A newly released GigaOm whitepaper by Muaayyad Al-Chalabi talks about an issue near and dear to our hearts here in Central Oregon – the trend by ISPs toward bandwidth-based pricing. Bend Broadband recently switched to just such a plan. As one of their customers and as a participant in the lively discussion that preceeded the new plan taking effect, it was with more than a little interest that I read the paper.
This paper was only recently released and makes several arguments about why such plans may be harmful to customers and ISPs alike. Since it comes from a fairly well known/respected website it will likely be oft-quoted in future debates on the subject. It seemed worth bringing this paper to your attention since it touches on some good points at the center of this issue and provides some interesting data. However I would be remiss if I didn’t also point out what appear to be a few of the fundamental flaws in reasoning used by the Author …
On Power Users & Scale Free Networks
A large part of the paper is given to a discussion of “power users” (the top 1% of bandwidth users) and how they act as “hubs in a scale free network”. Unfortunately the Author shows a surprising lack of comprehension not only of the nature of the problems that ISPs face, but of some pretty fundamental computer science concepts. In the example given in figures 1 & 2, the author claims a more hub-like network structure (figure 2) is 60% more “efficient”. This is incorrect – the structure has 60% fewer “routes” for data to travel, but does not reduce the amount of bandwidth required. Each route simply carries more data, so the system is at best the same when talking about the issue of bandwidth caps. (In fact, it’s probably worse due to the effects of uneven bandwidth usage on the “last mile problem“.)
The paper also claims that power users are critical hubs in social networks. However there is no substantiation for that claim, and given that social networks (e.g. MySpace, FaceBook, Twitter, and Linked In) are not inherently bandwidth intensive, it’s not clear what relevence this would have even if it were true.
On Trends and Pricing
One point the paper makes, that is irrefutable, is that household bandwidth consumption will be increasing in coming years. This is absolutely true. However the author incorrectly assumes that ISPs won’t be adjusting their price model over time. He combines the fairly conservative plan that Time Warner is experimenting with (5GB cap, $1/GB over) with Cisco’s prediction that in 4 years the average household will consume 200GB/month and concludes that users will be paying an average of $215/month in 2012. The likelihood of this happening is essentially zero. Demand from users, and competition from other ISPs will force companies to raise caps and lower pricing on a regular basis. We’ve already seen the effect of consumer pressure here in Oregon where Bend Broadband increased their cap from 20GB to 100GB in response to customer demand, even before their plan took effect. Thus, the $215/month figure would seem to be pretty meaningless – there are too many unknown factors unaccounted for.
[Update, Oct 6, 2008: Tom Lee over at TechDirt is a bit harsher in his criticism, and points out basically the same flaws.]
Loading ...