COMMENT ⊗   VALID 00002 PAGES
C REC  PAGE   DESCRIPTION
C00001 00001
C00002 00002				ARPA
C00174 ENDMK
C⊗;
			ARPA
∂21-JUL-75  0545		network site ISI
 Date: 21 JUL 1975 0543-PDT
 From: STUBBS at USC-ISI
 Subject: New Accounts at ISI
 To:   doris
 cc:   Blue, les at SU-AI, stubbs
 
 Doris-
 following just received from Les Earnest -
 18-JUL-75 21:15:40-PDT,661;000000000000
 Mail from SU-AI rcvd at 18-JUL-75 2115-PDT
 Date: 18-JUL-75  2:54 PM PDT
 From: Les Earnest (LES @ SU-AI)
 To:   Stubbs @ USC-ISI
 - - - -
  Dear Gene:
  
  We wish to request two more accounts at ISI, preferably on their "C"
  machine.  I assume that you are still the right person to approach.
  
  These accounts are to be for Brian McCune and Elaine Kant, both of
  whom are Student Research Assistants engaged in research on Program
  Understanding Systems under Prof. Cordell Green, which is sponsored
  by ARPA.  They need access to the Interlisp program, which does not
  run on our computer.
  
  Thanks for your help.
  			Regards,
  			Les
 -------
 
 +*+*+*+*+*+*+*+
 
 If you can help him out please let him know - [Les  at SU-AI]-
 thanx-
 ------
 Les- Doris Smith in ipto is now the Resource Manager and is the one to
 contact in the future for new accounts, usage etc. [Doris at ISI].
 She's a  nice gal and will be very helpful. That doesn't mean tho that 
 we should lose all contact. I'll still be around but in Program 
 Management Office rather than IPT so keep in touch!
 regards
 Gene
 -------


∂20-JUN-75  1502		S,LES
 Dear Lick,
 
 Sorry I missed you on the return call.  As I mentioned, we have four
 new dissertations coming out, but I haven't had a chance to read and
 digest them yet.  There have been a couple of interesting developments
 on computer vision projects still in progress, however.
 
 HORIZON NAVIGATION
 
 We have what I believe is the first example of visual guidance by
 computer in a natural setting.  A computer-controlled cart has been
 programmed to pick out distinctive features on the horizon and to use
 them as an orientation reference.  If it is commanded to move in a
 certain direction, it will head that way for a short distance, then
 visually recheck the horizon features, correct its orientation, and
 proceed again.  The distinctive feature recognizer is quite
 dependable (i.e. it almost never mistakes one feature for another).
 This system works reliably in the roads and fields around the
 Laboratory. 
 
 CAR RECOGNIZER
 
 We have some preliminary results from an attempt to automatically
 recognize automobiles in aerial photographs.  The recognizer that
 we have is rather slow, but it seems to pick out something like
 90% of the cars, with almost no false recognitions.  There are
 several ways in which it can be speeded up substantially and
 work is proceeding.
 
 			Regards,
 			Les
 CC: Licklider%ISI


∂09-JUN-75  1312		network site ISI
 Date:  9 JUN 1975 1312-PDT
 From: DORIS at USC-ISI
 Subject: VARIOUS
 To:   DALE
 cc:   LES at SU-AI, STUBBS, FIELDS, DORIS
 
 
 	1)  PLEASE SET UP A DIRECTORY IN THE "WORLD" GROUP
 FOR <STEINBERG> WITH PASSWORD OF "LOU" CHARGED TO SU ACCOUNT.
 	
 	2)  THE FOLLOWING ACCOUNTS SHOULD BE CANCELLED, WHICH
 ARE ALL LISTED UNDER THE "ARPA SLICE" UNDER NET-CF:
 
 			<APPLE>
 			<CORD>
 			<FRIED>
 			<SHRIMP>
 			<STUFFED>
 
 	IS THERE AN UPDATE ON THE <ULAPA> DIRECTORY?  IF SO,
 HOW CAN I GET IT, IS IT STILL IN THE DIRECTORY OF <TIMESHEET>?
 THANKS IN ADVANCE - DORIS
 -------

∂28-MAY-75  1355		S,LES
 Subject: ISI Account for Lou Steinberg
 Gene,
 
 As you may recall, several months ago we requested an account on one
 of the ISI machines for Lou Steinberg.  When I visited you last month,
 I got the impression that there would be no difficulty in securing
 access, but I have heard nothing further on it.  Since he is eager to
 proceed, I would appreciate it if you could check on the disposition
 of this request.
 			Thanks,
 			Les
 CC: Stubbs%ISI

∂28-MAY-75  1255		S,LES
 Subject: IJCAI Travel
 I'm looking for some general guidance on the handling of requests for
 travel support to participate in the forthcoming IJCAI Conference. 
 
 As things stand, there are apparently eleven people in SAIL who have
 had papers accepted or are invited lecturers or officials of the Conference.
 Even if you would stand for it, we don't have sufficient funds to pay
 the costs of sending them all.  A number of them have requested NSF
 travel grants, but current rumors indicate that the disposition of
 these requests will not become known before mid-June.  Most of these
 people (at least the students) appear determined to go whether or not
 they are funded.
 
 From a budgetary standpoint, it appears that we could afford to spend
 between $4000 and $5000 out of ARPA funds, which would cover about
 three people (@ $1.6k each).  Given my choice, I would like to await
 the NSF evaluation results, then offer at least partial (though not
 necessarily equal) cost reimbursement to those who need it.
 
 My question: does this approach sound plausible, or are there likely
 to be some additional constraints to contend with?
 CC: Licklider%ISI


∂06-MAY-75  1715		S,LES
 Subject: Suggested Additions to AI Roadmap
 
 "Accomplishments" Addendum
 
 Under "Computer Controlled Manipulators" you might add "Stanford"
 to the list of users.  Another sub-item would be "Manipulator control
 languages (WAVE, AL)".
 
 	o  Display terminal systems
 		o  Developed first display-based timesharing systems
 		o  Developed video switch concept
 			o  Shares limited number of display channels
 			   with larger number of terminals
 			o  Low cost/terminal
 			o  Used by Stanford, MIT
 		o  Devised "4-shift" keyboard
 			o  Permits touch-typing of large character sets
 			o  Facilitates "one-stroke" control and interactive
 			   page editing
 			o  Used by Stanford, CMU, MIT
 	o  Display editors
 		o  Text page editors (TVEDIT, E)
 			o  Far more efficient than typewriter editors
 		o  3D geometric editors (GEOMED)
 			o  Interactive design of 3D objects
 		o  Digital logic editors (SUDS)
 			o  Logic drawings, printed circuit cards, back
 			   panels
 			o  Used by Stanford, Digital Equipment Corp.,
 			   MIT, CMU
 	o  Interactive photo interpretation
 		o  Works with satellite photographs
 		o  Automatically locates corresponding features in
 		   photos of similar areas taken at different times
 		o  Locates "variable features" (things that have changed)
 		o  Applied to Mariner IV photos of Mars
 	o  Automatic indexing of text
 		o  Permits rapid retrieval of messages, news, or other
 		   text on the basis of content
 		o  Experimental system at Stanford uses Associated Press
 		   and UPI newswires (soon at NSA and CMU)
 	o  Programming language development
 		o  List processing languages (IPL-V, LISP)
 		o  Recursion in programming (e.g. above plus ALGOL)
 		o  Numerous new concepts in PLANNER, CONNIVER, INTERLISP,
 		   SAIL
 	o  Speech Understanding
 		o  AI research laid foundation for current development
 
 o  ORGANIZATIONAL PLAN
 	...
 	o  Select several application problems ...
 		...
 		o  Data retrieval  -- devise natural language "front end"
 		   for data retrieval systems
 			o  Answers questions about the kinds of data
 			   collections that exist, as well as retrieving
 			   data
 		o  Photointerpretation
 			o  Interactive coverage catalog searching
 			o  Feature-tracing aids for cartography
 			o  New-feature detectors
 CC: Russell%ISI


∂25-APR-75  1623		network site ISI
 Date: 25 APR 1975 1624-PDT
 From: LICKLIDER at USC-ISI
 Subject: AI Roadmap
 To:   Amarel, Earnest at SU-AI, Feigenbaum at SU-AI,
 To:   Newell at CMU-10A, Nilsson at SRI-AI, Winston at MIT-AI,
 To:   Russell, Blue, Carlson, Carlstrom, Fields, Kahn, Walker
 cc:   Licklider
 
 	Sorry to have been so long without communication after
 telling you this roadmap had a deadline of April 19.  Heilmeier
 gave me an extension so I could work on some other
 items with timing tied more tightly to the external world.  I am
 afraid that I do have to let him see this very soon to prove there
 is some movement toward a roadmap,
 but I wish this were much better.  Dave, would you and the other
 guys in the office fix this up in whatever ways come quickly to mind
 and let George know it exists -- and give it to him if he asks
 for it -- but otherwise hold it for further work based on iteration
 with the AI PIs.
 
