Why do some ISP's have bandwidth quotas?
Steve Gibbard
scg at gibbard.org
Tue Oct 9 19:10:13 UTC 2007
On Mon, 8 Oct 2007, Joe Greco wrote:
> It's arrogant to fix brokenness? Because I'm certainly there. In my
> experience, if you don't bother to address problems, they're very likely
> to remain, especially when money is involved on the opposite side.
There's a big difference between fixing brokenness and demanding that
somebody else do something that might make sense in your situation but not
in theirs.
People in different places deal with different economic factors, whether
in terms of telecommunications, or the movement of goods, or labor.
In the US, we automate a lot of household tasks (dish washing, clothes
washing, etc.) because the machines to do those tasks are cheap, and
peoples' time is expensive. Many people have cars, because they can
afford them, but hiring a chauffeur would be extremely extravagant. In
some other parts of the world, importing machines to wash dishes or wash
clothing would be considered extremely expensive. Anybody who could
afford such things would instead spend $10 per month on a servant who
would do what those machines do, and clean the house and cook meals.
Cars, made with expensive foreign materials by expensive foreign labor,
would mostly be unaffordable, but anybody who could afford one would also
spend the extra $10 per month on somebody to drive it for them.
The same goes for telecommunications connectivity. In the US, we've been
conditioned over the last ten years to think of long haul telecom capacity
between major cities as a seemingly infinite resource with a price
approaching zero, so expending extra effort to limit its use wouldn't make
much sense. In parts of the world where the same long distance capacity
that we take for granted sells for $5,000 per megabit per second per
month, and where $5,000 represents 10 or 20 years of income for the
average person, people look at it differently.
So, what drives the telecom cost differences? Distance and difficulty of
terrain are certainly partial answers, as are economies of scale. But, if
we look at what happened in the US, the story is a bit different. In the
late 1990s, we had a telecom construction boom. Demand for capacity was
surging and lots of investors wanted to invest in new capacity. There was
a set of cities that were seen both as where the money was, and where a
network had to be to be a peer of all the other networks. Once somebody
was laying a little bit of fiber, it didn't cost much more to lay a lot of
it, so lots of companies ended up burying lots more fiber than they had a
use for between the same set of points. Doing that was really expensive,
and the construction had to be paid for up front before the fiber could be
used. Presumably, the investors assumed that once the fiber was in, it
would sell for what fiber was selling for before the construction boom,
and they'd all make their money back.
Unfortunately if you're an investor, but fortunately if you're an American
consumer, the market didn't work that way. It became a competitive market
with high capital costs, a near infinite supply of the product, and lots
of competition. The fiber companies couldn't price their services to
recover their construction costs, because if they had charged more than
any of their competitors they wouldn't have made a cent. Instead, they
priced their services to be cheaper than the competition, trying to
salvage whatever money they could. The competitors responded by pricing
things even cheaper, trying to make sales of their own, and the cycle
repeated itself again and again as the price of capacity along those
routes fell towards zero. Eventually, the construction debt went away in
bankruptcy, the fiber got bought up cheaply by companies that hadn't lost
everything in building it, and what it had cost to build ceased to be
relevant at all.
In other words, capacity in the US is cheap because a bunch of investors
screwed up. That's nothing new; it's how the American railroads got built
in the mid to late 1800s, and it's how the original American phone
networks got built in the early 1900s. Investors will presumably keep
making similar mistakes, and society will be better off because of it.
But counting on them to make the same mistake while investing in the same
thing within the same decade may be pushing it.
Unfortunately for consumers, and fortunately for investors, this pattern
didn't repeat itself everywhere. Those who built fiber on paths that
were not seen as where the money was ended up with monopolies. They can
charge far more than their construction costs, as long as they can find
customers willing to pay. They're vulnerable, of course, to somebody else
coming along and building a parallel cable that forces their prices
towards zero, but such a cable would force its own price towards zero as
well, and generally wouldn't be a good investment. Second cables do
occasionally get built, but often their either on sufficiently different
paths that they still provide a monopoly on connectivity to somewhere, or
they're built by somebody who needs the capacity themselves.
If you're an ISP in an area served by an expensive long haul capacity
monopoly rather than a cheap competitive free for all, the economic
decisions you're likely to make are quite different than the decisions
made in major American cities. If you can always go get more cheap
capacity, encouraging your customers to use a lot of it and thereby become
dependent on it may be a wise move, or at least may not hurt you much.
It's probably cheaper than keeping track of who's using what and having to
deal with variable bills. But if the capacity you buy is expensive, you
probably don't want your customers using a lot of it unless they're
willing to pay you at least what you're paying for it. Charging per bit,
or imposing bandwidth caps, is a way to align your customers' economic
interests with your own, and to encourage them to behave in the way that
you want them to.
I should also note that those advocating "carrier neutral" fiber to the
home probably won't get high speeds at low prices, as long as that
"carrier neutral" fiber is a monopoly. Even regulated monopolies are
generally allowed to charge enough to make back their construction costs,
and if the ISPs are paying for the fiber monthly rather than up front,
that's going to put a pretty firm floor on what ISPs can charge. What
would theoretically push consumer prices down would be to have several
competing companies build fiber to the same set of homes, that they or
their post-bankruptcy successors would continue to control. We've sort of
got that in much of the US, with the DSL/cable duopolies, and speeds do
seem to be creeping up slowly, although not nearly as much as in some
other parts of the world. The question, of course, is how to convince
investors to go for such a thing.
-Steve
More information about the NANOG
mailing list