Net Neutrality. What good is a free operating system without a network?
Ben Scott
dragonhawk at gmail.com
Thu May 11 12:42:29 EDT 2006
On 5/11/06, Fred <puissante at lrc.puissante.com> wrote:
>> Why, exactly, should ISPs be required to charge a flat rate?
>> ... For just about everything in the world, the more you use,
>> the more you pay. ... Why is data transfer different from the
>> rest of the universe?
>
> Data, on the other hand, *is* in a different universe. Your ISP does not
> incur costs on a per-packet basis.
That's not actually true. We just like to look at it that way.
There are two types of costs directly involved in Internet
connections. (I'm ignoring physical plant, support, staffing,
advertising, etc., for purposes of this discussion.)
One is the cost for the dedicated feed to the subscriber (i.e.,
you). For DSL and leased lines, this is the local loop -- the pair of
wires from your house to the local telco CO. For cable, this is the
cable wire from your house to the tap on the pole. This can be seen
as a continuous, dedicated, unshared resourse. It's only for the
subscriber. It costs the same regardless of use. You statement is
correct about this part.
Once you actually get to the the interconnected, shared part of the
network, things change. Multiple subscribers share the data pipes
throughout the network. Switches, routers, concentrators, DSLAMs, etc.
Ethernet cables, leased lines, fixed wireless links. These are all
shared.
The shared network is always oversubscribed -- meaning the backbone
capacity is less than the aggregate of all the subscriber connections.
This is the only way to do it, and is a big part of what makes
packet-switched networks (like IP) so much cheaper than
circuit-switched networks (like the traditional PSTN). For most
subscribers, the feed is idle most of the time, so you can get huge
cost reductions once you're in the shared network.
But this also means that the more bandwidth a subscriber uses, the
bigger the slice of the shared network pie they consume. If there's
not enough pie to go around, the operator has to buy more capacity
(equipment, lines, etc.).
So my mom who reads email and browses the Ikea website has a *much*
different cost impact on an ISP than someone who spends all day
downloading porn, warez, music, and CD images via P2P networks.
Likewise, a huge web site (like Google) has a bigger impact on the
backbone then www.gnhlug.org does.
So saying all subscribers are equal is wrong.
Now, pricing structure isn't just about bits. In many cases, the
average subscriber usage pattern is good enough, so it isn't worth the
hassle of metered billing. Just charge people a flat rate and be done
with it. But that's a statistical business decision. The costs are
still there.
This is not a new thing, either. The private dial-up services (AOL,
Prodigy, et. al.) all used metered billing for a long time. Most
so-called "unlimited dial-up" accounts weren't really unlimited,
either -- if you tried to stay connected 24/7, you'd run into problems
with the ISP.
So, if a carrier wants to charge Google more money because they eat
up more infrastructure than GNHLUG does, that alone doesn't make me
want to take to the streets with a torch and a pitchfork.
The fact that there's a lack of competition in the space of big
Internet carriers *is* an issue, as it means Google doesn't have much
of a choice for carriers. There's no market force to keep costs in
check. As I said before, I suspect this net neutrality shitstorm is
really a symptom of that.
I haven't had a chance to review the documents that Greg Rundlett
kindly linked to yet, so I'm still incompletely informed.
> Data is not harvested from mines or wells at great expense. Data is virtual.
Sure, but the infrastructure to haul it around -- the "conduit", as
you put it -- is very, very real and very, very expensive.
> Having said that, I will also state that I am NOT in favor of government
> regulation. The less government intervention, the better.
Gee, I'm shocked to hear that from you, Fred. ;-)
-- Ben
More information about the gnhlug-discuss
mailing list