 	PIs, would you telephone to Dave (I am off for a week of
 vacation in the Arizona deserts) any immediate suggestions that
 you want to get into the roadmap before GHH gets his first
 look at it -- and would you prepare and send by SNDMSG to Dave and
 to me your more deliberate reactions.  I'll be back on May 5th
 and want to go immediately into the next iteration, so please let
 me have some response by then.  I think the roadmap will 
 come into being progressively over a period of several months, but
 the next few weeks will be important because all funding actions
 related to AI will be referred to the roadmap -- and nothing will
 move without it.
 
 	There are some places in this that are pretty bad, I
 realize.  (The characterization of MYCIN, for instance,
 needs to be changed to read "antimicrobial therapy",
 and so on.)  Please contribute suggestions.  We need a full
 list of accomplishments.  I can think of several more -- just
 got started, in fact, but it will help if you will suggest ones
 from your laboratories.  (Limiting the request in that way will
 avoid duplication and the wasted effort that duplication would
 represent.)  
 
 	With respect to scientific/technical objectives, I have
 had some advice to the effect that what I think is required
 cannot be done reasonably or effectively -- but I think
 we should plunge on and try.  The main objection is that one
 cannot foretell how long it will take to have an insight, to
 solve a problem, to find a searched-for key.  Okay, but
 if we can decide what the objectives are, we can fall back
 on something like Delphi for time estimates.  The main thing is
 to try to explain where AI is trying to go, and "forward" is not
 a sufficient answer.
 
 	Well, here is the roadmap as of now, poor thing that
 it is:
 
 
 
 
 
 		ARTIFICIAL INTELLIGENCE ROADMAP
 
 	[First-cut version, April 19, 1975, J. C. R. Licklider]
 
 o  INTRODUCTION
 
 	o  This version must be iterated with
 
 		o  ARPA front office
 		o  AI community
 
 	o  The issues underlying this effort to constuct a roadmap
 	   include:
 
 		o  ARPA has supported AI for 10 years
 		o  ARPA investment in AI > $35M
 		o  Contribution of AI to solution of DoD problems
 		   has been questioned (e.g., by Whitaker of ODDR&E)
 		o  Need to test/evaluate/prove capability of AI to
 		   solve real Defense problems
 		o  Not sound to have such large (75% ?) fraction of
 		   support of AI field come from one source (ARPA)
 		o  Potential of AI to increase effectiveness of
 		   many Defense systems and operations
 		o  Potential impact of AI on US economy
 
 		o  And so the need for a clear roadmap (plan) showing
 
 			o  Application objectives
 			o  Scientific objectives
 			o  Organizational structure
 			o  Test/evaluation mechanisms
 
 	o  Guidance from ARPA front office
 
 		o  Develop roadmap
 
 		o  Increase application-oriented effort
 
 			o  Without increased ARPA/IPTO funding for AI
 
 		o  Get other agencies to support more of AI
 
 			o  Especially more of the basic research
 
 			o  Get help from OMB in arranging it
 
 o  ACCOMPLISHMENTS DUE MAINLY OR IN LARGE PART TO AI R&D
    AND NOW USED BY DOD
 
 	o  Time sharing
 
 		o  Created a 200-company industry
 		o  Now used by many DoD organizations
 		o  Value of time sharing and interactive
 		   computing in software development and
 		   maintenance extolled by FCDSSA
 
 			o  FCDSSA using GENIE system
 			   developed by UCB with ARPA support
 
 	o  Stack (pushdown list)
 
 		o  Basic programming technique
 		o  Now used in almost all programming
 
 	o  Hash coding (associative addressing)
 
 		o  Basic programming technique
 		o  Now widely used in storing data in
 		   primary memory and secondary storage
 
 	o  List structure
 
 		o  Basic programming technique and data 
 		   structure
 		o  Now used in most symbol-manipulation work
 
 	o  Interactive programming and debugging
 
 		o  Now used in several DoD programming shops
 		   (e.g., AFDSSC, FCDSSA) and commercial
 		   computer and software firms (e.g., TRW, CSC,
 		   Honeywell, DEC, parts of IBM)
 
 	o  Mathematical assistant programs
 
 		o  MACSYMA, REDUCE
 		o  Used to aid applied mathematicians
 		o  Automatically solve difficult indefinite
 		   integrals, reduce comples polynomials
 		o  Used by NRL, ERDA, NASA, others
 
 	o  Other specialist problem solvers
 
 		o  DENDRAL
 
 			o  Mass spectrometry expert
 
 		o  MYCIN
 
 			o  Medical diagnosis
 			o  Recommends chemotherapy
 
 	o  Computer controlled manipulators
 
 		o  Scheinman arm
 
 			o  Used by:
 
 				o  JPL
 				o  SRI
 				o  GM
 				o  NBS
 				o  Univ. of Illinois
 				o  Purdue Univ.
 				o  Boston Univ.
 
 o  TOP-LEVEL GOALS OF AI R&D
 
 	o  Understand the processes involved in intelligent
 	   problem solving and decision making
 
 		o  Especially including the effective use of
 		   knowledge
 
 	o  Develop technology base for intelligent systems
 
 		o  Functions
 
 			o  Planning
 			o  Perception
 			o  Reasoning
 			o  Manipulation and locomotion
 			o  Natural language
 			o  Representation of knowledge
 			o  System organization and control
 			o  Learning
 
 		o  Support systems and techniques
 
 			o  Programming
 			o  Computer system architecture
 			o  Interfaces with users, other systems
 
 	o  Construct and demonstrate three kinds/levels of intelligent
 	   system (IS)
 
 		o  Highly specialized for specific applications
 
 			o  Therefore more cost-effective for those
 			   applications
 			o  E.g., weapon and vehicle guidance
 
 		o  Capable of expert-human-level performance
 
 			o  Easily replicated (overcoming shortage
 			   of top-level human expertise)
 			o  Usable in hazardous environments
 			o  More cost-effective than people
 			   if training is taken into account
 
 		o  Capable of performance beyond expert human level
 
 			o  Faster -- for time-critical applications
 
 				o  E.g., target acquisition,
 				   threat assessment
 
 			o  Capable of deeper analysis and use of
 			   more complex algorithms in analysis
 
 			o  More comprehensive knowledge bases
 
 	o  Transfer AI technology into use
 
 		o  Select good application vehicles
 
 		o  Develop, demonstrate, and prove cost-effectiveness
 		   of AI applications
 
 		o  Develop AI expertise in selected DoD and
 		   DoD-contractor organizations
 
 o  ORGANIZATIONAL PLAN
 
 	o  Rename "Knowledge-Based Computer Applications", which
 	   is too long a name, to "Artificial Intelligence
 	   Applications"
 
 	o  Put Commander Hollister in charge of AI when he
 	   reports to ARPA
 
 	o  Set up AI Steering Committee representing potential
 	   users of applied AI and government participants in AI
 	   technology transfer
 
 		o  AI contractors will meet with AISC but not
 		   (so AISC will be an in-government committee)
 		   be formal members
 
 	o  Reorganize AI to improve connection from research
 	   through to appliation and evaluation
 
