Monday, February 1, 2016

U.S. Internet Progress: Lies, Damned Lies and Statistics

The latest Federal Communications Commission report on the status of Internet access does not vary much in one respect from earlier reports: people in rural areas tend to be underserved or unserved.

Beyond that, one has to interpret the results, as the report paradoxically suggests Internet access is getting worse, when in reality it is getting better.

Specifically the FCC report suggests the nation is going backwards, in terms of capability, when in fact capacity has been growing at nearly Moore's Law rates. The report, as always, is right to point out gaps between urban and rural availability and performance.

But it is completely wrong about the trend.

Where in 2012 “100 percent” of urban consumers had “fixed advanced telecommunications capability,” that dropped to 57 percent in 2013 and then to 54 percent in 2014, the report states.

Cable TV operators, who have done the most to improve Internet access speeds, for the most people, have increased speeds  at nearly Moore’s Law rates--doubling nearly every 18 months--for years.  

And now we have Google Fiber, plus gigabit investments by third party providers and telcos including AT&T and CenturyLink.

In fact, over the last year, U.S. Internet access speeds grew 105 percent, ironically using figures the FCC itself provides.

“Between September 2013 and September 2014, we observe a 105 percent increase in the maximum advertised download speeds among the most popular service tiers across participating ISPs weighted by the number of participants using a given ISP,” the FCC said.  

“We find that, over the course of our reports, the average annual increase in actual download speeds by technology has been 28.2 percent for DSL, 61.2 percent for cable, and 19.2 percent for fiber,” the agency said.

The point is that the most-recent FCC broadband progress report has to be heavily interpolated, even as referenced against other recent FCC reports on the subject.



To be sure, updating definitions is not an unreasonable step, over time. Since end user requirements change over time, while supplier capabilities likewise increase, it does make sense to use reasonable “current market” tests. The problem is that comparisons over time then become difficult.

“What” one measures matters. Measuring “fiber to the home” (the means for delivering) instead of capacity leads to distortions, for example. Those sorts of distortions will grow over time as new platforms become widespread, in addition to making comparisons over time less useful.

To be sure, the FCC report points out the gap between urban and rural service, which is real. For an agency whose mission includes universal service, that makes sense.

On the other hand, because the definitions keep getting changed, the sense of progress is blunted.

On the other hand, even using the new 25-Mbps definition, not the older 4 Mbps definition, over the past two years alone, the number of people who have “no fixed advanced telecommunications” service has been cut in half, while the percentage of people living in urban areas without that same level of access likewise has been cut in half.

The report also suggests 10 percent of all residents have “no access.” That is a “platform specific” approach, and will come as news to satellite broadband providers, for whom that group is the primary customer target.

Eventually there will be some other new wrinkles that could skew the results. Mobile Internet access has become more important, and eventually will have to be accounted for, as it now is impossible to evaluate Internet access globally without reference to mobility.

No comments:

"Tokens" are the New "FLOPS," "MIPS" or "Gbps"

Modern computing has some virtually-universal reference metrics. For Gemini 1.5 and other large language models, tokens are a basic measure...