COMMENT ⊗   VALID 00002 PAGES
C REC  PAGE   DESCRIPTION
C00001 00001
C00002 00002	\chapter {35.}{Networks Considered Harmful for Electronic Mail}
C00012 ENDMK
C⊗;
\chapter {35.}{Networks Considered Harmful for Electronic Mail}

Electronic mail over computer networks has been in use for almost 20 years.
The widespread use of telefax is more recent.  However, unless e-mail is freed
from dependence on the networks, I predict it will be supplanted by the telefax
for most uses in spite of the fact it is more advantageous.  Information is
transmitted more cheaply as character streams than as images, and multiple
addresses are readily accommodated.  Moreover, messages transmitted as
character streams caa be readily filed, searched, edited and used by computer
programs.

Unless e-mail is separated from special networks, telefaxing will prevail
because it works by using the existing telephone network directly.  To become a
telefax user, it is only necessary to buy a telefax machine for a price between
\$1,000 and \$5,000 (depending on features) and to publize one's fax number on
stationery, on business cards and in telephone directories.  Once this is
done anyone in the world can communicate with you.  No complicated network
addresses and no politics to determine who is elegible to be on what network.  
Telefax is already much more widely used than e-mail.  Japan estimates that 5 
percent of the Japanese household population will have telefax machines by 1995. 
By the year 2010, the devices will cost about \$200 and be found in 50 percent
of the homes.

E-mail could work  the same way at similar costs, but because of a mistake
by DARPA aaout 20 years ago, i.e., making a special-purpose, special-politics
ARPANET network the main vehicle for e-mail, it was combined with other network
uses that require higher bandwidth and packet switching.

Another mistake was UUCP.  It uses the telephone network but three features
inherited from its use within Bell Telephone Laboratories made its widespread
adoption a blunder.

\item{1.} It assumes that both parties are using the UNIX operating system
rather than using a general mail protocol.  This is only moderately serious,
because some other systems have been able to pretend to be UNIX sufficiently
well to implement the protocols.

\item {2.} It requires that the message-forwarding computer have log-in privileges
on the receiver.  This has resulted in a system of relaying messages that
involves gateways, polling and complicated addresses.  This reslts in politics in
getting connected to the gateways and often causes addresses to fail.

\item{3.}  Today, forwarding is often a service provided free and therefore of
limited expandability.

\noindent There has been a proliferation of networks and message services on a
variety of time-sharing utilities.  Some of them are commercial and some serve
various scientific disciplines and commercial activities.  The connections
between these networks require politics and often fail.  When both commercial
and noncommercial networks interact, charging complications arise.  A whole
industry is founded on the technologically unsound ideas of competitive
special-purpose networks and storage of mail on mail computers.  It is as
though there were dozens of special-purpose telephone networks and no genera
network.

The solution is to go to a system that resembles fax in that the ``net addresses'' 
are just telephone numbers.  The ssmple form of the command is just MAIL <use>@
#<telephone number>, after which the user engages in the usual dialog with the
mail system.

The sending machine dials the receiving machine just as is done with fax.  When the
receiving machine answers, the sender announces that it has a message for
<user>.  Implementing this can involve either implementaion of protocols in a
user machine or a special machine that pretends to be a user of the receiving
machine or local area network.  The former involves less hardware, but the
latter involves less modification to the operating system of the receiving
machine.
I have heard various arguments as to why integrating e-mail with other network
services is the right idea.  I could argue the point theoretically, but it
seems better to simply point out that telefaxing is already far more widespread
outside the computer science community.  Indeed, it is often used for
communicating with someone who is thought to have an e-mail address when
forwarding connections seems too complicated.

\bf {35.1  The World of the Future}

Eventually, there will be optical fiber to every home or office supplied by the
telephone companies.  The same transmission facilities will serve telephone,
picturephone, telefax, e-mail, telnet, file transfer, computer utilities,
access to the Library of Congress, the ``National jukebox'' and maybe even a
national video jukebox.  In the meantime, different services require
different communicatation rates and can afford different costs to get them.
However, current telephone rates transmit substantial messages coast-to-coast for 
less than the price of a stamp.  Indeed the telefax success, not to mention 
Federal Express, indicates that people are willing to pay even higher costs.

What will happen to e-mail over the next 20 years?

There are two kinds of problems, technical and political.  Guess which is 
easier. The main technical requirement is the development of a set of 
point-to-point telephone maal protocols.  Any of several existing network mail 
protocols could be adapted for the purpose.

Presumably the same kinds of modems and dialers that are used for fax would be 
appropriate but would give better transmisson speeds.

Organizationally speaking, perhaps the simplest solution would be to get one or 
more of the various UNIX consortia to add a direct mail telephone protocol
to UUCP.  Such a protocol wuld allow mail to be addressed to a user-id
at a telephone number.  The computer would require a dialer and a model with
whatever characteristics were taken as standard and it would be good to use
the same standards as have been adopted for telefax.  It must not require 
pre-arrangement between the sending and receiving computers, and therefore cannot
involve any kind of log-in.  Non-UNIX systems would then imitate the protocol.