 		o  Retain basic research capability in universities
 
 		o  Add applications-oriented R&D capability in
 		   FCRC(s) and/or an institute 
 
 			o  E.g., RAND, Aerospace, Mitre, USC-ISI
 
 		o  Add applications capability in industry
 			o  E.g., TRW, Boeing
 
 		o  Add appliations capability in Service system-
 
		   development organizations
 
 			o  E.g., NELC, ECOM
 
 	o  Select several application problems as foci for integrated
 	   research-application-evaluation projects.  E.g.,
 
 		o  Morse Code -- reception in presence of noise and
 		   other-code interference -- including encyphered/
 		   encrypted code
 
 			o  Work now under way at MIT
 
 		o  Maintenance of vehicles or weapon systems --
 		   based on monitoring of records and indicators --
 		   includes diagnosis/prognosis with aid of computer
 		   models and data bases
 
 			o  SRI exploring this area -- main SRI project
 			   is closely related: computer-based
 			   consultant on maintenance of electro-
 			   mechanical equipment (which is an AI
 			   application of very high potential impact
 
 		o  Computer-aided command -- including threat
 		   assessment, selection of strategies and tactics,
 		   allocation of forces, monitoring of battles, and
 		   command decision making
 
 			o  Fleet C-cubed program under discussion with
 			   NAVELEX and NELC
 
 			o  Planning workshop scheduled for June 2-4
 			   at NELC -- will involve fleet, Navy research,
 			   and ARPA AI-contractor people
 
 			o  Army C-cubed research also under discussion
 
 		o  EW -- rapid analysis of signals, identification of
 		   equipment and tactics, and selection and 
 		   implementation of countermeasures
 
 			o  EW applications planning contracts
 			   being negotiated by Avionics Lab for ARPA
 
 		o  Natural language -- monitoring for key words or 
 		   phrases, identification of language being spoken,
 		   recognition of particular speakers
 
 			o  These problems will be addressed in the
 			   Speech Processing program
 
 		o  Intelligence Analysis
 
 			o  This application area is the focus of much of
 			   the work in Intelligent Terminals and Very
 			   Large Data-Base Systems
 
 		o  Photointerpretation
 
 			o  This area is being addressed by the Image
 			   Understanding program
 
 		o  ASW
 
 			o  IPTO will support TTO in this area
 
 	o  Analyze application requirements in selected
 	   problem areas to determine scientific/technical requirements
 
 	o  Analyze ongoing research efforts and define their
 	   objective structure 
 	o  On basis of results of two preceding steps, define objective
 	   structure for AI (including both scientific/technical
 	   and application milestones) for the next 3-5 years
 
 	o  Have the AISC hold program reviews at frequent intervals
 	   (every two months at first) to monitor and guide the
 	   shaping of the reorganized AI program
 
 [More to come.  The next part will try to deal with the
 objective structure in a substantive way but will be only a place
 holder for the analysis mentioned in the foregoing.]
 
 				J. C. R. Licklider
 				At this point on April 25, 1975
 
 
 
 				Regards
 
 				Lick
 -------

∂18-APR-75  1435		1,QIB
 Dr. Licklider's secretary, Hilda called and wanted to know when you
 were going to see Dr. Licklider - as she wanted to arrange his schedule -
 she thought you were coming back to Washington.  Said to call Monday
 as she was on her way out for the rest of the day.  (202) 694-4001.


∂14-APR-75  1321		S,LES @ ARPT
 ∂25-MAR-75  0623		S,LES
  Subject: Request for Computer Account
  Dear Gene:
  
  I wish to request that one of our Student Research Assistants, namely
  Louis Steinberg, be given access to one of the Tenex machines at ISI.
  He is working for Cordell Green on Program Understanding Systems,
  under ARPA sponsorship, and needs some Interlisp service in support
  of that work.
  
  Our computer system does not yet support Interlisp and probably won't
  for a year or so.  If it would help for load leveling, Steinberg
  could restrict his work to certain specified hours of the day or
  night. 
  
  I realize that those machines are fairly heavily loaded already, but
  if you can reasonably squeeze on another one, we would appreciate it.
  
  		Regards,
  		Les
  CC: Stubbs%ISI
 
 
 CC: STUBBS%ISI


∂16-APR-75  1002		ESS,JMC
 I agree with the content of your message.  I can't go into detail, because I'm
 returning to Japan in 10 minutes.  However,
 	1. Why don't we suggest a two weak AI applications study group
 in Washington this summer.  The first two in August would suit me, but
 there could be others from Stanford.
 	2. I agree that Heilmeier's suggestions are dubious, but I got
 an idea that maybe his attack on AI has run out of internal steam and
 all he wants now is some kind of symbolic redirection.  As you say, it is
 hard to imagine enlarging the Morse effort, and the language recognition
 is for the speech group, leaving ASW as the only possibility which might
 be hard to declassify.
 	3. We have decided to pursue the car-finder that I mentioned
 without further ado in the hopes of having a silver bullet in three
 months.  I don't know if you remember what I said about my meeting
 with Tactical technology in which they said they would be impressed by
 such a demo into thinking that scene analysis might have target finding
 applications.
 	4. I am a bit happier with the list of AI science topics that
 Russell put on the board than I am with the list in your memo, because in
 xx it includes the problems that interest me.  In general, if Russell gets
 his way, we don't have a qualitative problem we can't adjust to - only
 some budgetary problems.
 	5.  If someone had the time, the best use of it in the next week
 or so (sorry about weak earlier in the message) might be to put flesh on
 the list of scientific problems that Russell posted and to formulate
 a proposal for the summer study that a number of groups could agree to
 including how to sanitize classified info.
 CC: newell%CMU


∂16-APR-75  0956		network site CMUA
 **** FTP mail from [A310AN02] (NEWELL)
 o: Licklider @ ISI
 rom: Newell @ CMU-10A
 ate: 16 Apr 75
 e: AI ROAD MAP MEETING
 C: Carlstrom @ ISI, Fields @ ISI, Russell @ ISI, Simon @ CMU-10A
    JMC @ SU-AI, LES @ SU-AI, CCG @ SU-AI, PHW @ MIT-AI, 
    Feigenbaum @ ISI, Amarel @ ISI, Nilsson @ SRI-AI, 
    Sacerdoti @ SRI-AI
 
 ick:   Twenty-four  hours  has  permitted some  reflections  on
 onday's session and the problems attendent thereto.
 
 1)  To  repeat  what was  generally  accepted,  implicitly  and
 xplicitly, by  the AI  scientists at  the meeting:       It  is
 ppropriate, given the current general context and specific ARPA
 ontext, for  the AI field to attempt  a series of applications.
 uch  applications would  be a good  thing for  AI generally, as
 ell as for its specific relations with DoD.
 
 2)  Also  to repeat:       Application  opportunities  must  be
 iscovered, verified  and exploited.    Some institutional means
 ust  be found to  expedite this.  For  it is clear  that the AI
 ommunity  by  itself  does  not  have  the  expertise  nor  the
 onnections  to find high payoff  activities.   Dave Russell, at
 he end of the day,  strongly posited a mechanism of a Rand-like
 gency  with the  mission of  finding application opportunities,
 erifying them,  building a bridge to the  AI Labs, etc.   There
 ould also  exist, as  an adjunct  to this,  an AI  Applications
 echnical   Group   (or  some   such   title),   consisting   of
 epresentatives  of the  various Labs, CMU,  MIT, SRI-AI, SU-AI,
 U-HP, plus  maybe others in related  programs, such as BBN-SUS,
 DC-SUS, Amarel, etc.  This group would  be  a prime  forum  and
 nitiation point for these applications.        I expressed some
 oncern that such an agent could come into being in short enough
 rder to  satisfy  the  needs of  the  day (implying  that  some
 emporary vehicle would have  to be erected), but Russell seemed
 onfident that  such expedients  were unnecessary.  It would  be
 etter his way.
 
 3) It  is extremely important to be sure  that the payoffs of a
 pecific  application  are  real.     It  is  too  easy  to  get
 andbagged  to have a  seeming application turn  to dross. Given
 hat  ARPA is  prepared  to  spend  large fractions  of  its  AI
 ommunity (a  relatively precious  resource) on  producing  some
 pecific  applications, it is critical  to substantiate the need
 nd acceptability  of  an  application.    ARPA  itself,  though
 nside the DoD and much closer to the application sites than the
 I   community,  does   not  itself  have   the  expertise  and,
 mportantly, the  time to examine the  situations enough to make
 eliable assessements.
 
