Qu’ils mangent de la brioche?
As the sun rose this morning, America’s broadband policy wonks were greeted with news that the Open Technology Institute is about to bless us with a new addition to their report series on the state of America’s broadband networks, The Cost of Connectivity 2013. The previous edition charged that Americans overpay for broadband service. In OTI’s words:
The results showed that, in comparison to their international peers, Americans in major cities such as New York, Los Angeles, and Washington, DC are paying higher prices for slower Internet service. While the plans and prices have been updated in the intervening year, the 2013 data shows little progress, reflecting remarkably similar trends to what we observed in 2012.
As usual, OTI emphasizes the high end of the broadband market: They’re more interested in Ferraris than Fords and more concerned with Gulfstreams than Greyhounds. That’s perfectly fine for the broadband connoisseurs among us who spend their time obsessing over Speedtest figures, but it may not be quite as meaningful to the average citizen as measurements of normal speeds and median prices would be. For all practical purposes, we live in a 15 megabit per second broadband world today (plus or minus 5), so the price and availability of 100 Mbps or 1000 Mbps (Gigabit) services is not really a serious issue to most people.
In fact the report does a very poor job of showing compelling motivations for extremely high speed networks. Is there an application that the folks in Hong Kong and Kansas City are enjoying with their gigabit services that the rest of us can’t touch, or are they simply watching the same Netflix streams and YouTube clips as the rest of us? It’s an important question because Netflix is still a three (3) Mbps service on Google’s gigabit connections in Kansas City despite the awesome capacity and potential the network brings to the home. OTI simply says speed is kind of important for unspecified reasons:
The speed of your Internet connection determines your ability to view web pages, download and upload content, and use applications and services like voice over IP and two-way videoconferencing. As network technology has improved over the years, the general trend has been an increase in broadband connection speeds. At the same time, however, the amount of bandwidth required by the average consumer increases as streaming video and sharing user-generated content becomes commonplace. Speed, therefore, remains an important metric for consumers when evaluating their broadband options and considering how they plan to use the Internet.
Even this mild claim isn’t true: According to the Akamai State of the Internet Report, web pages don’t load substantially faster in cities with the highest network speeds than they do in the average American city. This is simply because network speed is less likely to be the limiting factor than is server capacity. The average performance of the average TCP connection measured by Akamai in the United States is about 8.7 Megabits/second, and the corresponding figure in Hong Kong is 10.8 Megabits/second. Consider the cost of upgrading all of America’s broadband networks to fiber to the home with the latest electronics – the National Broadband Plan put it at $350 Billion – and then tell me whether it’s worth the cost to do such an upgrade right away. Better yet, tell me how much you’d be willing to pay for an extra two megabits/second of average speed and why you’re not buying that upgrade today, given that it’s probably available to you.
It’s also the case that video streaming is a 2 – 3 megabit/second application, and video conferencing runs at roughly the same rate divided between the upstream and downstream directions. The gap between the requirements of today’s applications and the capacity of today’s networks is already ridiculously large, so there’s no rational argument for increasing it. It should be noted that Gigabit networks have been available to homes, offices, campuses, and wide-area commercial connections since the late 1990s, and the only use for them so far is to support large numbers of users: in an office setting, a 1000 Mbps connection serves hundreds of users. Your kids may be loud, but you don’t have hundreds of them unless you’re a fish.
In addition to its elitist bias, the OTI report is simply wrong in many of its factual assertions. The most glaring is its claim that American cities are as densely populated as those with faster and cheaper broadband:
Many American consumers take high prices and slow speeds to be a given, but our data demonstrates that it is possible to have faster, more affordable connectivity in cities of comparable density and size.
In fact, America’s large cities are not nearly as densely populated as Seoul, Hong Kong, and Tokyo: New York City’s population density is 26,953 people per square mile and Seoul’s is 44,691.
But the more disturbing error is OTI’s claim that low prices for high speeds broadband services in some foreign cities are the result of competition between multiple providers:
Our data also shows that the most affordable and fast connections are available in markets where consumers can choose between at least three competitive service providers.
