Voice Economics

Gregory Alan Bolcer (gbolcer@gambetta.ICS.uci.edu)
Wed, 14 Oct 1998 21:45:07 -0700

Voice is expensive. You look at the cost
to support 2000 minutes per month and you will
understand. When you think about it though,
how many of those minutes are spent in dramatic
pause? People speak in half-duplex, why does
voice need to be full-duplex anyways? We as a culture
have become conditioned to need to speak over each
other and interrrupt each other. Speaking over
each other is useful in face-to-face, but does it
really need to happen over digital sound? The reason
VOIP is so hard is that last little and extra mile
of supporting the 3% of conversation that overlaps.

Art Hitomi bought an Nextel i1000 today. One of
the features that I was skeptical about until I got
home and found the ad stuffed in my mailbox was that
it supplies a 2-way radio good for most of souther California.
Similar to how instant messaging combined the
timeliness of of chat with the usefulness of text
and routing, radio has the potential to provide most
of what we use phones for, but at a fraction of the

Imagine a phone conversation where only one person
was allowed to speak at a tiem. Is that a restriction
you could live with in some situations? Now imagine
that you only are charged for the on air time
that you actually speak. Imagine also that the time
you speak can be compressed and delay sent and
instanly received and de-compressed. All of
a suddent you have a near-free communication mechanism without
having to pay for the luxury of speaking and interrupting someone.
This seems to work for most cases of communication (I assume
with the exception of phone sex). Now, imagine a
conference call where the voice gets multi-casted
to all listening participants.

I think that the economics of telephone communications
is going to change radically. The process went from
focusing on the phone, to selling the service and giving
away the phone, and now to absolute billing. I am
starting to see the benefit to the consumer for
per-packet billing.