 4) To be concrete, on reflection I am not at all convinced that
 he items on Heilmeier's list  are all really in the category of
 enuine  application opportunities.    I cannot  speak about the
 SW  problem, since that  is not  a single problem,  but many --
 amely, the question of where  to apply AI techniques all up and
 own  a complex  system.  But the  other two  are bitty problems
 imed at  highly specific targets. I did not  get any sense that
 PRA really  knew in  detail whether  the payoffs  were real  or
 imply  ephemeral, momentary opinions  of one or  two high level
 eople in  the organizations  connected with  the  applications.
 eilmeier's carriage-trade  philosophy  requires a  really  good
 arketing  and marketing research  arm if it is  to succeed.  It
 ill do ARPA no good if it squanders its substance on a bunch of
 rrelevant  mirages.   The  military scene is  littered with the
 ead bones  of expensive solutions which  were only monuments to
 omeone's folly.
 
 5) It appears that MIT-MAC is already spending substantial sums
 relative to the apparent size of the problem) on the Morse-code
 roblem.  Surely  it would not be fruitful  for the AI community
 o  get further involved in  that one.  I am  not quite sure why
 his  one showed up on  Heilmeier's list, but maybe  it was just
 eant to be illustrative of what ARPA had no decided to do.
 
 6)  On the language-spotting  task, I need to  reiterate what I
 aid  at the  meeting.            First, I  do not  think the AI
 abs (in counterdistinction to  the SUS Labs) should take on the
 roblem,  they  simply  would  have to  build  up  much  of  the
 xpertise that  the SUS  Labs  have, which  would be  a  genuine
 uplication.  Second,  I consider  that the  SURG has  not  been
 sked to consider taking on that task. I agree that you possibly
 entioned   it  to  the   SUSC  (though  I   have  forgotten  it
 ompletely), but since it was not brought up in the context of a
 erious confrontation  with the SUS 5-year goals,  I do not take
 t as a serious proposal. It would have been derilict of any SUS
 roup to take on the  task, given how tightly the SUS program is
 trapped down to the 5 year goals.
 
  do  believe we can consider this task  and, if it is important
 nough, we can consider  folding it into the present contractors
 n some way.  But we do have to face the potential effect on the
 -year goals and to see how to work around them.  As chairman of
 he SUSC, I  am quite willing to go around  on that issue, but I
 eed a signal from you  or Dave Carlstom that indicates you want
 o  do that. This is  a serious point and  proposal I am making,
 ince I do not want  to be accused later of having fiddled while
 ome burned -- of having  not picked up this problem when it was
 mportant to  do so.      I  would like  some feedback  on  this
 pecific  matter.            The  problem  itself  seems  rather
 traightforward, given the current art.  I would estimate half a
 an  year for the technical  work, if done at  CMU given all the
 acilities.  To this must be added the whole custumer interface,
 hich might be as much again. Much depends on details which I do
 ot  know, of course (eg,  how must it be  packaged and how much
 ust it cost).  If the problem could wait until after Nov76, you
 ould  surely get  it taken  on by  the SUS  Labs if  it were as
 mportant as Heilmeier stipulates.
 
 7) Another  example of a SUS-related  application is the Korean
 n-line  communication aid,  which you  raised as  a problem and
 hich I  suggested a solution approach to some  time ago.   I do
 ot know  what became of that.  Again, it  would be derilict for
 he  SUS community to  fold that in without  at least explicitly
 acing the 5 year goals. All this stem, of course, from the fact
 hat  we  (read: ARPA  and  possibly  Newell-cum-SURG-initiator)
 anted  a  program   tightly  fixed  on  impressive  goals,  and
 herefore not  with much  slack  for such  things. Again,  I  am
 illing to consider this.
 
 8)  I cannot  believe that the  CBC is not  on the  track of an
 mportant  application.  It  has two  things wrong  with it: (1)
 here is  not an immediate customer eager and  ready to pay; (2)
 otwithstanding SRI's  search, there  does not  seem to  be  one
 iding out  there quite yet (though I do  not know how intensive
 hat search was or is).   Yet, it does not seem to me profitable
 - for  ARPA, even  on its  own current  terms --  to jerk  that
 ffort  up at  the roots and  radically redirect  it.  Rather it
 eems to me critical to widen the scope to "Real-time operations
 onsulting" (namely, how to  help someone carry out an operation
 n real  time) and to search for  applications within this wider
 phere.   The core of work  on the CBC remains  in fact strongly
 elevant; and the new applications can be grafted on.
 
 9) What  should go in the  Road Map? It seems  clear to me that
 he  Road  Map  for Friday  has  its  action  component  defined
 ndependently of  its  substantive  component.     To  wit,  the
 ormation  of the application-finding  mechanism, defined above,
 ill  not be  justified, nor  require justification,  from the a
 tatement of the current art or a statement of future scientific
 oals.  However, this application proposal  will differ from all
 ther such attempts by  the promise, implicitly extracted at the
 eeting, by the AI Labs to enter into such an application-search
 holeheartedly.
 
 he  substantive mode must  perhaps still be  there by Saturday.
 ou don't have much to work with, in terms of what was generated
 efore and during the meeting.  Thus, I would attempt to get the
 ction component to stand in  for the rest.  Let me discuss each
 f  the substantive components  a little, and then  come back to
 his.
 
 10) When a set of the worlds best scientists, being asked about
 he  their very  own scientific domain,  becomes tongue-tied and
 roduces answers unsatisfactory  in a first year qualifier, then
 he conclusion is not that the science doesn't exist, it is that
 he  question  was  posed wrongly  or  the  situation  inhibited
 dequate  response.    You  asked us,  I think,  to do something
 nder constraints that communicated:  (1) that none of our prior
 ttempts was to be considered satisfactory -- that something new
 nd  different  was  required;  (2) that  we  adopt  a  form  of
 pecification of results and  of expectations that is foreign to
 S and  AI, and largely foreign to science  (namely, to state in
 dvance the content of  the scientific results to be expected up
 o several years in the future, so that the questions would only
 e whether or when the result would be attained.  This pre-empts
 he science and leaves us tongue-tied).
 
 or instance, in the CMU proposal I have just finished writing a
 tatement  about the  basic scientific  questions of  AI and the
 igh level propositions that characterize what we have found out
 n  AI.  Apparently that is  to be discarded as  not adequate or
 ppropriate  to the  task  --  and  I  am to  find  yet  another
 tatement,  different from that,  that is to be  adequate to the
 ew (yet identical) task.
 
 or instance, Nils has  just finished writing a paper (IFIPS 74)
 evoted to a  summary of what AI has done  and what areas it has
 orked in. Apparently that is to be discarded as not adequate to
 he task -- and Nils  is to find yet another statment, different
 rom  that, that  is to  be adequate  to the  new (yet idenical)
 ask.
 
 et  me   strongly  suggest,  for  instance,   that  as  far  as
 haracterizing the  present state is concerned,  you take a copy
 f Nils  IFIPS  paper and  underline in  red  the items  in  the
 ibliography that are done in  the ARPA AI Labs, and on the many
 harts  that draw a  map of the  area, circle in  red these same
 tems.  This will give (1)  a direct picture of the coverage and
 cope of the field of AI that ARPA has given birth to; and (2) a
 irect  picture of the  extent to which ARPA  is responsible for
 hese results and for the important ones.
 
 et me  further suggest that you put in  front of Nils paper the
 irst  section on AI  goals from  the CMU proposal,  as giving a
 igh level coherent picture of  what AI as a science is striving
 o achieve and what in global terms it has found out.
 
 hese two items answer  only the question:  What are the results
 n AI in its own scientific terms.  They do not answer it fully,
 ut they will do as well as what you can put together in yet one
 ore attempt in a few hours.
 
