Skip to main content

Expanding the Tubes

Chris Hoofnagle, adjunct professor of information | November 5, 2009

Inherent in the network neutrality debate is the interest in providing high quality of service for internet communications that are delay intolerant, such as VoIP and video. But could we provide high quality of service by simply increasing bandwidth?

There are many economic barriers to investing in architecture and capacity. A recent FCC presentation suggests an approach to improving speed competition among ISPs: get them to advertise with, well, less lying.

My home ISP advertises speeds up to 6 Mbps. Affiliates are even more strident, eliminating the “up to” language. But my frequent tests have never exceeded 3.5 Mbps, which is similar to Comscore’s findings included in the FCC’s presentation. You get about half of what is advertised.

Advertising a theoretical maximum speed is very appealing to consumers, it causes them to optimistically believe that they will enjoy that sweet speed. What if we changed the advertising rules to require disclosure of average speed, and consistent with advertising law principles, these average speeds would have to be substantiated? Would truth in advertising cause consumers to demand faster speed, and in the process, more investment in the network?

Comments to “Expanding the Tubes

  1. In reply to Mr. Sinatra’s comment, throwing bandwidth at the problem does not make the problem go away in this case. It follows the same model as advertised fuel mileage for a new vehicle, except that the auto industry gets its information from a fairly outdated standard provided by the government. The “theoretical” maximum bandwidth advertised is accurate in the fact that you can achieve these speeds in the best case scenario. Therefore, there is no false advertising from ISP’s.

    With that being said, I do believe that there is a serious lack of responsibility on behalf of these providers. These providers do have a responsibility to provide real-world information regarding up-time and average speeds. Since most internet providers require a contract, this would be the ideal place to put the average speed and up-time for their servers.

    It also seems that the internet and its capabilities grow at a much faster pace that the technology which carries it. If we continue to develop applications which require more bandwidth faster than the bandwidth can grow, then we will see bottlenecks and poor service. This is not a shortcoming of the ISP’s.

    I am responsible for a low-bandwidth site,, and I do not suffer from poor service on my servers or any reported poor service from my users. Youtube, on the other hand, provides some video which can be viewed in HD quality. I have attempted to watch HD videos there, and never see the video uninterrupted. If my average bandwidth from my high speed provider matched the advertised speed, this may not be the case.

    What we, as consumers, may need to be aware of is that the advertised speed of our connections my only be obtainable at 3:00AM.

    • You’re confusing the notion of “bandwidth” as a marketing term used by ISPs for “broadband” connections and “bandwidth” as an accurate measure of link capacity, which is how I am using the term. My point was that ISPs can generally invest more easily in actual bandwidth rather than in tools to restrict bandwidth (the latter of which are often why you see download “speeds” that don’t match the capacity of the link). But we have been taught by equipment vendors that doing something easy like investing in bandwidth somehow won’t work and that you need to use a lot of fancy traffic-engineering tools to make things work. In some cases, this is true, such as for large science data flows over long-distance links. But in the case of people who want to watch video on Youtube, throwing bandwidth at the problem does work.

      Of course, if ISP’s wish to create artificial scarcities, that’s another matter.

  2. For many years, network equipment vendors have been pitching quality-of-service mechanisms in favor of the “throw bandwidth at the problem” solution. That phrase, often uttered by network gear salespeople, is frequently followed up by various sorts of arguments as to why “throw bandwidth at the problem” is, was, and most importantly, will be an ineffective or overly-costly strategy for dealing with real-time network applications. The fact is, these arguments have been going on for years and, in general, “throw bandwidth at the problem” has worked just fine as a strategy. Moreover, it has turned out to be cheaper and easier when the maintenance and complexity issues surrounding QoS have been factored in.

    While I agree that there are serious problems with the impact that the strict interpretation of net neutrality as it applies to legitimate traffic-engineering and QoS, I also think that the network engineering community (and its suppliers) are paying too much attention to finessing the problem with QoS bells and whistles when brute-force bandwidth gives consumers better service all around.

Comments are closed.