A recently released GAO report (pdf) concludes that as ISPs adopt usage-based pricing, it is increasingly important for consumers to understand how much bandwidth applications use so they can sensibly manage bandwidth consumption. This is a good thing. But the report holds ISPs responsible for providing that information and blames them for doing a poor job of it, which is ridiculous.
Participants in GAO-conducted focus groups expressed confusion about how to estimate their bandwidth usage, and the report uses the following chart to illustrate a source of that confusion. The chart shows that ISPs provide conflicting information about how much bandwidth an application type can be expected to use. Based on such confusing information, consumers may buy a higher- or lower-bandwidth service than fits their needs.
But how can the GAO expect an ISP to know how much bandwidth any application may use? The following figure shows the GAO chart amended with named applications as examples of high- and low-end ranges for each application type. The added data is based on NetForecast’s research into application bandwidth consumption (disclosure: I am an employee of NetForecast).
The conflicting ISP consumption numbers reflect the fact that, just as different household devices consume different amounts of energy and different cars average different gas mileage, different applications (e.g., streaming applications like Netflix and YouTube) have different bandwidth usage efficiency.
In the report, the GAO wrongly states that "differences [in consumption estimates] may be due to different technologies providers use." The underlying ISP technology (i.e., copper, fiber, or cable) has no effect on how applications operate, nor the speeds (in bits per second) at which they operate. NetForecast's research into actual speeds experienced by Netflix users, for example, found that they are primarily driven by the consumer's device and Netflix viewer choices (see How the Netflix ISP Speed Index Documents Netflix Congestion Problems [pdf]). Even when Netflix users experienced the highly publicized "poor performance" period in early 2014, it was caused by congestion at transit providers and not by the consumer broadband ISPs that are the subject of the GAO report.
Rather than blame ISPs for providing confusing information, the GAO should encourage application creators to provide information about how much bandwidth their applications consume. ISPs aren’t in the business of testing the bandwidth efficiency of applications. Asking them to do so would be like asking your power company to tell you how much energy each device in your household consumes, or asking Exxon or Shell to tell you the fuel efficiency of different makes and models of cars.