The Internet is hiding in the weeds

The Internet continues to diffuse to households. That trend should not surprise anybody who has been awake this past decade. Some of the details behind the trend might surprise many people, however.

These details never make headlines. Headlines are about new facts, new trends, new developments, and new surprises. The details behind slow moving trends do not make headlines. There are too grinding, too nuanced, and in some cases, just plain boring.

Let me put it this way. Rarely does the diffusion of a new technology appear like an elephant stampede — namely, with lots of thump-thump-thump and distant rumbling that eventually becomes a deafening roar and overtakes everything in its path. Nope, never that loud.

Rather, most technologies spread like the Internet. The Internet spread like a mice or rabbit infestation. The signs were not visible at first, and certainly there was not much noise from any rabbit stampede. The animals crept up on the neighborhood, quietly perhaps, hiding here and there, but traces of their presence appeared all around, never directly in view, but always out of the corner of the eye. They appeared first in one field and then another. Eventually they ended up everywhere.

You heard it here first. The Internet spread like rabbits hiding in the weeds. (Perhaps that is not such a great metaphor, but now that I have written it I am not taking it back.)

More to the point, the details behind the Internet’s diffusion do not get much notice because they are not noisy new factoids or noisy new events. Yet, they actually say a lot about the state of the Internet in the US today, so they are worth a closer look. And that is what I am going to do in this post. We are going to get into the weeds.

The graph

Let’s start with this marvelous graph. It comes from the recent publication, “Digital Nation: 21rst Century America’s Progress Towards Universal Broadband Internet Access”. This is put together by some terrific employees at the National Telecommunications and Information Administration from information collected by the US Census as part of a supplement to the Current Population Survey.

Notice a couple things about this graph.

• First, this measures the adoption of the Internet by households.  As of October, 2009, 68.7% of US households have an Internet connection.
• Second, notice the trends. Notice the impressive replacement of dial-up connections with broadband connections in the last decade. Only 5.3% of US households have dial-up connections for the Internet access. That means only 7.7% of the households using the Internet use dial-up. Wow. What a change from a decade ago.
• Third, in theory all diffusion is the same — it follows an “S-curve” pattern of adoption, but in practice it is not. Recall that households are not the same as individuals. Households with children tend to be larger than households without. Also, US families tend to be larger than families in other developed countries. Finally, if you read the report carefully you will see that US households with children tend to be more prevalent adopters of the Internet than those without children. Which implies two things: when 70% of households adopt it means MORE than 70% of the US population has adopted; it also means comparisons across countries are meaningless without accounting for household size precisely because US households tend to be bigger than those in other countries (By the way, if you did make that adjustment you would realize the panic about the slow US Internet adoption is not quite as bad as sometimes portrayed…).
• Fourth, notice the missing data in 2005. There should be a survey there, but there is not. Therein lies a tale of political prejudices interfering with statistical science. <Warning: Long sarcastic digression coming>. Briefly, after much tussle somebody in the Bush administration won a bureaucratic war and cut the funding for the survey supplement to the CPS that yielded the data for this publication. Now, I am not very astute about politics, but I do know something about Internet statistics. Until that bureaucratic war cut the funding these statistics were the longest continuously administered unbiased survey about household US of the Internet. Then the survey got stopped, future history be damned (not that I am bitter). More amazing for myopia, this was done right in the midst of the diffusion of broadband. So at that moment in time nobody in the US government had any official statistics to tell them what was happening. Policy was flying blind (once again, not that I am bitter). It gets even better. Later something even funnier happened — the administration discovered/decided that broadband was important for US competitiveness and productivity growth, and it discovered that somebody had cut the funding. So the White House restored the funding. That is why the survey restarts again in 2007 after most households had adopted broadband, and why there was funding for another in 2009. <End of digression>
• Fifth, and finally, the rate of adoption seems to be slowing (though it is a bit hard to tell for sure with the 2005 data missing). We are at close to 70%, and it sure looks like many households simply refuse to adopt. In marketing circles this is often called “reaching a saturation point.” Demand growth has slowed.

Which motivates the question as to why more than 20% of US households refuse to adopt. For the answer look at the following pie charts, also taken from the same NTIA publication as above:

Three explanations stand out: can’t get it, don’t want it, and costs too much. This too deserves comment:

• The two biggest explanations relate to demand, such as lack of desire (don’t want it) and lack of low prices (too expensive). I am not sure how to interpret “no computer”, which could either be lack of desire or lack of income or both.
• Lack of supply (can’t get it) is actually rather small. It is fourth on the list, and, for all intent and purposes, it only exists in rural areas.
• It has become politically correct to throw up a graph about the adoption of the Internet around the world, compare the US to other countries, and panic because the US has lower adoption rates than other countries. Look at those pie charts and think about it. Lack of supply is surely not the primary issue. If the biggest reason for non-adoption is “doesn’t want it” then what is all the panic about? Is there any reason to panic because the US has the highest percentage of Luddites in the developed world? Is there any reason to panic if “not available” is such unimportant for explaining why households do not adopt the Internet?
• More seriously, what can policy do about lack of adoption? If anything, there are three basic options: make it available where it aint’, subsidize/regulate vendors so they lower prices, and subsidize purchases at households who have low income. I am not saying anything new. The US already does a mix of all three in telephony. And the Internet already has a bit of this too. The US has been creeping towards a mix of the first two options with its universal service funds called e-rate.
• Let me rephrase that. Only 11% of the non-adopters in rural areas got that way due to lack of supply. That is not an insurmountable problem. It is no surprise that many policy wonks suggest redirecting the funds for subsidizing telephone service into subsidizing broadband build-outs in rural areas. These subsidies have long been far higher than needed to subsidize rural telephone service, and should have been cut a while ago. Only political expediency has kept them alive (after all, there are many senators from rural states….). Well, I can see the sense of those who argue that if we have to collect such subsidies, let’s at least use it for a good purpose, like subsidizing broadband.

In the weeds

The Internet continues to diffuse to households. And when you look in the weeds there are some rather interesting things going on.

That is quite an infestation.