Interesting People mailing list archives

Re: cost of 1 gig GOOD POINT


From: David Farber <dave () farber net>
Date: Wed, 15 Apr 2009 10:42:53 -0400



Begin forwarded message:

From: "Savage, Christopher" <ChrisSavage () dwt com>
Date: April 15, 2009 8:58:52 AM EDT
To: <dave () farber net>
Subject: RE: [IP] Re:   cost of 1 gig GOOD POINT

Richard makes an assertion below that is not, in fact, generally true:
that "we have the fairly undeniable fact that services pay for
infrastructure in America."

The most basic infrastructure one can think of is roads.  Services don't
pay for roads, taxes do.  Similarly education.  Street lights.  Storm
drainage.  Sewage treatment.  There are fees for public transit but they
overwhelming majority of funding, I believe, is simply subsidies from
tax payments.

Health care is paid for by a combination of taxes (think Medicare,
Medicaid, insurance for government workers), insurance, HMO-like
arrangements, etc.  Actually paying on a service-by-service basis is
notional rather than real.

Richard's assertion masks what is, in the broadband and other arenas,
probably the critical question: What do we actually MEAN by
"infrastructure"?

I'd throw out, as a first approximation, infrastructure is something
that: (a) is capital intensive; (b) is long-lived; (c) is a necessary,
or at least a key, input to a wide variety of unrelated productive and
leisure activity; (d) is technologically relatively stable; and (e)
exhibits strong positive externalities.

The higher something scores on these five criteria, the more plainly it
is "infrastructure" that will likely be provided via some kind of public
mechanism, as opposed to via reliance on market incentives.  As the
thing in question scores lower on any/all of these five, it looks less
like infrastructure and more like a normal private market-produced good
or service.

Of course, sometimes the public relies on market incentives (by creating
monopolies) to motivate private capital markets to build the stuff
needed, rather than taxing people directly for it.  E.g., toll roads;
monopoly ferry routes; monopoly electricity or gas or telephone
franchises. But over time as the technology stabilizes and positive
externalities make it impossible for private providers to capture the
full benefits of their investment (causing them to under-invest), it
will make more and more sense for the public to step in.

Obviously I'd welcome thoughts and critiques of this proto-model.  But
the point is, we need SOME systematic way to think about what makes
something "infrastructure" and, if something is "infrastructure,"
whether its particular characteristics suggest any particular public
policy regarding its provision.

Chris S.

-----Original Message-----
From: David Farber [mailto:dave () farber net]
Sent: Wednesday, April 15, 2009 7:04 AM
To: ip
Subject: [IP] Re: cost of 1 gig GOOD POINT



Begin forwarded message:

From: Richard Bennett <richard () bennett com>
Date: April 14, 2009 8:34:14 PM EDT
To: dave () farber net
Subject: Re: [IP] Re:  cost of 1 gig GOOD POINT

This is a good point, and it goes to one of the two underlying issues
in the usage-based pricing debate. On the one hand, we have some
quantifiable infrastructure costs that are needed to support peak
load, most of which have been exposed in Brett's message and the
various critiques and clarifications of it (and kudos to Brett for
having the nerve to show his work, BTW, which takes much more nerve
that hand-waving.)

But on the other hand, we have the fairly undeniable fact that
services pay for infrastructure in America. If we're going to have
FTTH or FTTN, we need to pay for a one-time upgrade in wiring. The
least painful way to do that is to bundle triple-play services with
Internet access, and this has been demonstrated not just in commercial
contexts, but in muni fiber projects. Burlington, VT and Morristown,
TN have muni fiber with bundled TV and phone service, and Burlington's
founder has said that the uptake would have been pathetic without
triple play. So triple play is essential to the financing of  upgraded
cabling.

As people migrate off traditional TV and phone services to IP-only
service plans - and this certainly appears to be the inevitable
migration path - the revenue void needs to be filled for a time (until
fixed infrastructure costs are paid) and usage-based pricing does the
trick. I don't think that's a big secret, or that it's especially
nefarious.

We're at the beginning of this migration right now, so we see great
disparities in IP usage between the early adopters who watch TV via
IP-
delivered video files and the traditionalists who watch TV over cable
packages and limit their IP to browsing web sites and maybe watching a
few cats on treadmills on YouTube. The point is simply that the
economics of TV delivery are disrupted by the download-over-IP model,
not to mention the economics of advertising and subscriptions.

