Home
2012-11-28

Some Causes of the Slow US Internet

The Problem

Broadband in the US is slow and expensive compared to international broadband.

The Problem Quantified

Every six months, the FCC releases a report quantifying the availability, speed, and pricing of Internet connections in the United States. The latest Internet Access Services report, hereafter the IAS, is for the period ending June 30, 2011 (FCC, "Internet Access Services: Status as of June 30, 2011"). In the first six-months of 2011, the nature and extent of broadband connections changed dramatically in the United States. The number of US connections over 200 Kbps increased by 31% year-over-year (IAS, 1).

Yet even with this explosive growth, US broadband speeds are still incredibly slow relative to the rest of the world. The FCC is required by law to issue an annual report comparing US speeds to international speeds. It released the International Broadband Data Report (Third), hereafter the IBDR, on August 22, 2012. In terms of average download speed, the US is still 24th (IBDR, 24). Countries at the top ranks (Korea, Hong Kong, Sweden) have nearly twice the average download speed as the US (IBDR, Appendix A, Table 3c).

In terms of price, the FCC breaks the comparison down into three speed tiers: 1-5 Mbps, 5-15 Mbps, and 15-25 Mbps. In every category, the US has higher than average prices, and as speeds grow greater, the disparity between US prices and international prices increases. Of the 32 countries measured in the 15-25 Mbps range, the US ranks 26th out of 32. The United States' peers are Iceland, Slovenia, and Chile (IBDR, Appendix C, 8).

This economic reality shapes the nature of broadband in the United States. There are extremely few Internet connections at the average speed in the United States. Rather, there are a small number of incredibly fast Internet connections, and a huge number of extremely slow Internet connections. 61% of Internet connections in the United States have downstream speeds of 3 Mbps and below (IAS, 3).

The Argument

In the near future, wireless broadband will not solve the fact that broadband in the United States is slow and expensive, for the following reasons:

  1. The physical nature of wireless networks limits supply.
  2. Enormous pent-up demand will keep prices high.
  3. The structure of the US wireless broadband market leads to weak competition and inefficient market decisions.
  4. The US already leads the world in adoption of the latest wireless technologies. Prices remain high and speeds remain slow.
  5. The countries with the fastest, cheapest Internet access have large, well-developed Fiber-to-the-Home networks. The US does not.

The Old Guard and the New

In the United States, cable and DSL were the first to provide last-mile broadband access on a broad scale. They leveraged existing infrastructure to allow them to transmit Internet data over circuits once reserved for voice and video transmission. The actual last-mile cable used is largely unchanged from what it was twenty years ago. (Cat 3 cable was introduced in the early 1990s (Wikipedia, "Category 3 cable").) Upstream, the infrastructure has changed tremendously.

The new guard is wireless broadband, commonly known in the US as 3G, 4G, LTE, and WiMax. These rely on securing the exclusive use of a portion of the wireless spectrum and building new towers to expand coverage and increase capacity.

Wireless Explosion

Wireless broadband expanded dramatically in 2011 in the United States. In 2011, investment in wired and wireless network infrastructure increased by 24%. The US now has the highest rate of LTE adoption in the world. This expansion shows no sign of abating. Analysts predict total US investment in wireless data networks will be $25-$53 billion in 2012-2016. (IBDR, 2).

Wireless Limitations: Physical

Despite this wireless explosion, US broadband speeds remain low and prices remain high. One cause is the nature of wireless networking. There are fundamental limitations that make scaling wireless networks more difficult and costly than scaling wired networks. With wireless, the communication medium is inescapably shared. Various ingenious protocols, like High Speed Down-Link Packet Access (HSDPA), have been developed through pragmatic, real-world testing and optimization. These protocols allow a huge number of devices to efficiently share a limited band of spectrum. Adding more towers closer together can also increase capacity, by enabling all devices to transmit data using less power . A large number of low power transmissions traveling short distances creates much less interference than a small number of high power transmissions traveling long distances, improving performance.

Even with the multitude of exciting wireless network improvements on the horizon, the physical reality cannot be escaped. The communication medium will, at root, remain shared. At any given location, as the transmissions of one user increases, the available bandwidth for all other users at the physical level decreases.

