Wednesday, November 12, 2014

Genuine competition is best for broadband Internet

Genuine competition is best for broadband Internet:

The third edition of the New America Foundation’s poorly named “Cost of Connectivity” report was released last week, immediately garnering predictable reactions from credulous writers worried the U.S. suffers from slow broadband Internet service. The New York Times declared a lack of competition has caused the U.S. to fall behind the rest of the world. Matt Yglesias at Vox urges the federal government to subsidize networks like the Chattanooga power company’s system even though it is in hock to taxpayers to the tune of $112 million in federal stimulus funds plus another $390 million in bond payments. And consumer warrior David Lazarus demands “open access” regulations in the U.S.

Mandated sharing of high-speed fiber is no panacea

In reality, broadband speed and price comparisons don’t make the case for open access, or allowing rivals to use a provider’s wires or fiber. American broadband activists don’t seem to realize it, but the fastest networks in our competitor nations are generally exempt from open access regulations, even in nations that apply open access rules to DSL. The fastest networks in the G7 are either completely free of open access rules or are subject to rules so limited they have little effect. The 2013 OECD study “Broadband Networks and Open Access” shows no use of open access rules for cable broadband networks in the OECD except in Denmark and Canada, and cable nearly always outperforms DSL.

The very fastest networks use fiber, but most nations with extensive fiber have either granted providers “regulatory holidays” from open access or adopted “contingent model” rules that require rival ISPs who don’t own their wires to lease strands of fiber from incumbents in groups (I explain this in detail in my paper, “G7 Broadband Dynamics”). The net result of contingent model rules in Japan is a greater than 70 percent market share in fiber broadband for incumbent NTT-East/West.

Subsidies and partial government ownership

The Korean picture is similar, with Korea Telecom enjoying a 67 percent share of the fiber to the home (FTTH) market. Japan’s FTTH network is cross-subsidized by profits from NTT DoCoMo’s high-priced mobile broadband network, and Korea’s network costs are borne in significant part by landlords supplying in-building facilities; both nations have provided tax incentives for fiber construction by carriers still owned in part by the government.

Japan’s FTTH network hasn’t prevented the continuing stagnation of that nation’s economy, and the most readily discernable impact of the Korean FTTH network is a national fascination with gaming that some term a national peril; the nation is now considering a “Game Addiction Law” to restrict access to this apparently risky activity.

Inflated claims, limited utility

The Speed ranking of Internet service systems must account for two realities that NAF scarcely considers: first, advertising claims about broadband speeds are notoriously inflated in many nations; and second, the point of diminishing returns in consumer broadband is far below peak speeds in most developed nations. The European Commission has found that broadband providers in the EU deliver less than 76% of advertised speeds, and anecdotal studies in Japan find an even larger gap. Real broadband capacity in the U.S. actually exceeds advertised speeds across the board.

Second, the property we call “speed” in broadband systems (bandwidth) is more accurately termed “capacity”. This is important because we understand intuitively that capacity is dictated by usage. Nations that provide capacity far in excess of actual usage don’t achieve much benefit, and the only nation in which users download and upload more data over residential broadband networks than the U.S. is Korea. The FCC notes that increases in broadband capacity over the normal 10 Mbps standard are unlikely to make web pages load noticeably faster, even if they provide benefits for other applications. Programmers at Mozilla, Google, and other browser creators fret about the coordination of workflows (“parallelism”) and rendering performance, because they know broadband capacity isn’t the whole story.

Network capacity already on natural growth trajectory

The usage gap between the U.S. and Korea is narrowing, and a year from now the U.S. will probably surpass Korea despite our allegedly “pitifully slow and overpriced” infrastructure. It makes sense to build networks with more capacity than needed by today’s applications; it’s not obvious that they need to be overbuilt by 100 times when network capacity tends to double every three years simply as a result of advances in technology. Lest we forget, the cable modem plans that offered 1.5 Mbps at their introduction in the late ‘90s now exceed 100 Mbps in most areas and xDSL has jumped to 40 Mbps or higher in many areas (including my home in Colorado).

We should also look askance at any study that seeks to estimate the cost of connectivity on the basis of prices quoted at the city level. The best deal I’ve found on broadband in the U.S. is the package offered by Sonic in the East San Francisco Bay community of Brentwood: 1000 Mbps broadband and telephone service for $40/month.

NAF excludes Sonic from its survey, as well as the Bay Area’s second best plan, 1000 Mbps from Paxio in Oakland, San Jose, and the Peninsula. Neither Paxio nor Sonic offers fiber service within the city of San Francisco, but it’s not for lack of desire: the Planning Commission simply won’t allow it. This is a matter of local politics, not a national policy.  The best prices are offered by selective local carriers because they don’t have to cross-subsidize rural customers with the profits they earn in urban areas.

The prices that consumers pay for broadband service are dependent in large part on the costs of providing service today and tomorrow, and these costs depend on volume, distance, investment, and consumer demand. The NAF report adds nothing to the broadband policy debate because it overlooks these factors and substitutes cherry-picked prices for real costs.

Reprinted from The Open Standard.