COMMENT ⊗   VALID 00003 PAGES
C REC  PAGE   DESCRIPTION
C00001 00001
C00002 00002	NOTES FOR THE NSF PROPOSAL ON HOME TERMINALS
C00010 00003	necessary to have input and output equipment that would handle  upper
C00016 ENDMK
C⊗;
NOTES FOR THE NSF PROPOSAL ON HOME TERMINALS

0. General plan
	The work for which we requesting support will consist of  two
parts.   The first will be the development and testing of a number of
home terminal applications.   This  will  involve  placing  terminals
connected  to  the AI Lab's computer in selected places on campus and
in selected homes.  The extent to which  the  applications  are  used
will be monitored, and an attempt will be made to determine the price
a current set of applications can command.

	The second area  of  work  will  be  in  supporting  computer
science and engineering.

	A  long  range  goal  which  is  probably too ambitious to be
realized in the time scale of this proposal is to create a collection
of  services  for  which  there  would  be  a  substantial  number of
customers at a price of $50 to $100 per month and to do the  computer
work necessary for them to be offerable at such a price.

	A major rationale of the proposed research is that computer
systems have hitherto permitted companies to offer the public essentially
unchanged services at a lower cost to the firm permitting higher
wages at not so much increased prices.  The public, however, sees
little direct benefit from the use of computers.  This is beginning
to change as the airline companies begin to calculate fares and
print tickets thus saving the traveller waiting time.  The car rental
computer systems that save filling out the form every time a car is used
are another example of direct payoff to the public.

	However, the main advantages of such systems will come when the
public can use them directly.  This requires a debugged uniform way for
the public to ask questions and make requests that will not have to be
learned anew for every system.  We propose to devise such a way and
experiment with connection to such commercial systems as will let us in.




1. Experimental applications.
	a. improved APE
	b. route finding
	c. reading
	d.  connection  to  at least one reservation system through a
program that simulates an agent.
	e.  some start on the discussion system and journal.  We must
mention Engelbart's work and discuss our different point of view.
	f. calculator
	g. local information system updated by clerk
	h. connection to TELCO directory system, to a bank account system.
The general idea is that we provide a uniform user interface to such
systems and debug the idea of making the systems directly usable by
the public.



2. Qualifications of the lab.
	
	The Stanford Artificial Intelligence Laboratory has existed
since 1963 and its computer facility was established in 1966.  This
facility now contains the following elements relevant to the project
we are presently proposing.
	a. PDP-10 and PDP-6 processors with 256K 36 bit words of
core.  
	b. Swapping disk for time-sharing.
	c. 600,000,000 bytes of on-line storage on an IBM 3330 disk
file.
	d. About 60 keyboard and display terminals in the laboratory
and 3 remote display terminals.  The terminals are in offices rather
than in terminal rooms.  This permits a more realistic test of the
utility of home terminal type facilities than would be possible if
the user had to go somewhere to use them.
	e. A time sharing system that can support about 45 users at
a time.

	Besides the facilities, we have relevant capabilities as
follows:
	a. Our systems programming group developed most of the code
in our present time-sharing system and has developed editors,
compilers, text handling programs.  We developed the first time-sharing
system based on displays rather than typewriters.




3. Starts. APE, reading program, find
		The  idea  of  a  home  terminal is fairly old in the
Stanford Artificial Intelligence Laboratory, and some of us have  had
terminals at home for some time, at first teletypes and later display
terminals.  The paper reproduced as Appendix A was presented in 1970.
The  first  conclusion  stemming  from  our experience is that a bare
terminal in the home connected to a time-sharing  system  is  of  use
only  for  writing  and  debugging  programs.   Use in every day life
requires an extensive collection of programs.

	The first non-programming activity for which we have used our
terminals  extensively  is  writing  and  editing.    For this it was

Feldman remarks: included public terminal both for economics and
because it's a good thing.  Peripheral uses: message center,
reservations, etc. Massive applications like IR and CAI.
CATV
necessary to have input and output equipment that would handle  upper
and  lower  case  without  special  setup, and this was a goal of our
laboratory from the time we acquired our PDP-6 computer in 1966.   It
also  requires  editing  programs that are easy to use.  When we were
using teletypes for input and output, most people who had that option
still  had  secretaries  make the final copies of their writings, but
when we were able to put display terminals in offices,  there  was  a
decisive  change.     Now  almost  all but mathematicians write their
papers directly into the computer.


4. Relevant topics in computer science and engineering.
	a. a general purpose terminal
	b. standardizable set of display commands
	c. standardisable set of commands for human  interaction  and
interaction with files. semantically specified.
	d. proposals for the national file system
	e. a standard commercial interface
	f. proposed standards for documents  in  arbitrary  character
sets.
	g. a paper study of the  computer,  file,  and  communication
requirements for cost-effective home terminals.
	h. universal time-sharing systems.
	i. dial-up communication conventions compatible with the ARPA
network.
	j.  how to make the programs easy to learn to use and easy to
maintain proficiency in.  This probably requires relatively  standard
conventions  on how programs are controlled and an attempt to get the
best advantages of short commands and mnemonics together with  really
good  help facilities.  Our system shall not require six hours of use
each quarter in order to maintain proficiency.

The object of all these standardizability efforts is not to  get  the
standards  accepted although if we are successful enough in designing
them, they might be accepted.  All we really  propose  to  do  is  to
study  the  problem  from  a  non-parochial point of view and to make
proposals that could be standards or at least a basis for  discussion
of standards.

5. Staff
	a. J. McC a small fraction of time
	b. research associate
	c. two or three graduate assistants
	d. one professional programmer maybe two
	e. one clerk

6. Hardware
	a. About six remotable terminals.  These will be good quality
terminals with graphic capability.  No effort will be  made  to  meet
cost  goals that would later have to be met in an economically viable
system of home terminals since the technology to do  this  will  come
later.   If  we  can get them soon, the preference will be for single
Yale type terminals.  Communication facilities must be  provided  for
and we shall probably have to increase the capacity of the PDP-10 for
handling low speed interfaces.  Additional system software  work  may
have  to be done to make sure that light services such as reading and
table lookup on files can be done without  loading  the  time-sharing
system.

7. Budget
	This will be determined by the requirements for equipment and
staff  as  described  above, but the figure to shoot fore is $100K to
$150K in initial hardware and $100K to $150K per  year  in  personnel
and  perhaps $30K per year in subsequent hardware.  Maybe the project
will have to contribute to general laboratory  expenses.   We  should
try  for a three to five year grant which we may try to expand if the
initial results are promising and there are good ideas  that  warrant
it.