 12)  I do  believe that  several additional  descriptions of AI
 cientific results  are possible  that will  appear to  be  more
 atisfactory to upper ARPA than the two above items (if, indeed,
 nything  is   satisfactory).    I   cannot  carry   out   these
 escriptions in the time available, indeed I think it would take
   couple of months of  very hard work (maybe  more).  But I can
 ketch and  illustrate one part of it (which  is indeed based on
 ast efforts to systemitize).
 
 rogress in  AI proceeds  in terms  of increases  in  scientific
 nowledge about the various components of the intelligent agent,
 omponents that are  defined functionally.  A standard division,
 hich corresponds in part to Nilsson's core areas, is:
 
 > Recognition and description (Perception)
    > Vision
    > Speech
    > Language
 > Representation
 > Problem Solving Methods
 > Control Structure
 > Assimilation & Accomodation (Learning)
 
 ithin  each component one  can describe a  series of structures
 or mechanisms) that  are possibilities for this component.  The
 iscovery of each such structure and mechanism is an advance for
 I and  a  result.  Verification,  of  course, is  required;  it
 omes,  usually, from  incoroporation in  several total systems.
 nowledge  about each  mechanism grows  with experimentation and
 heoretical  sharpening.  Such knowledge,  again  when  verified
 xperimentally,  constitutes  scientific results  for  AI.    It
 onsists mostly  of  statements of  adequacy or  sufficiency  in
 pecific task environments.
 
 hus, the statement "What are the results of AI" at a given date
 s  a listing  of the  various mechanisms  (usually described by
 onventional technical names), plus the associated statements of
 dequacy.  This  list grows  over time,  and it,  rather than  a
 arametrization of how good are the systems that can be produced
 onstitutes the core transferable knowledge of AI.  This core is
 ndeed  transferrable,  precisely because  it  consists  of  the
 bstracted mechanisms which have been shown experimentally to be
 seful in several task environments.
 
  cannot  produce the  lists  of results  for the  total  field,
 ostly  because  they  have not  been  extracted,  labelled  and
 rganized  in this way.   I can do  it for one  subpart, that of
 roblem solving  methods.  Here, much that we  know can be given
 y  specific methods  (analagous  to  the methods  of  numerical
 nalysis).  A fairly good list is:
 
 > Generate and test
 > Hill climibing
 > Heuristic search
    > Search stragegy:
       > Depth first, Breadth first, Best first, 
         Progressive Deepening
    > Evaluation
       > Evaluation functions, level of aspiration, 
         duplication avoidance, external limits
 > Matching
 > Hypothosize and match
 > Means ends analysis
 > Substitute & eliminate
 > Range restriction
 > Abstraction planning
 
 o  find a  short  way  to say  what  we  know, eg,  about  Hill
 limbing, takes more energy than I have at this wee hour.  We do
 now  the major  things  to  beware of  (Multi-modality,  Mesas,
 idges, Cliffs),  we do have some empirical  things to say about
 hen  Hill Climbing seems  to work  and when it  doesn't.  We do
 ave a  way of  classifying the  refinements of  the method  (as
 implifed models  of the  hill, which  are used  to predict  the
 ptimum hill-step to take). And so on.
 
 he  existence  of  this  list implies  a  large  kit  of  tools
 vailable  to be  used  in  applications, and  indeed,  when  Ed
 eigenbaum  says they  used "standard  AI" in  Dendral, he means
 hat  the techniques in  Dendral pretty much are  drawn from the
 ist of such known and characterized methods.
 
 o  plot the  success of  the Problem  Solving Methods component
 ver time  is  to watch  this list  grow  and/or the  amount  of
 nowledge about each such component grow.
 
  do  not know how much  each of the other  components can be so
 haracterized, though I expect it could be pushed quite far. But
 ot tonight!
 
 13) Future  goals  of  AI must  be  in  general to  extend  the
 echanisms and structures of each area and to show that they are
 dequate to wider and  more difficult problems.  The discover of
  new method  is not to be asserted in advance  as a goal, if it
 ould then the new  method would have been found!  Sometimes one
 nows  enough about a method  or structure to specify  as a goal
 hat it is  to be explored. To find the  true scope of the range
 estriction method, is such a (small) goal.
 
 oals, in the sense that you (read: ARPA) want them, are only to
 e associated with systems or with instruments (Physics has such
 oals  for the  energy  of  interactions its  accelerators  will
 each,  or the resolution  of microscopes).  There  has not been
 ny  difficulty,  as  far  as I  know,  in  determining  how  to
 arametrize the  structure and  performance of  specific  narrow
 lasses of  systems (such as SUSs  or Dendral-like systems) when
 he task  comes close  enough to  feasibility to  make it  worth
 dopting  a system  as an  AI goal.  It  can surely  be done for
 arious  other specific classes  of systems, though  it will not
 over, thereby, all of AI's goals.
 
 I have run out of gas  here -- I believe more can be said about
 tating AI pure goals, but it just escapes my fog-bound mind.  I
 ove on.]
 
 14)  Applied goals, I  believe are to  be stated in  one of two
 ays.  These  provide other ways of describing  what AI has done
 n terms of how it can contribute to these goals.
 
 ne  way is  how we  started out  to do  it at  the meeting.  An
 pplied system is posited  (ie, a total military system, such as
 n ASW system). Then,  within that some points of AI application
 re found,  from which one attempts to derive  the AI that might
 ake a difference.   This is a form of systems analysis, and one
 hat can lead to a backward chain of available relevant research
 esults and,  as  well, of  still  needed research  --  methods,
 tructures, knowledge, experimentation,  , etc. needed to do the
 ob.  I  think we should do a substantial  amount of this, and I
 elieve  quite satisfactory  road-map results would  come out of
 t.    Unfortuneately,  I believe  that the  effort per  complex
 otal  system must  be a summer  study-group sort  of thing, ie,
 bout  what we  put into  the SUS  initial report.   But this is
 xactly what  is to be done by this  Rand-like agency (plus some
 f us).
 
 he second  way is  to specify  applied technologies.   The  two
 rototypic examples  at hand are the notion  of a SUS technology
 nd  (more pertinent) the natural  language front-end technology
 hat we all were talking about.  One can take the development of
 uch a  technology as a goal and describe  both what existing AI
 lready provides  and what  new research  is needed  to get  it.
 his can  be carried  out  much more  within the  AI  community,
 hough some  sense for what is really  required to make specific
 pplications go  is important. But again, it  takes a fair sized
 ffort to lay out  such a technological alternative.    We could
 ommission such  explorations.   (It would depend,  I guess,  on
 pper ARPA being prepared to consider such expansions.)
 
 15) We did assert rather strongly that there have been a number
 f  civilian applications of  AI, eg, in  management science, in
 esign, etc.  Ferreting these out and asking whether any of them
 ould be  applied to  military systems  would be  an  additional
 mportant  task for  this applications  organization (along with
 ome of us). This, of course, is yet one other way of stating AI
 esults.
 
 ******
 
  am really  to the end of my rope  tonight (this morning) and I
 ill send this  out after Herb gets a chance  to look it over. I
 m willing to work on  expanding or modifying any piece of this.
  am pretty much around from here through the weekend.
 
 .N.
 
 .S.   Recall  that  I am  expecting  feedback on  the  Language
 potting issue.
 

∂10-APR-75  2236		S,LES
 Regarding your planned visit on the 23rd, everything is fine except
 for a conflict with David Luckham's attendance at a major software
 conference in Los Angeles, April 21-23.  Luckham would like to make
 the third day there, but is willing to return early if you would
 like to learn about our Automatic Deduction program at this time.
 
 Alternatively, he is likely to be visiting Washington soon and would
 be happy to go over his program with you then.  We will be guided by
 your wishes in this matter.
 