This claim is neither accurate nor meaningful. In Japan, Korea, and Hong Kong, high speed fiber services are provided by the same firms that provide DSL services, so the only real competition for service quality and price on wired networks is between the former telephone company and the cable company, exactly the situation we have in the United States. The fact that there may be a number of retail service providers offering plans on a common wire plant doesn’t improve quality and price, although it may affect the ordering and service experience.
In fact, extremely high speed, low price broadband networks only come about when there’s a deliberate effort by the government to subsidize and support them. Where the market is left to its own devices, we get network connections that are fast enough to satisfy actual user needs and cheap enough not to prevent the mildly curious from subscribing. Left to follow our own wishes, most of us would prefer improvements in our mobile service over gigabit connections at home that will never be exercised at more than two or three percent of capacity: Bandwidth has no value unless it’s used, after all.
Finally, OTI fails to provide any analysis of the benefits, if there are any, that have come to the cities and nations that have gone down the path to highly subsidized, ultra-fast networks. Have Hong Kong, Japan, and Korea experienced profound economic growth, better education, and improved quality of life and wellness as a result of their network building programs? Such data would get policy makers’ attention these days, when the most difficult problems faced by lawmakers concern the looming pension liabilities that developed nations all face. Unfortunately for those who insist that only gigabit networks will do, research data shows greater benefits for LTE networks than for arbitrarily fast residential broadband. Netflix is nice, but it doesn’t grow the economy.
Another demand for tax dollars must be justified by something more than a desire to achieve bragging rights in an international race to provide questionable services. Until the champions of ultra-fast networks can show some actual benefits their desires don’t deserve the attention of policy makers.
This is not to say that there are not areas in which the market for broadband services isn’t working and may never work: reasonably high speed broadband is not as widely available to rural America as in urban areas, with the exception of LTE and satellite services with limited capacity (satellite capacity is fine today, but today’s systems have a limit of about one million combined users that could become a problem if LTE doesn’t go deep enough into the boondocks.)
In addition, many city dwellers cannot afford both wired and mobile broadband connections and the devices to use them. Income inequality is increasing in America, and its effects are far-reaching and pernicious. This is a larger problem than broadband policy can solve by itself, but subsidies for the working poor and those actively seeking work are sensible investments in workforce development.
All in all, OTI’s emphasis on Cadillac broadband plans and lower prices for HBO is a bit mystifying. OTI is supposed to emphasize the broader social good, but it’s hard to see any connection between their wishes and complaints and the actual needs of the disadvantaged and downtrodden. “Let them eat gigabits” is much less compelling in the English translation than it was in the original French and Chinese.
[…] This blog originally appeared on High Tech Forum. […]
This is a good article. I have seen some similar arguments, but they were not as well laid out as this. I don’t get what’s wrong with, as you put it, “Fast enough to satisfy user needs.” Isn’t the actual user experience what we should be focused on? It’s amazing how many people get talking about something as stupid and irrelevant as the findings in the OTI report. I hope people in Washington in charge of broadband see your article before they spend $350 billion trying to satisfy the elites.
No, satellite is not fast enough to satisfy users need, which is precisely why the majority of internet subscriber’s choose alternative technologies when available . Also, LTE is nice for mobile use but doesn’t provide enough utility for your average American families needs. Which is why 83% of smartphone users also subscribe to wired broadband services.
15 megabit/second is plenty fast enough for Netflix, which works well at 2.5 mbps.
Oh yeah? How many movies have you watched on Netflix with satellite?
Many people get all of their videos from satellite services such as DirecTV and Dish. The Internet is not a cable TV replacement, it’s a whole different thing.
You’re right, it’s not only about speeds and realistically, although copper will likely co-exist with fibre in the next decade, it’s also sensible to bet on technical progress in the last mile when over-the-top services are in infant stages.
And there’s a catch: for the same demand levels, doubling the QoS in mobile lte is much more costly than doing it on any type of wireline networks.
Having a hard time locating the part of the Akamai report that says websites don’t load faster in cities with higher broadband capacities. Do you have a page number, or is it from a previous report? Thanks in advance, great article.
I’m using Akamai’s Average Connection Speeds for the US as a whole vs. Hong Kong. ACS measures web page loading as it’s a measure taken at the TCP level with multiple TCP connections running in parallel.