP data caps. However, a single ISP did answer the challenge: mid-sized California ISP Sonic.net. CEO Dane Jasper has been saying for years that data caps aren’t necessary to manage Internet usage, and saw Felten’s call for data as a way to put his money where his mouth was. And, while it’s based on data from only a single ISP, Felton’s analysis (published by Diffraction Analysis and summarized on his blog) is telling. While the top 1 percent of data customers do account for 20 percent of overall bandwidth consumption, they represent only 14.3 percent of the top 1 percent of data users during peak hours.
In other words, over 85 percent of bandwidth consumption during peak hours did not come from the heaviest data users. If Sonic.net’s data can be generalized to other broadband providers — and, of course, there’s no way to know for sure — it would mean that data caps do a very poor job of punishing people for causing network congestion.
The threat of Internet video
So why are ISPs so interested in data caps? One possible answer is suggested by Sonic.net’s Dane Jasper: TV.
Most of the United States’ leading ISPs (Comcast, Time Warner, Verizon, AT&T, Charter, Cablevision) also offer television and voice services. These so-called “triple plays” bundle phone, television, and Internet together onto a single bill. While all three services generate significant revenue for these companies, the video portion is the most expensive to operate, thanks in part to the myriad of regulations surrounding the paid-television industry but also the complex licensing involved in carrying cable channel packages. That’s why, in a typical “triple play” bill, cable television service is often the greatest portion of the fee (depending on channel package).
The problem, of course, is that consumers are increasingly turning to the Internet for their video entrainment, opting for so-called “Over The Top” (OTT) solutions like Netflix, Amazon Video-on-Demand, and Hulu over broadcast television, cable service, and television operators’ video-on-demand services. Increasingly, consumers are considering these companies to be broadband providers, rather than television providers, so they’re tending to cancel cable services while signing up for broadband.
This is generally a bad trend for paid television providers. Instead of collecting money from customers who subscribe to television packages, the money goes to the likes of Netflix. From the network operators’ point of view, they took all the trouble to build out their network, bring broadband to people’s homes, and get everything operating, and now OTT video providers are swooping in and threatening to break off one leg of their three-legged revenue stool. (Another leg, voice services, is under similar assault from VoIP solutions like Skype, Google Voice, and Vonage.) That’s one reason why big ISPs like Comcast are looking to charge services like Netflix usage fees for accessing their customers.
One way television providers can dissuade customers from canceling their television service is with Internet usage caps. Data published this January by research firm Nielsen (PDF) finds that, on average, Americans watch 32 hours and 47 minutes of television every week. That converts to over 4 hours and 40 minutes every day. And that’s an average: While some consumers watch considerably less television, an equal number watch considerably more.
Now imagine if that television were all shifted to OTT providers like Netflix, Hulu, Amazon Video-on-Demand, and others. Netflix’s top high-definition streams run at about 4.8Mbps, or about 36 MB per minute. Multiply that across the average television consumption of 4 hours and 40 minutes a day, that’s a little over 10.1 GB of data consumption per day. Take a typical 30-day month, and average consumers come in at just over 300GB of bandwidth utilization just for television alone.
Many ISPs begin capping data at 250GB per month; some cap data usage at lower levels (often corresponding to lower-priced broadband packages); some have higher caps.
Granted, not all customers are going to pull in reasonably high-quality high-definition content, so their bandwidth utilization will be lower. But the point stands: If typical television watchers were to shift their habits to broadband viewing, they would suddenly find data caps very limiting.
The dirty secret
ISPs imposing data caps claim that “data hogs” — loosely, customers pulling down more than 250GB a month in data — are disproportionately consuming network resources, causing congestion, and impacting Internet performance for other customers. However, it’s important to remember that broadband providers like Comcast, Time Warner, AT&T U-Verse, Verizon FiOS, and others are delivering their television content to their customers via the same IP network infrastructure that supports their broadband Internet service.
If a television customer watches the average four hours and forty minutes of programming a day in high definition (or, perhaps, has multiple televisions or DVRs accessing standard-definition content) they are already consuming bandwidth in the same ballpark as data usage the providers categorize as abusive. Of course, television and broadband Internet access are separate services for which providers bill separately, but from the point of view of network congestion, it’s all just data going down the same pipes. In general terms, average television users utilize the provider’s network as heavily as “data hogs.”
With the FCC’s proposed regulatory framework for network neutrality being challenged by the likes of Verizon, there is very little regulatory framework that can be applied to ISPs or backbone providers management of their networks. For now, they’re essentially free to do what they want, whether that be blocking particular applications (like file sharing) or imposing data caps.
However, as consumers and governments increasingly look upon Internet access as a right, rather than a privilege, ISPs may find it difficult to justify limiting or severing Internet service for simply for being used. Few question that ISPs need to be able to manage their networks, both in general terms and on a real-time basis. However, it’s increasingly apparent data caps are an artificial policy that seems based on broadband operators’ pocketbooks, rather than their network capabilities. Hopefully, responsible ISPs will develop new polices that more transparently manage network loads, such as limiting top-bandwidth consumers only during periods of peak network loads, regardless of how much bandwidth they may consume in an hour, day, or month.