Occasionally I run into articles about how the internet is doomed for economic reasons, even today, like this article by John Michael Greer. But they’re wrong. Greer’s basic argument is that internet services like Facebook are being heavily and unsustainably subsidized by money-losing venture capitalists and that the internet as a whole is heavily and unsustainably subsidized by fossil fuels.
Every time I’ve mentioned the future of the internet on this blog, I’ve gotten comments and emails from readers who think that the price of their monthly internet service is a reasonable measure of the cost of the internet as a whole.
It actually is a reasonable measure, because the majority of the cost of operating the internet as a whole is the cost of operating the so-called “last mile” service. The total cost is maybe twice that.
For a useful corrective to this delusion, talk to people who work in data centers.
I do, all the time. They pay substantially less for their bandwidth and their computers than residential internet users do. This should be unsurprising, because they are able to drive bargains with equipment and bandwidth providers that residential users can’t, and they can place their computers right next to the internet backbone instead of stringing cables through urban neighborhoods.
Amazon may be the biggest retailer on the planet, for example, and its cash flow has soared in recent years, but its expenses have risen just as fast, and it rarely makes a profit. Many other content provider firms, including fish as big as Twitter, rack up big losses year after year.
Twitter’s current operating expenses are around two billion dollars a year, with 302 million active users. That’s six dollars per user per year, or 50¢ per user per month. This is very small compared to the cost of the price of monthly residential internet service.
Facebook’s operating expenses are around ten billion dollars a year, with which they serve 1.44 billion active monthly users, 936 million active each day. Again, that’s about six dollars per user per year.
Wikipedia’s operating expenses are only US$58.5 million, but of course that doesn’t pay the editors, just the server infrastructure. Because of Wikipedia’s privacy policy, they don’t know how many people read Wikipedia every day, but comScore says it’s about half a million. That means Wikipedia spends about 10¢ per user per year, or about 1¢ per user per month.
Now, it’s true that getting some money from the users who benefit from an internet service into the pockets of a company who provides it is a pretty big problem, and one that Greer correctly points out many companies are all too happy to put off solving. But that doesn’t imply that the total costs per user are high — as we’ve seen above, Twitter, Facebook, and Wikipedia (three of Alexa’s top 10) are spending 50¢, 50¢, and 1¢ per user per month. Indeed, in all likelihood, if Twitter and Facebook weren’t desperately struggling to get more users, their operating expenses would probably be lower. But Twitter’s only been able to put off profitability for so long because their cost per user is so low.
(If we extrapolate these costs down to a small three-person company, with each person costing US$100k per year and equipment costing a similar amount, we end up with US$600k / 50¢ ≈ 1.2 million users as the smallest viable online service these days.)
Once this happens, the companies that dominate the industry have to stay in business the old-fashioned way, by earning a profit, and that means charging as much as the market will bear, monetizing services that are currently free, and cutting service to the lowest level that customers will tolerate. That’s business as usual, and it means the end of most of the noncommercial content that gives the internet so much of its current role in popular culture.
I often hear nonsense like this from people who are relatively new to the internet and therefore don’t remember what it was like before privatization. Noncommercial “content” has a history on the internet far deeper than what is suggested here.
What we’re seeing in practice is that, when you lower the cost of communication sufficiently, commercial “content” like Encarta or Encyclopedia Britannica or Microsoft Windows has a really hard time competing with noncommercial “content” like Wikipedia or FreeBSD (the free software MacOS is built from). The big websites like YouTube, Wikipedia, Facebook, and Twitter are not sources of content; they’re just ways for people to talk to each other that happen to be more convenient and efficient, and no more expensive, than running peer-to-peer software on their own computers. The companies that currently “dominate the industry” (if by that we mean Facebook and Twitter and not Comcast and VSNL) are only able to do so because they are successfully able to deliver people to each other.
As long as some people can control the software running on their own computers and get those computers to talk to each other, companies like Facebook will have to compete with “free”. (And as one of the commenters on Greer’s post points out, BitTorrent already accounts for the majority of internet traffic.)
the internet is simply a cheaper and more convenient way of doing things that people were doing long before the first website went live, and a big part of the reason why it’s cheaper and more convenient right now is that internet users are being subsidized by the investors and venture capitalists who are funding the internet industry. That’s not the only subsidy on which the internet depends, though. Along with the rest of industrial society, it’s also subsidized by half a billion years of concentrated solar energy in the form of fossil fuels.
We’ve already shown above that the degree to which internet users are being subsidized by money-losing investors and venture capitalists is very small, but the energy subsidy question here is interesting, and Greer’s position is spectacularly wrong.
It’s true that our fossil fuels accumulated over half a billion years, but that does not mean that they are half a billion years of concentrated solar energy; the vast, vast majority of that solar energy was reflected or re-radiated into space and lost forever, not stored away in coal and oil to await humanity.
Current proven coal reserves are 861 billion tonnes, which at 24 GJ/tonne are about 21 zettajoules. Oil and gas reserves are somewhat less. The earth’s radius is 6400 km, and it receives about 1000 W/m² of solar energy, which is about 128 petawatts over its circular cross-sectional area.
That means that current proven coal reserves are 45 hours of concentrated solar energy, not half a billion years, not a billion years, not a million years, not a thousand years, not a century, not a year, not even a week. Every two days, we receive from the sun as much energy as all of our currently-known coal reserves.
Historically, it has proven difficult to use solar energy — in part because, as Greer correctly notes, fossil fuels are more concentrated and therefore hard to compete with in an economic sense, but also because of our limited materials and fabrication technology, much as nitrate mined in Chile was a crucial strategic ingredient for explosives until the Haber process was invented.
But using solar energy is no longer particularly difficult, if we measure difficulty in dollars — Juan Cole collated several stories of recent utility-scale solar energy, among which is the news that Nawaz Sharif in Pakistan just inaugurated the hundred-megawatt Quaid-e-Azam Solar Plant in Punjab province built “by China’s Tebian Electric Apparatus Stock Co Ltd (TBEA) [in] a year…at a cost of $190 million”. That’s US$1.90 per watt. Spain is already at 70% carbon-free electricity, mostly nuclear, hydroelectric, and wind.
Speaking of nuclear, the amount of energy available in easily fusible elements in seawater is comparable to the amount available in solar, if burned over a reasonable period of time; thorium is somewhat smaller, while uranium is substantially smaller. Fusion energy, hot-dry-rock geothermal (aka “enhanced geothermal systems”), and solar are each individually capable of getting human civilization to Kardashev Type 1.