COMMENT ⊗   VALID 00002 PAGES
C REC  PAGE   DESCRIPTION
C00001 00001
C00002 00002	COMMENTS ON FARBER AND BARAN
C00009 ENDMK
C⊗;
COMMENTS ON FARBER AND BARAN

by John McCarthy


	I think that Farber and
Baran haven't made the case that communication and
computation are inseparable.  I think they are separable, and it is
important to separate them, because communication is a natural monopoly, 
and computation is not.
However, here are a few points:

1. The battle over the Teletype Model 40 is cited as an example of the
inseparability.  I regard it as a minor issue.  By historical accident,
AT&T has managed to stay in the terminal business, although the regulators
have forced it out of every other business that supplies products.
The Teletype Model 40 is just another terminal.  The telephone network
is no more adapted to it than to any other terminal.  It might be argued
that AT&T should get out of the terminal business, and its competitors
in the terminal business are trying to achieve that result via the FCC,
by citing the Model 40's communication capability.  However, the future
of computer communications depends in no way on whether AT&T is allowed
to sell terminals, so long as it isn't allowed to forbid other manufacturers'
terminals.  In my opinion, AT&T deserves to be allowed to continue in
the terminal business, because the Model 33 Teletype was a major contribution
to time-sharing in its day, and contributions should be rewarded.  The
Model 40 is unlikely to be as successful.  However, whether virtue is
rewarded seems to be a minor issue.

2. It is important that cheaper digital communication become available.
We need transcontinental terminal at reasonable rates.  Such a service
needs to be on a switched network, preferably the ordinary telephone
network between the user and the central office.  It needs high peak
transmission rates (1200 bits/sec is tolerable and 9600 would be nice),
but its average transmission rate can be small, and the transmission
speeds can be different in the two directions.
Ideally, when a telephone is dialed, a special prefix should announce
that a terminal type connection is wanted, and the telephone system
should then route the call through equipment that make such connections
economically.  In that way, any telephone could be used for terminal
communication, even those in telephone booths.

	Demand for terminal communcication will
be very elastic down to $.10 per hour connect time, but any reduction
in the present $35.00 per hour daytime and $13.00 per hour night time
transcontinental rates
will be important.  Because there is substantial demand at high rates,
optimizing AT&T's profits will not necessarily lead to the lowest rates
compatible with reasonable profits.
Moreover, regulators will be inclined to subsidize home telephones at
the cost of pricing digital service out of the market.
Therefore, the public interest
in low terminal rates needs forceful representation before the FCC.

3. The short inquiry is another important service.
This is much more cost-effective in Western Europe than in the U.S.,
because the telephone rates are linear in time used down to a few
seconds.  Of course, the low rates may not hold up, once the service
is used extensively for business purposes.

4. Dialnet is based on the opposite point of view to Farber and Baran.
Its view is that the present AT&T services are underexploited so that
a great improvement in communication can be made without any new
offerings.  As the Dialnet paper shows, the present offering is
quite cost-effective for message and file transfer and for local
telnetting, failing only for long distance telnetting.

5. My one fear is that regulators, inspired by papers like Farber's
and Baran's, will outlaw Dialnet while they contemplate the optimum.
However, I think we will get it going very quickly, and within three
years, it will be too big to step on.  I would welcome any help I
can get in achieving this goal.