 				Regards,
 				Les Earnest
 CC: Russell%ISI


∂11-APR-75  2131		S,LES
 AFTER DISCUSSIONS WITH AL NEWELL AND JOHN MCCARTHY, WE CONCLUDE
 THAT HE SHOULD COME TO THE MONDAY MEETING.  HE EXPECTS TO BE
 THERE WITH PROBABILITY .99 .
 CC: LICKLIDER%ISI


∂10-APR-75  1957		network site ISI
 Date: 10 APR 1975 1956-PDT
 From: LICKLIDER at USC-ISI
 Subject: Should McCarthy Come Back from Japan to Attend Meeting?
 To:   Earnest at SU-AI, Newell at CMU-10A
 cc:   Licklider
 
 	This evening John called from Japan.  I told him I could not
 quite come to the point of suggesting (or agreeing) that he should
 come all the way from Japan to attend the meeting here Monday on
 ARPA's changing attitude toward AI.  I said I would get your advice,
 Allen, and talk with Les again (I talked with him on the phone this
 afternoon on the same subject), and John said he would call Les
 tomorrow before deciding whether to return or not.  The reason I
 want to bring you in on this, Allen, is that you know a lot about the
 situation and see it from a university PI's point of view.  If you
 think John should come back, that will switch me over to that con-
 clusion.
 
 	This evening, after talking with John, I called Allen and
 found, as I should have known, that he was enroute to SF.  I then
 called you, Les, and left word asking you to call Allen at the SF
 Airport Hilton.  If you do that, and put together the results of it
 with those of the conversation with me this afternoon and with this
 rather complicated message, you will have the basis for advising
 John when he calls.
 
 	The connection with Japan was fantastic, and I am entertaining
 the hypothesis that John is not in Japan at all -- but maybe hiding
 out in east Rosslyn.
 
 				Regards
 
 				Lick
 -------

∂10-APR-75  0404		network site ISI
 Date: 10 APR 1975 0404-PDT
 From: LICKLIDER at USC-ISI
 Subject: AI "Roadmap" Meeting
 To:   Amarel, Earnest at SU-AI, Feigenbaum, Newell at CMU-10A,
 To:   Nilsson at SRI-AI, Winston at MIT-AI
 cc:   Licklider
 
 	Apologies for my falling behind schedule in defining
 the arrangements.  Evidently, no date suits all, and Monday (Apr. 14) suits most -- all but
 Nilsson.  Nils, could you (since you, yourself, cannot make it on
 Monday) send Bert or Peter with Sacerdoti so we can go
 ahead with the meeting?  Pending the firming of that way of handling
 the problem, would you all hold Monday (Apr. 14) for the meeting.
 Also, please let me have your individual, preliminary versions of a
 roadmap by tomorrow.
 
 	There is no major change in the situation here.  We have not
 yet heard from the Apportionment exercise.  I still consider the
 problem serious and demanding of real decision making and response from
 the community but not the occasion for
 panic or precipitous action.
 
 	Looking forward to seeing you.
 
 				Regards
 
 				Lick
 -------

∂10-APR-75  1333		network site ISI
 Date: 10 APR 1975 1323-PDT
 From: RUSSELL at USC-ISI
 Subject: PROGRAM REVIEW SCHEDULE, 21-23 APRIL 75
 To:   PIRTLE at I4-TENEX, CRACRAFT at I4-TENEX, MADDEN at OFFICE-1,
 To:   RDA at I4-TENEX, NILSSON at SRI-AI, LES at SU-AI, WESTDIV,
 To:   TOM at CCA
 cc:   HEILMEIER, LICKLIDER, HYDE, RON at I4-TENEX, BLUE, FIELDS,
 cc:   CARLSON, HILDA, RUSSELL
 
 CONFIRMING OUR PREVIOUS DISCUSSIONS, I AM PLANNING THE FOLLOWING
 SCHEDULE OF VISITS TO THE SAN FRANCISCO BAY AREA THE WEEK OF 21 APRIL
 1975.  PLEASE ADVISE IF THIS FINAL VERSION CAUSES ANY PROBLEMS.
 	
 MONDAY, 21 APRIL 1975
 	
 0830 - 1230	SRI  - AI
 		(ACCOMPANIED BY ROBERT MADDEN (NSA))
 	
 1330 - 1800	ILLIAC IV RELATED PROGRAM REVIEWS INCLUDING -
 	
 		     I4 SECURITY - HAWAII, WESTDIV
 		     IVTRAN - MCA
 		     DATACOMPUTER - CCA & AFGWC
 		     BENCH-MARKS - RDA
 	
 	
 TUESDAY, 22 APRIL 1975
 	
 0900 - 1500D	ILLIAC IV
 		PROGRAM TUTORIAL AND REVIEW FOR NSF & NSA
 		(ACCOMPANIED BY EUGENE BIERLEY OF NSF AND ROBERT
 		MADDEN OF NSA)
 	
 		(FOR MEL PIRTLE:  PLEASE SCHEDULE 30 MINUTES FOR
 		BIERLEY, GATES AND ME TO MEET WITH DR. MARK)
 	
 1500 - 1700	RAND CLIMATE DYNAMICS, ILLIAC IV PROGRAMMING
 		REVIEW
 	
 		(FOR TOM CRACRAFT:  PLEASE MAKE ARRANGEMENTS WITH
 		 RDA & MCA TO ATTEND)
 	
 WEDNESDAY, 23 APRIL 1975
 	
 0900 - 1400	STANFORD AI
 	
 DAVE RUSSELL/ARPA/IPTO
 -------
 


∂07-APR-75  0234		S,LES
 Subject: Response to Easter and Roadmap Messages
 We greatly appreciated receiving your perspectives on the current
 ARPA environment.  Your perseverance in the face of a very difficult
 situation is remarkable. 
 
 A note on message addressing: the problem in getting the Easter
 message to me was that our computer thinks I'm really "Earnest". 
 
 			ROADMAP
 
 Regarding the proposed roadmapping project, we will give it our best. 
 I can come anytime in the period April 11-16.  I have a Soviet
 visitor scheduled the 16th, but could find someone else to handle
 that.  I plan to choose our second participant from among Luckham,
 Green, Binford, and Winograd as soon as I get a chance to talk with
 them. 
 
 I am worried about the kinds of questions being asked, not because
 there aren't plausible answers, but because the way they are being
 asked fits a world model that predicts further doom.
 
 Frankly, we're all a little low on energy at the moment, being still
 in the final stages of choosing and informing the students who are to
 be layed off.  Though I feel bad about having to do this, I do not
 have a sense of persecution.  This Lab has enjoyed 11+ years of stable
 support from ARPA without a single prior slash (but with a fair
 amount of downward pressure on the lid).  That is a remarkable
 record for relationships of this kind.
 
 You mention that "Intelligent Terminals already has a definite plan". 
 We would very much like to see something like that, both as a sample
 and to help identify crossties.  Other relevant plans would be
 Software Technology and Image Understanding, if they exist.  Any
 chance of getting a look? 
 
 			EASTER
 
 Based on your analysis of the situation, it appears that we (SAIL)
 should attempt to do three things concurrently, as follows.
 
  1. Continue trying to establish the connections between perceived
  DoD goal structures and our programs.  As always, this can involve
  modifications to the programs, to the descriptions of the programs,
  and to the perceived goals.
 
  2. Take some of the things that we were planning to do on a limited
  basis as support activities and push them harder as separate goals. 
 
  3. For any programs that appear to be "off the road" from your
  viewpoint, we should seek alternate funding sources.  Clearly we have
  the main responsibility for doing this, but we would appreciate any
  help and guidance that you can provide. 
 
 In the balance of this note, I offer some specific remarks under 1 & 2.
 
 		1. The DoD Connection
 
 You attribute to GHH the view that DoD should not be supporting
 formal reasoning because it is applicable to mathematical theorem
 proving and checking.  We certainly have not advocated it for that
 reason, but rather for its application to certifying program
 correctness.  Surely it is not considered bad that projects have
 beneficial side effects in other fields.  Assuming that GHH
 appreciates the importance of software certification, I conclude that
 we have failed, somehow, to establish credibility that these goals
 can be achieved in a reasonable time.  I think that there should be
 room for a reconciliation of interests here. 
 
 A recently initiated subtask that we neglected to mention in our
 proposal is formal verification of digital logic design.  That is,
 one of our grad students plans to couple representations of actual
 digital logic circuits with formal reasoning techniques to prove
 correctness (i.e. that the designs meet their specifications). 
 