With computers, dramatic improvements in performance and dramatic drops in price are de rigeur. The human time in discovering how to manipulate bits of silicon in clever ways is the primary driver of cost, not the silicon itself, as demonstrated by the rapid improvement and spread of ARM processors.

Wireless networks are different, more akin to oil than RAM. Wireless networks center on a limited, physical resource and extensive, ongoing investments in infrastructure to harness that resource.

Wireless Limitations: Demand

As speeds continue to increase, existing users will send more and more data, and do more and more of their network communications over wireless networks.

Most users who possess wireless broadband with their smartphones also have wired broadband, either at home or at the office or both. Only 17% of cell phone owners do most of their browsing on their phone, and the majority of these do so for convenience, not because their phone is their only option. (Pew Internet Project, "Cell Phone Internet Access").

This is due to the limitations of relying solely on wireless broadband. With tethering, there is no technical limitation on all of a user's traffic being sent using devices already in the home. Except for the network. All other things being equal, users would prefer to rely on a single Internet connection that is always fast, always reliable, and always with them.

For wireless broadband, demand from existing users is poised to increase just as quickly as supply. This is not a recipe for falling prices and expanded access.

Wireless Limitations: Weak Competition

Wireless broadband also faces much weaker competitive pressures than wired broadband. The central physical constraint of wireless networking (a shared medium) means that a certain band of spectrum must be managed by one and only one carrier. This permits only a limited number of providers to exist. In the US, there are four: Verizon, AT&T, T-Mobile, and Sprint. Other providers can and do exist, but they do not manage their own networks. They simply sell access to one or more of the Big 4. It places the Big 4 in the awkward position of being both a vendor and a competitor. If competition from the smaller companies grows too fierce, the Big 4 can raise rates or reduce service quality to reduce competition. Regulatory action can and does mitigate this. But regulatory action will not limit acquisitions, except for among the Big 4. It makes for an awkward, imperfect competitive environment.

This imperfect competitive environment may or may not lead to strange economic arrangements like the two-year smartphone contract, centered on the "free" or nearly free 2-year upgrade. This "free" upgrade is in reality one of the most expensive financing plans ever devised. The subsidy received is paltry, a few hundred dollars at most. Over the course of two years, the aggregate premium paid by the consumer is many multiples of that. (If a credit card company tried to offer financing at similar rates and terms, the consumer outrage and governmental response would be breathtakingly swift.) No doubt a part of this financing is cultural, centering on American's peculiar fascination with credit and tolerance for paying high interest rates in exchange for immediate gratification. Also, the deep American longing to get a "deal," and readiness to, with the tiniest provocation, perceive whatever economic choice they make as a deal.

Whatever the cause of the two-year contract, it makes an imperfect competitive environment significantly less competitive. Switching monthly, the consumer could conceivably try the offerings of the Big 4, as well as a handful of the sub-carriers, in a fraction of the year. But with a two year lag time between experiments, it would take more than a decade to sample the same number of networks.

Two years, let alone a decade, is an eternity in wireless networking. Especially at the moment. In August 2012, independent test showed that T-Mobile had real-world average download speeds of 9.2 Mbps in Portland. Just eleven months before, independent tests by the same company showed that T-Mobile had real-world average download speeds of 3.2 Mbps in Portland. Verizon performance remained the same for both tests, with a download speed of about 14.6 Mbps. (Rootmetrics. "Compare Carriers: Portland").

Take the hypothetical Portland consumer who had a T-Mobile contract from 2008-2010. Experiencing the slow download speeds, they decided quite rationally to switch to Verizon. Their Verizon contract then expired in 2012. They may have heard vague tales of T-Mobile being faster. But the economic information they are going to trust is their own experience from 2008-2010. Two-years is a long commitment. If they have come to rely on the fast data speeds, as most smartphone users do, they will be extremely wary of taking a chance on T-Mobile again, no matter how clever or pervasive its advertising, no matter how attractive its pricing.

It is a recipe for a group of customers consistently having inaccurate data and making suboptimal decisions. It is a recipe for market failure.

