Marcus Didius Falco wrote:
> During the Internet bubble of the late 1990s, predictions of explosive
> growth in Web traffic fueled a construction boom for fiber-optic
> networks. The long-distance industry has since been plagued by price
> wars and a huge glut of unused capacity.
That is the primary reason for the demise of AT&T and MCI. The
product (carrying a long distance call) has become so cheap that the
skills needed to carry it (managing a long distance network) aren't as
This has also happened in the computer world. Once programmers had to
carefully track every character to avoid wasting scarce computer
memory, but today memory is so cheap those tracking skills (and
products) aren't as important anymore either.
Historically, the steam locomotive required much more maintenance than
a diesel which is why diesels replaced steam and many of the skilled
AT&T and Mountain Bell used to advertise their microwave towers built
in very tough places yet they did whatever it took 24/7 in the worst
weather to keep them going.
The high capacity of fibre optic is really no big deal when looking at
long distance over the long haul. In the beginning, it was one very
thick wire barely carrying one call at a time. Longer calls required
setup of trunks connecting one town to the next as well as the
appropriate amplification (too much wouldn't work -- it had to be just
right). In the 1930s every long distance call needed considerable
workup by operators consulting routing charts, and engineers tinkering
with routing and amplification and transmission.
But they discovered carrier signals and some automation so that a
physical circuit could handle more calls and spread the cost. Setup
time was reduced saving money too.
After the war, the Bell System installed coax, microwave, and
automated switching (first for operators, then for the public) that
greatly reduced setup costs and improved efficiency. To the present
day, the improvements continued lowering cost.
(It bothers me that critics of the Bell System think the technology
they used in 1983 was no different than that of 1933 or even 1953.)
> A traditional phone call requires a continuous circuit to be
> "opened" across the network. "When you have to dedicate a line to
> connecting two people on opposite ends of the country, that uses a
> lot of resources," With VoIP, "You can cram a lot more phone calls
> into the same amount of network at the same time."
I'm not sure this is entirely true. Yes, the packet-switching nature
allows much more flexibility, but also risks conversation trouble if
the network happens to be crowded at a moment and the packet is
delayed. I believe the Bell System used a voice packet time of
transmission on overseas cables -- using the natural pauses in a
conversation to sandwich in other calls on the same channel.
Further, a long distance call doesn't exactly have a "continuous
circuit". One is being multiplex among a great many other calls and
data sharing a communication trunk. True, there is setup at either
end, but VOIP requires ISP servers doing the same thing at either end
I expect that ISPs will be forced to add server capacity and raise
prices if they find themselves with heavy VOIP traffic on their
systems. The stuff isn't switched for free.