 		2. Pushing Support Subgoals
 
 		2.1 Intelligent Terminals
 
 We have developed an excellent keyboard and text editing system for
 local displays and would like to develop a similar system for remote
 displays.  The same is true of our display-oriented news service
 (currently working with Associated Press and New York Times
 newswires).  Both of these tasks might reasonably mesh with your
 intelligent terminals program. 
 
 Incidentally, we received an inquiry a week or so ago from a group at
 NSA.  They would like some help in adapting our News Service program
 to their needs and appear willing to provide separate funding.  In
 view of recent developments it might be better to tie this more
 directly to the IPTO program. 
 
 		2.2 Documentation
 
 We would like to continue the development of representation schemes
 for high quality documentation.  With the advent of large data base
 systems, the question of how to do this efficiently becomes rather
 important.
 
 Surprisingly, there appear to be no examples of computer
 representations of complete documents (including text in various
 fonts, sizes, and orientations, line drawings, photographs, etc.)
 that are also reasonably efficient.  We need to develop both
 low-level, device-dependent representations (e.g. for XGPs and
 similar raster printers and for various displays) and flexible higher
 level representations that permit presentation on various devices. 
 
 We have been interested in this problem for some time and have done
 some of the pioneering work.
 
  a) Our (higher level) document compiler, PUB, was the first to go
  beyond traditional text justification to add symbolic crossreferences
  and automatic compilation of tables of contents and indexes.  This
  work has been substantially extended by groups at CMU and Xerox PARC.
  One of our (unpaid) staff members has developed another compiler,
  called POX, that is remarkably fast, versatile, and uncomplicated. 
 
  b) Members of our staff have developed several kinds of specialized
  graphics editors: e.g. GEOMED (for 3D objects), SUDS (digital logic
  drawings, printed circuit card design, etc.), and FLOW (flow chart
  layout).  We see a need for both more general graphics editors (able
  to handle a wider range of tasks) and more complete ones (having
  specific "built-in" knowledge of a problem domain).  Such editors
  should have outputs in a representation that can be incorperated in
  finished documentation. 
 
  c) We have the first working example of an interactive drawing
  program that works over the ARPAnet, using Sproull's network graphics
  protocol.  This also was put together by an unpaid staff member. 
  (Gee, maybe if we stopped paying people, productivity would increase!)
 
 We would be happy to gin up a proposal in this area and would be even
 happier if we could get some funding for it.
 CC: Licklider%ISI


∂5-APR-75  0536		network site ISI
 Date:  5 APR 1975 0539-PDT
 From: LICKLIDER at USC-ISI
 Subject: Roadmap for Intelligent Systems
 To:   Newell at CMU-10A, LES at SU-AI, Winston at MIT-AI,
 To:   Nilsson at SRI-AI, Feigenbaum, Amarel
 cc:   Licklider
 
 	George Heilmeier wants me to produce a 'roadmap' for the
 IPTO Intelligent Systems program -- actually, for the
 union of Artificial Intelligence and Knowledge-Based Computer System
 Applications, since Intelligent Terminals already has a
 definite plan.  Yesterday I received word that the roadmap should be
 ready for discussion on April 19 (which turns out to be a Saturday),
 and that creates some schedule pressure.
 
 	Instead of just putting together something from the local
 IPTO point of view, I'd like to build a plan out of
 inputs from all of you, something that will represent concensus and
 constitute a step toward solution of the general problem I tried to
 set forth in my 'Easter Message'.
 
 	The main content of a roadmap should be the
 problems to be addressed and the objectives in terms of which the
 effort should be structured during the next (say) three
 years.  The pressures from the front office to which the roadmap
 should respond (either by acceptance and incorporation of
 guidance or by presenting strong counterarguments and setting forth
 a counterplan) include: creating several application projects
 that will prove the power of AI technology in Defense problem areas,
 eliminating undesirable duplication, redirecting work that is aimed
 at solving problems that are not important to DoD (an example, here,
 is MIT's work on maintenance of electronic equipment at the circuit
 card level -- GHH [Heilmeier], who knows a lot about electronic
 devices, thinks LSI and blind automation are outflanking that problem
 area, and another example is McCarthy's work on formal
 reasoning, which GHH sees as so basic in its application to math-
 ematical theorem proving and checking that DoD should not be supporting
 it), and showing what has been achieved as a result of ARPA's
 expenditure of the '50 million' since 1964.  The last pressure
 suggests that we have to go a bit beyond roadmapping and
 must include an 'accomplishments' section.
 
 	As I see it, the roadmap should include both scientific
 objectives and application objectives.  On the scientific side, it
 should try to make the chain of reasoning from scientific objective
 to technological capabiltiy to application system  -- make that chain
 definite and real.  The thing to avoid is a set of very far out
 scientific objectives that will make it appear
 that no useful product can be expected for another decade.
 
 	Could we proceed as follows -- I'll even be specific about
 a date, but you may have to make counterproposals at whatever levels
 need changing:
 
 	Each of [Newell, Ernest, Winston, Nilsson] identify an 
 additional member of your group and get him involved as a central,
 working member of this circle, which would then number 10, not
 counting me.  (That would leave Feigenbaum, whose project is the
 smallest, and Amarel, whose project does not fall in AI-KBCSA in
 IPTO bookkeeping but who is nevertheless centrally in AI, as single
 members.)  Each person or contract-group  prepare a roadmap
 for AI-KBCSA (magnifying the areas in which his group is 
 concerned, but not wholly neglecting the others -- and not neglecting
 applications, either -- indeed, moving to embrace and include
 application(s) insofar as eager, willing, able) and circulate it t
 to all the rest by Friday noon this week.  Over the week-end, each
 person study all the roadmaps.  All come to Washington on Monday,
 April 14th for a day of discussion, planning, evaluation.  On that
 Monday, jointly create the required IPTO roadmap.  Later in that
 week, I would put it into ARPA format and play it back to the
 group, and put your reactions into a final iteration.
 
 	The additionally selected people should be scientific/
 technical/working-researcher types and not administrator/manager.
 
 	Would you all, ASAP, send me messages indicating whether or
 not you can come to ARPA Monday the 14th and ALSO -- since probably
 some will not be able to -- what days between Friday the 11th and
 Wednesday the 16th you can come to Washington.  [Except the 15th,
 on which I have to make a twice postponed trip to RADC.]  Please
 let me have that information by Monday evening at the latest.
 
 	What I have in mind for a roadmap is not a full and complete
 report on the importance and health of AI.  It is just a few sheets
 of paper or viewgraphs.  But it ought to be right because it is going
 to be the basis for whatever redirections are made in the near future,
 say in the next year.  There are longer terms things to discuss,
 and we'll have time (in a full day) to discuss them, but this
 roadmap will have to be the main order of business, the sine qua non.
 
 	Please figure on arriving, ready to start, at 9:00 and
 working through the dinner hour.  Indeed, east-coasters, please plan
 on taking last planes back, and, west-coasters, on staying
 over.
 
 				Regards
 
 				Lick
 -------


∂2-APR-75  0534		network site ISI
 Date:  2 APR 1975 0534-PDT
 From: LICKLIDER at USC-ISI
 Subject: Retransmission of 'Easter Message' Earlier Sent to 'Ernest'
 To:   LES at SU-AI
 cc:   Licklider
 
 	The purpose of this Easter note is to bring you up to date
 on a development in ARPA that concerns me greatly -- and will, I think,
 also concern you.  It is the continued and accelerating (as I
 perceive it) tendency, on the part of the ARPA front office, to devalue
 basic research and the effort to build up an advanced science/technology
 base in favor of applied research and development aimed at directly
 solving on an ad hoc basis some of the pressing problems of the DoD.
 Let me be clear that I am strongly in favor of ARPA's contributing
 maximally to the solution of pressing DoD problems.  What concerns
 me is that, whereas I see the main hope in the creation of new methods
 and far-advanced systems based on new methods, the prevailing direction
 in ARPA is to do research within the specific
 contexts of military problems -- and not to do research
 that does not have a military 'buyer' ready to take it over as soon as
 the concept gets well formulated.
 