In a sense, usage pricing clears an obstacle in the path of the
migration toward IPTV; some bandwidth cap systems simply cut off users
who exceed the cap (Comcast) while Time Warner makes high users pay
more up to a certain point and then lets them rip; after you pay TW
$150/mo you can download and upload as much as you want. $150/mo is,
not coincidentally, about what you'd pay for the top tier of triple
play.

The thing that I find most remarkable about all of this is the fact
that the Internet is rapidly becoming the world's biggest TiVo. For
all the rhetoric about innovation, building communities, free speech,
and empowering the downtrodden and the neglected, it appears that the
average American citizen wants nothing more that better entertainment
options.

The next time South Park does a show on Internet congestion, they'll
have to replace the oversized Linksys home router they used to
symbolize the net on the last one with a TiVo HD.

RB

David Farber wrote:


Begin forwarded message:

From: Adam Lynch <alynch () gmail com>
Date: April 14, 2009 7:46:15 AM EDT
To: dave () farber net
Subject: Re: [IP] Re: cost of 1 gig

I'm surprised nobody is up in arms about this one, yet, but:

What about all the traffic that never goes off-net- like Akamai'd
content, and other distributed content? Customers will then be
charged
for usage that *never actually leaves TW's network*?

-Adam (who lives in the affected area of Rochester, NY)


On Tue, Apr 14, 2009 at 6:13 AM, David Farber <dave () farber net>
wrote:


Begin forwarded message:

From: James Seng <james.seng () gmail com>
Date: April 14, 2009 5:26:12 AM EDT
To: Dave Farber <dave () farber net>
Cc: ip <ip () v2 listbox com>
Subject: Re: [IP] cost of 1 gig

Ignoring the math aside, and I doubt time warner pays their
bandwidth in
blocks of T1, let's   use market rate of CDN as a basis, you can
get about
20cent/GB. Larger players can get below 10cent. (Disney for one
gets that
price as far as I know).

Of cos pricing varies depend how far you are from the exchanges but
CDN
pricing is as good estimate as you can get, bear in mind the price
I quote
above is US market rate, profitable to CDN operator.

Sent from my iPhone

On Apr 14, 2009, at 5:05 PM, David Farber <dave () farber net> wrote:

So what does it cost off peak?? djf

Begin forwarded message:

From: Brett Glass <brett () lariat net>
Date: April 13, 2009 10:06:43 PM EDT
To: "Dave Farber" <dave () farber net>
Subject: P.S.

See my posting regarding the incremental cost of 1 GB at


http://bennett.com/blog/2009/04/pitchforks-in-austin-time-warners-
bandwidth-cap/comment-page-1/#comment-427947

@matthew: Time-Warner (or any ISP) isn't just paying for transit,
though
that is an expensive component of their costs. They're also paying
a
backbone provider. And since the Internet has "rush hours" (the
busiest time
of day is "prime time," when people all seem to want to start
streaming
video or browsing at once), they have to buy and provision enough
capacity
to keep up with the peaks in demand.
$1/GB is actually pretty close to the incremental cost of pumping
another
gig of data through the system during "rush hour." Think about it:
you have
an hour and a half to get that gigabyte through the pipe to the
user who's
streaming the movie. That's 8 billion bits in 90 minutes. That
means that
you have to add another 1.48 Mbps - about another T1 - of
capacity.
What does this cost? On a monthly basis, this much bandwidth costs
$5.50
if you are co-located at an Internet peering point. The cost is
about $30 if
you're a well connected cable provider with a leased line into
that peering
point. It costs you$148 if (like me) you're a rural ISP paying
$100 per Mbps
for bandwidth; and $450 if you're a rural ISP in some other areas
of America
where the only way to get bandwidth is via expensive bonded T1
lines.
But back to Time Warner's case: the $30 divided by 30 days is $1
per GB -
exactly what Time Warner is charging.



-------------------------------------------
Archives: https://www.listbox.com/member/archive/247/=now
RSS Feed: https://www.listbox.com/member/archive/rss/247/
Powered by Listbox: http://www.listbox.com




-------------------------------------------
Archives: https://www.listbox.com/member/archive/247/=now
RSS Feed: https://www.listbox.com/member/archive/rss/247/
Powered by Listbox: http://www.listbox.com








-------------------------------------------
Archives: https://www.listbox.com/member/archive/247/=now
RSS Feed: https://www.listbox.com/member/archive/rss/247/
Powered by Listbox: http://www.listbox.com




-------------------------------------------
Archives: https://www.listbox.com/member/archive/247/=now
RSS Feed: https://www.listbox.com/member/archive/rss/247/
Powered by Listbox: http://www.listbox.com


Current thread: