Lies, Damn Lies and Internet Statistics

Share it on Twitter  
Share it on Facebook  
Share it on Linked in  

The issues of Internet capacity, Net neutrality, broadband investment and network management are interconnected. And make no mistake about it: Those interconnections are laced with dollar signs.


Last week, I interviewed University of Minnesota mathematics professor Andrew Odlyzko, who said that the rate of growth in Internet usage is slowing from about 100 percent to about 50 percent per year. He said current investments and the natural tendency of equipment to become more efficient and higher capacity over time should be enough to handle that increase, though he said there are potential spikes -- such as an unexpectedly radical move of video from the airwaves and cable infrastructure to the Internet -- that could upset the apple cart.


I asked Odlyzko why there are so many people screaming about the impending meltdown of the Internet. He suggested that the big issue is that the idea that the Internet is in trouble fits the agenda of powerful forces, such as politicians and equipment suppliers. Odlyzko said there were no quantitative studies that had findings significantly different from his.


Fat Pipe traces the growth slowdown. It notes, for instance, that Cogent's traffic actually declined during the second quarter. Other interesting and rather inconsistent numbers include Level 3 (65 percent growth), Global Crossing (140 percent) and AT&T (more than 50 percent for its backbone). Numbers garnered by Oldzyko's study suggest less than 30 percent annual growth in Japan and single digit growth in Hong Kong. The general trend indeed does seem to be toward slower increases in traffic.


The importance of the perception of whether the Internet is a bit or a byte away from digital Armageddon is reinforced by this study, which was done by a group called Speed Matters. The group ominously says that real-time download speeds in the U.S. are 2.3 megabits per second (Mbps), the Internet equivalent of a gravel road. The study says that speeds are 30 times faster in Japan, 25 times faster in South Korea, eight times faster in France, and three times faster in Canada.


Scary stuff. While it clearly could be true, the fact that Speed Matters is run by the Communications Workers of America should make everyone feel better. A finding by a union that great investments are needed is about as surprising as a survey by a hotel chain finding that people need to travel more.


The survey is taken to task in this Silicon Alley Insider post. The writer begins by accepting the argument that broadband speeds are too slow in the states. He then excoriates the report, noting that its assessments of speed are based on metrics gathered by people who happen by and take a speed test offered on the Speed Matters site. The writer notes that Speed Matters got its numbers for the other nations from a study by the Information Technology and Innovation Foundation, but opted for its own numbers in assessing the U.S. speeds. The ITIF U.S. speed was 4.9 Mbps, more than double the number it derived from its site visitors.


Statistics are dangerous things. Mark Twain said that there are lies, damn lies and statistics. Clearly, numbers will be used to push the U.S. regulatory framework in one direction or another. It is important that those making the choice use the statistics, not the lies.