> Please tell me what's wrong about my analysis below. The basic
> conclusion is that regulators are making a terrible mistake by
> not requiring ILECs to construct and sell dark fiber to end users.
While I'm in basic sympathy with your goal, and I spend a fair amount of
time trying to liberate some of the 24 strands of SBC-owned fiber in my
prem, there are some regulatory goals which wouldn't be served by a really
simple implementation of the model you're suggesting.
One of the big regulatory goals is to keep cars moving on the roads,
rather than just backhoes. That means that street-digging permits need to
be issued sparingly. In a country less wealthy than the U.S., digging the
same street multiple times _simply can't be afforded_ by anyone making
rational decisions on behalf of national (or even just municipal)
interest. Does that need to hamper competition on those only-dug-once
paths? Not at all... Look at the model that isn't yet being implemented
in Kabul, or which was done mostly-cleanly in Auckland.... Require that
whoever does dig the street put in enough facilities for themselves _and_
all of their present and future competitors, with the excess going into a
"fiber bank" that bids against the first holder for any fiber sales on
that path, to keep prices in check.
Now, you can't implement a model like that unless all the runs in the
"bank" can be crossconnected. That means that you should continue to use
a "wire center" hub-and-spoke distribution model. So far so good.
But if individual end-users (at the consumer level) start asking for
home-runs to their ISPs, now your crossconnect facilities get really
out-of-hand. Aside from it just being phenomenally inefficient use of the
fiber resource, relative to a packet-switched model.
So I tend to like the Wellington or Kabul models better than the Palo Alto
model... Fits better with our Ethernet-at-L2/IP-at-L3 world-view, and
makes more efficient use of the precious resource, which is PROW
disruption and the associated time/labor.
-Bill
--
general wireless list, a bawug thing <http://www.bawug.org/>
[un]subscribe: http://lists.bawug.org/mailman/listinfo/wireless
