perm filename III[S76,JMC] blob sn#218707 filedate 1976-06-07 generic text, type C, neo UTF8
COMMENT ⊗   VALID 00002 PAGES
C REC  PAGE   DESCRIPTION
C00001 00001
C00002 00002	.require "memo.pub[let,jmc]" source
C00014 ENDMK
C⊗;
.require "memo.pub[let,jmc]" source;
.cb III AND OFFICE AUTOMATION


	The object of the this memorandum is to show how III might enter
the office automation field with profitable results.

.bb What is full office automation?

	Full office automation involves the following equipment and
services:

.item←0
	#. Each person whose duties require a desk also has a keyboard and display
terminal.  This includes all secretaries, executives and professionals
who spend even part time at a desk.  The terminal must be capable of
displaying at least 30 lines of 80 characters with upper and lower case
and whatever special characters are normally used in the business.
Since the terminals are used for intra-office communication, it is
important to executives that all their subordinates have terminals.  Once
adequate performance is achieved, the most important characteristic is
cost, so that offices will fully equip themselves.
A user will normally log in to the system in the morning and remain
logged in all day.

	#. The terminals are connected to a time-shared computer system
with cost-effective disk files.  The current imitations of the IBM 3330
model 11 are sufficiently cost-effective, and some of the lower priced
smaller disk systems may be.  Since offices vary in size, it is worthwhile
to be able to handle small ones.

	#. It will be an important plus if facilities exist for message
communication with the rest of the world.  In some cases this may mean
a connection to ⊗Telenet or one of its competitors, but I think the
cheapest system will be an idea being developed by the Stanford Artificial
Intelligence Laboratory called ⊗Dialnet.  The equipment for it is just
a  computer operated telephone dialer rented from Pacific Telephone and
some lines.  With suitable protocols such a system can communicate with
anyone else similarly equipped.

	#. The system needs a printer capable of producing documents of
at least office memo quality.  A printer that can produce report quality
and brochure quality documents will have still wider application.

	#. The first application is keeping current versions of reports
in process and other changeable documents in the system with any user able to
read them at his desk or print copies for use elsewhere.
The commands required to read and print documents will be on the terminal
placard and will be usable immediately without practice or training.
For example, if this document were being developed by several people
into a plan for III entry into the business, it would be important
that many people be able to read its current state and add sections.

	#. Anyone whose job involves preparing or modifying documents
will use the %2editor program%1.  This includes secretaries and
executives.  This is more work to learn, but we have found that
temporary secretaries can work under supervision of more experienced
secretaries with 15 minutes instruction.
Programs for checking spelling are used, and programs for finding
doubtful grammar can be devised.

	#. The next application is message sending.  To send a message
to an individual or existing distribution list takes 5 seconds plus
the time to type the message.  The message will be received immediately
if the receiver is at his desk and the next time he logs in otherwise.
Users accessible by network are treated the same as users in the
office provided the user has the ⊗address or the address is on the
distribution list.  The relative advantages of this and the telephone
are arguable in the abstract, but experience with the ARPA net has
shown that the facility is much valued and heavily used by people
who otherwise don't use terminals.  I receive about 100
messages per month and don't need secretarial help with them, but
people who receive 500 or more ARPAnet messages per month require
such help.

	#. Next we have the preparation of high quality documents.
This requires a program that will compile a %2manuscript%1 into
an output %2document%1.  Our system for this is called PUB.
A text editor is used to prepare the manuscript and include font
specifications, table layouts, justification and other text layout,
symbolic section names, and commands that transfer information
to tables of contents and indexes.  This document is PUB output.
Our secretaries use PUB and enjoy making high quality documents, but
students and faculty also use it directly.  Its advantage is that
it keeps the reproduction cycle for high quality documents very short.
Changes to text can be made and new versions produced in a few minutes.
To be able to update a high quality document in minutes will be
very valuable for sales organizations, law offices, and executive
offices generally.

	Xerox is developing an office system based on the idea that
the preparer of the high quality document can see the finished
product on his screen.  In my opinion, this is excessive gold-plating
at the present stage of technology and is one of the main reasons
why III or other organizations can beat Xerox to the market in
spite of their early start.  Namely it requires them to standardize
on very expensive terminals - perhaps five times as expensive as
what may otherwise be required, and they still won't be able to
handle the very fanciest documents.  Moreover, their commitment
to displaying the document as it is edited gives them a very difficult
software task that they haven't completely solved even after several
years.

	The facilities mentioned above have existed at a number of
the PDP-10s on the ARPAnet for a number of years and have led to
great enthusiasm for this style of work among  many people who don't
otherwise use computers.  The dissatisfactions have mainly concerned
unreliability, slow response when there is too much competition from
computer bound users, and insufficient quantity or quality of terminals.
To my knowledge, no organization has yet got enough terminals and
log-in slots so that everyone can remain logged in all day.  We will
achieve this in another year, but at present, while we have 60 terminals,
only about 35 can be logged in at a time, and this produces an
undesirable strain on facilities.  It is especially important that
the secretaries and managers be logged in all the time, since effective
use requires frequent switches between the terminal and other activities.


.bb TECHNOLOGY

.ITEM←0
	#. The computer can be a mini in speed, but it is probably
unwise to put a large investment in a system with a small address,
because there will be a natural expansion of functions over the
fifteen year period that should be allowed for.

	#. As stated above, IBM 3330 model 11 technology is good
enough.  It would be advantageous to afford a small system.

	#. TV terminals with centralized memory still seem to have
a cost advantage over terminals with local storage.  Stanford is
developing a system based on a bit map that might be used, but
a still cheaper system with centralized character memories might
also be considered.  There must be a compatible terminal that can
be used over a telephone line.  These will be more expensive.

.bb FINANCIAL

	III might find adding such capability to its existing PDP-10
system cost-effective for internal management purposes apart from any
intention to go into the business.  However, faster progress would
require a large investment which I think is warranted.  A project that
invested $200,000 in hardware and six man years in a two year period
might produce a marketable system.
.skip 2
.begin verbatim
John McCarthy
.end