The current system also disproportionately favors incumbents. Incumbents have the greatest access to capital and expertise, which in turn secures the greatest access to spectrum and ability to deploy the most advanced technology. At times, Verizon is not faster than the competition. But never will it be slower. For precisely this reason, one hears very few complaints about Verizon's network, except for its price. Indeed, Verizon's whole advertising strategy centers on the superiority of its network. The fact that Verizon's network is only occasionally faster and more reliable than its competitors tends to get lost.. The enormous lag time before consumers gain an accurate perception of current network performance, a lag time greatly increased by the two-year contract, creates a sort of fortress around the perception of the incumbent's superiority.

Fiber-to-the-Home

By contrast, wired networking does not face the same fundamental physical constraints as wireless. There can be billions of independent circuits carrying the data of millions of users through a tiny physical space. This is more than just a theoretical possibility of wired networks. At the core routers of the Internet, this happens every second of every day.

At the home, one could conceivably have a hundred different a hundred different Internet Service Providers, each with a separate fiber optic cable, all serving the same building. There is an economic limitation why this does not happen. But there is no physical constraint.

This fundamentally different physical reality leads to a very different economic reality. With wired networks, no user's use limits any other user's use. So bandwidth caps are 50x higher (or more) versus wireless broadband.

Fiber-to-the-Home (FTTH) is the next stage of wired Internet access. Yet for all the US investment in the latest wireless Internet technologies, it is not investing in FTTH. This should give regulators, city officials, consumers, and entrepreneurs pause. Looking at the countries with the fastest Internet and the best prices, they all have one thing in common: FTTH.

There are 85 million wired Internet connections in the United States. In 2008, only approximately 13.1% of wired connections were FTTH. By contrast, the rate of FTTH in Korea is 67% and in Japan is 86.5%. Since 2008, the rate of FTTH penetration has increased in other countries, but has stagnated in the US. Verizon's FiOS was the largest FTTH program in the United States. It began in 2005, but by late 2009, Verizon was already winding down its expansion.

There is one well-publicized new FTTH effort: Google Fiber. But this effort is currently limited to a single city. This seems unlikely to change soon. Being an Internet Service Provider is far from Google's core business. Rather, Google's efforts are a transparent attempt to invigorate the moribund, or at least calcified, US consumer FTTH space. By that standard, it has failed. (Or perhaps just not yet succeeded.)

What of companies whose core business is being a Internet Service Provider? Take Verizon, the largest FTTH provider and largest wireless network provider. It's FTTH expansion plans halted immediately after its long-term multi-billion dollar content bundling and spectrum deal from Time Warner Cable.

One of the great advantages of using cable as a way to deliver Internet is the "natural" bundling it provides. Consumers will pay a large premium for video data called "cable," over and above what they pay for video data from other sources like the Internet. FTTH would mitigate these bundling advantages. While there is no technical reason why copper wire rather than optical fiber should carry the video data called "cable" into the home (quite the opposite actually), there is a cultural reason. Cable is cable.

Also, better wired Internet might cannibalize Verizon's much more lucrative wireless data network. Users might decide they could live with 2 gigabytes of monthly data, rather than 5.

In other countries, where cable is much less widely spread (physically or culturally or both), these profit sources are much less powerful. Thus, FTTH has and continues to make a great deal more sense than other alternatives. While the US has stagnated, FTTH becomes ever more prevalent outside of the US. This is a recipe for US Internet to become ever slower and more expensive relative to the rest of the world.

Conclusion

US Internet is slow and expensive. In part, this stems from the fact that the center of US investment in expanded network capacity is wireless, not wired. In part, this stems from the structure and incentives of US Internet Service Providers.

Entrepreneurs, city officials, regulators, and consumers themselves can all help to create a more efficient market that yields faster, cheaper Internet. Specific solutions are beyond the scope of this essay. But I hope I have shed light on the scope of the problem and some of its causes.

Sources

FCC, "International Broadband Data Report."
FCC. "Internet Access Services."
Ingrid Lunden. "Post Its $3.6 Billion Cable Deal, Verizon Puts Brakes On FiOS Expansion." Web.
Pew Internet Project. "Cell Phone and Internet Access." Web.
Rootmetrics. "Compare Carriers: Portland."
Wikipedia. "3GPP Long Term Evolution."
Wikipedia. "Category 3 cable."
Wikipedia. "High Speed Down-link Packet Access."
Wikipedia. "Verizon FiOS."


Blog Index