[100510] in North American Network Operators' Group
Re: Internet access in Japan (was Re: BitTorrent swarms have a deadly bite on broadband nets)
daemon@ATHENA.MIT.EDU (Tom Vest)
Thu Oct 25 10:04:33 2007
In-Reply-To: <20071024104034.X17322@sprockets.gibbard.org>
Cc: nanog@merit.edu
From: Tom Vest <tvest@pch.net>
Date: Thu, 25 Oct 2007 15:56:04 +0200
To: Steve Gibbard <scg@gibbard.org>
Errors-To: owner-nanog@merit.edu
On Oct 24, 2007, at 8:11 PM, Steve Gibbard wrote:
> On Wed, 24 Oct 2007, Rod Beck wrote:
>
>> On Wednesday 24 October 2007 05:36, Henry Yen wrote:
>>> On Tue, Oct 23, 2007 at 09:20:49AM -0400, Leo Bicknell wrote:
>>>> Why are no major us builders installing FTTH today? Greenfield
>>>> should
>>>> be the easiest, and major builders like Pulte, Centex and the like
>>>> should be eager to offer it; but don't.
>>>
>>> Well, Verizon seems to be making heavy bets on replacing significant
>>> chunks of old copper plant with FTTH. Here's a recent FiOS
>>> announcement:
>>>
>>> Linkname: Verizon discovers symmetry, offers 20/20 symmetrical
>>> FiOS
>>> service URL:
>>> http://arstechnica.com/news.ars/post/20071023-verizon-discovers-
>>> symmetry-of
>>> fers-2020-symmetrical-fios-service.html
>>
>> While probably more "good" than "bad", it is my understanding that
>> when
>> Verizon (and others) provide FTTH (fiber to the home) they "cut" or
>> physically disconnect all other connections to that
>> residence..... so much
>> for any "choice"...
>>
>> Exactly. And because they installed fiber, the FCC has ruled that
>> they do not have to provide unbundled network elements to
>> competitors.
>
> It's this last bit that seems to be leading to lots of complaints,
> and it's the earlier pricing of "unbundled network elements" at or
> above the cost of complete service packages that many CLECs and
> competitive ISPs blamed for their demise. Some like to see big
> conspiracies here, but I'm not convinced that it wasn't just a
> matter of bad planning on the parts of the ISPs and CLECs, perhaps
> brought on by bad incentives in the law.
>
> The US government decided there should be a competitive market for
> phone services. They were concerned about the big advantage in
> already built out infrastructure the incumbent phone companies had
> -- infrastructure that had been built with money from their
> monopolies -- so they required them to "share." This meant it was
> pretty easy to start a DSL company that used the ILEC's copper, but
> seemed to provide little incentive for new telecom companies to
> build their own last mile infrastructure. Once the ILECs caught on
> to the importance of this new Internet thing, that meant the ISPs
> and the new phone companies were entirely dependent on their
> biggest competitor for services they needed to keep functioning.
> The new providers were vulnerable on all sorts of fronts controlled
> by their established competitors -- pricing, installation
> procedures, service quality, repair times, service availability,
> etc. The failure of the new entrants seems almost inevitable, and
> given that they hadn't actually built any infrastructure, they
> didn't leave behind much of anything for those with better plans to
> buy out of bankruptcy.
Consider the implications of this line of reasoning.
A rational would-be competitor should expect to build out a new,
completely independent parallel (national) facilities platform as the
price of admission to the market. Since we've abandoned all faith in
the use of of laws or regulation to discipline the incumbent, we
should expect each successive national overbuild to be accomplished
in a "very hostile" environment (Robert De Niro's role in the movie
"Brazil" comes to mind here).
A rational new entrant should plan to deliver service that is
"substitutable" -- i.e., can compete on cost, capacity, and
performance terms -- for services delivered over one or more
incumbent optical fiber networks -- artifacts of previous attempts to
enter the market. The minimum activation requirements for the new/
latest access facilities platform will create an additional increment
of transport capacity that is "vast" ("infinite" would be only a
slight exaggeration) relative to all conceivable end user demand for
the foreseeable future. The existence of (n) other near-infinite
increments of parallel/"substitutable" access transport capacity
should not be considered when assessing the expected demand for this
new capacity.
A rational investor should understand that capex committed to this
new venture could well be a total loss, but should be reassured that
the new nth increment of near-infinite capacity that they help to
create will be useful in some way to whomever subsequently buys it up
for pennies on the dollar. The existence of (n) other near-infinite
increments of parallel access transport capacity should not be
considered when estimating the relative merits of this or future
access facility investments. Every household will become equivalent
to a core urban data center, with multiple independent entrance
facilities -- unless of course the new platform owner determines that
it would be it more rational to rip the new facilities -- or the old
facilities -- out. (Any apparent similarity between this arrangement
and Mao's Great Leap Forward-era backyard blast furnaces is purely
coincidental).
A rational government should welcome the vast increase in investment
created by successive attempts by would-be competitors to enter the
market, and by the liberation from all responsibility for the causes
and consequences. The FCC can be dismantled, but cashiered former
telco regulators can find new employment in the booming network
facilities construction sector -- or perhaps in the new Resolution
Trust Corporation that will handle the administration/liquidation of
successive would-be facilities-based competitors -- and the financial
institutions that bankrolled them.
The current "incentive problem" is basically a one-time problem.
However, the first full optical network platform that reaches
households will be the last one, and the problem of access segment
market power will be with us forever. History and the existence of 200
+ other simultaneous experiments in national network economics
provide abundant information on how to solve (or fail to solve) the
problem.
But maybe "Brazil" really is the right reference...
TV
> I don't think this was what was intended. My impression is that
> the wholesale copper was supposed to be a temporary bridge to allow
> the new entrants time to build infrastructure of their own. That's
> why the rules about sharing didn't apply to infrastructure built by
> the ILECs later. But new entrants building their own infrastructure
> generally didn't happen. Instead, the end-user ISP operators I was
> dealing with at the time generally seemed outraged that the evil
> phone companies, which should have been there to sell wholesale
> services to them, were instead competing in their markets.
> Unfortunately for them, the phone companies not only undercut them
> on cost, but generally built better networks. Given the impending
> obsolescence of the phone companies' traditional businesses, what
> else would the phone companies have been expected to do?
>
> The exception to this was the cable companies. They already had
> some physical plant of their own, but they invested a lot of money
> in a lot of new construction. Many of them didn't do financially
> well on the deals, but even those who ran out of money left behind
> infrastructure that is now effectively competing.
>
> This isn't to say the original encouragement of CLECs using ILEC
> copper in the 1996 telecommunications act wasn't without benefits.
> I rather doubt the ILECs would have gotten as interested in DSL as
> they did, if there hadn't been the threat of losing the business to
> competition. But given that improvements in speed since the
> initial crushing of the upstarts have been mostly limited to trying
> to match the capabilities of the cable companies, perhaps it wasn't
> the best strategy for the long term. If those who want to compete
> need to build some infrastructure of their own, and if anybody is
> successful in doing so, that should have a much bigger impact in
> terms of putting long term pressure on the ILECs to provide better
> service.
>
> -Steve