 	The present indications of this direct-
 application-oriented trend are strong pressures form the new Director,
 George Heilmeier, that IPTO 'redirect' the university AI efforts
 to work on problems (vehicles) that have real DoD validity, criticisms
 of the Speech Understanding and Image
 Understanding programs for not being tied directly into application
 projects that will be taken over and supported by the Services, strong
 emphasis on Software Technology efforts that will have effects in the
 short term (with strong pressure to de-emphasize longer-term research
 such as that on fully automatic programming), and, in general, a
 tendency to evaluate IPTO programs by asking people in the DoD offices
 (which would use or preside over the use of eventual applications of
 our stuff) what they think of the programs.
 
 	In the case of AI (or Intelligent Systems, IS, as the sub-
 element of the IPTO program is called in local paperwork -- IS now
 includes three sub-subelements: AI, Knowledge-Based Computer System
 Applications, and Intelligent Terminals), the situation is complicated
 by the fact that ARPA has been supporting basic research at a
 rather high level for more than ten years (has spent more than
 $50 million on it), and it is natural for a new director, or even an
 old one, to ask, 'What have we gotten out of it in terms of improvements
 in national defense?'.  [Supermodern punctuation convention right
 there!]  Unfortunately, most people who are asked that question answer
 'I don't know' or even 'nothing'.  IPTO tries to establish that pushdown
 lists, interactive debugging in source language, and even a
 big part of time sharing came out of AI and that AI systems such as
 Dendral, Mycin, and Macsyma actually do have expert-human-level
 capabilities or (in some areas) better, but there is in fact a big
 gulf between the perceptual sets of AI buffs and DoD administrators,
 and the latter really mean, when they ask about the payoff, where are
 the intelligent weapon systems?  They would understand intelligent
 support systems, but weapon systems dominate support systems in the
 prevailing psychological space, and even the Secretary of Defense has
 to work at it to get his people to realize that more is spent on
 operations and maintenance than on personnel and more on personnel than
 on procurement.  In any event, this perceived lack of specific
 payoff from $50 million is a major source of dissonance.
 
 	Needless to say, we have all been working hard to accomplish
 two things:
 
 	1.  To educate the new Director into a stronger appreciation
 	of truly advanced technology and into a realization that
 	the future is not to be won by making a lot of minor
 	technological advances and moving them immediately into the
 	Services.  We point to time sharing and interactive computing,
 	systems like DENDRAL and MACSYMA, languages like LISP, the
 	ARPANET, highly realistic graphics with brightness
 	gradations, color, and kinematics, and other such exemplars
 	of real advances that have already been made and have had
 	major effects.  We point to the big advances that are now in
 	the offing -- about which more later.
 
 	2.  To improve the connections between IPTO programs and the
 	DoD offices that ought to know about them, appreciate them,
 	say good things about them, and be ready to take over and
 	apply their applicable outputs.  There is no doubt that
 	IPTO is (and always has been) weak in respect of that kind of
 	connection with the Department of which it is a part.  During
 	the time he was in DDR&E, George Heilmeier evidently heard
 	a lot of criticism of ARPA, and of IPTO in particular, from
 	the various DoD offices, and now he is determined to do
 	something about it.  The fact is, we in IPTO were working
 	very hard to do something about it, but now we are under
 	strong Directoral pressure to accelerate the movement.
 
 Many of my remarks at the recent PI Conference were related
 to what I am telling you now, but I mean what I say here to be a 
 stronger statement of the problem and an assertion that 
 the problem is deeper than I then realized.
 
 	During the time since George Heilmeier arrived, we have
 taken advantage of every opportunity to brief him and discuss
 programs with him.  He has given us a lot of time and participated
 actively in the discussions.  He is bright and energetic; there is
 no problem about getting his attention.  The problem is that the
 frame of reference with which he enters the discussions is basically
 quite different from the frames of reference that are 
 natural, comfortable, and familiar to most of us in IPTO -- and, I
 think, to most of you.  In my frame -- or in our frames -- it is a
 fundamental axiom that computers and communications are crucially
 important, that getting computers to understand natural
 language and to respond to speech will have profound consequences
 for the military, that the ARPANET and satellite packet communications
 and ground and air radio networks are major steps forward into a new
 era of command and control, that AI techniques will make it possible
 to interpret satellite photographs automatically, and
 that 10↑10-bit nanosec memories and 10↑12-bit microsecond memories
 and 10↑15-bit millisecond memories are more desirable than gold.  In
 George's frame, and to a greater or lesser (I think the latter)
 extent also in Alex Tachmindji's frame, none of those things is
 axiomatic -- and the basic question is, who in DoD needs it and is
 willing to put up some money on it now?  We are trying hard to
 decrease the dissonance between the frames, but we are not making
 good progress.  As one of my colleagues put it Friday, 'I think we are
 slowly holding our own with George'.
 
 	A little over a week ago, we had our 'Apportionment Review'
 in which adjustments to FY 76 funding levels were discussed.  We
 have not heard yet exactly what the front office's conclusions were
 or will be (though we have been interactiong on the question); we
 expect to have a written statement on Tuesday, along with the other
 ARPA offices.  My expectation is that there will be major deferrals
 in all our basic research programs.  But we shall
 know more definitely very soon, so I'll not speculate further on that
 subject.  The important thing will be not so much the size of the
 deferrals as whether they are defined as hedges against Congressional
 fund cutting (Such deferrals are necessary because we do not yet know
 what the action of the Congressional committees will be.) or as
 diversions of funds from basic research (or even exploratory
 development projects that do not have definite technology-transfer
 routes established) into definite applications.
 
 	As I mentioned at the PI Conference, the concept of
 'Silver Bullets' is important in ARPA, in George Heilmeier's view
 of what ARPA should accomplish.  One of his main silver-bullet areas
 is underwater sound and sonar, and IPTO is in the process of 'buying
 in' on the HASP Project (Ed Feigenbaum's AI approach).  Another is
 maintenance of vehicles with the aid of sensors and indicators that
 predict needs for maintenance, and we are trying to establish the
 fact that computers necessarily must play a central role in
 maintenance diagnosis and prognosis.  A third silver-bullet area is
 Software Technology -- an all-out effort to solve DoD's software
 problem(s).  (This is an IPTO area, and we are working hard to formulate
 it.  George wants to get the Services and the software houses
 into the effort, as do I, since I have embraced the basic goal of this
 and see the Services and the companies that do DoD's software work
 as essentially the targets (and we need to have the targets working with
 and for us).  An issue in ST is the degree to which AP can be kept in
 the program as the main hope of achieving a really fundamental solution.
 An important staffer on an important Congressional Committee seems
 (still) to be set against AP and even against ST, and all the other 
 powers that be seem to be constitutionally against anything that won't
 get finished while they are still in their present jobs.
 
 	At present, George's list of important things IPTO can do for
 DoD is:  
 
 	Get computers to read Morse Code in the presence of other code
 and noise.
 
 	Get computers to identify/detect key words in a stream of
 speech.
 
 	Develop speech-understanding systems (if there really is a 
 clear use for them in the military).  [This is a major come-about
 during the last few days.  Earlier, he was very cool toward SUS.]
 
 	Solve DoD's 'Software Problem'.
 
 	Make a real contribution to Command and Control.  [George is
 not fully convinced about packet communications, yet, but he thinks
 we may have something in there somewhere.]
 
 	Help the Tactical Technology Office do a good thing
 in sonar.
 
 	[end of list]
 
 	Sadly, that list does not include some of the main items
 that are on mine.  It is too late on Easter evening to give my 
 full list, but here is one item from each of our seven
 programs:
 
 Intelligent Systems
 
 	Develop a system that will guide not-sufficiently-
 trained maintenance men through the maintenance of complex equipment.
 
 Advanced Memory Technology
 
 	Learn how to handle very large, distributed, redundant 
 databases.
 
 Image Understanding
 
 	Develop automatic photointerpretation.
 
 Climate Dynamics
 
 	Develop the basis in modeling and array computing for 
 evaluation of effects of major human projects/activities on climate.
 [As you know, ARPA is transferring this
 one to NSF -- but what I listed is more or less accomplished.]
 
 
 Software Technology
 
 	Take the excessive cost, delay, and error out of software
 development and maintenance.
 
 Speech Processing
 
 	Make it possible for people to communicate with computers in
 natural, continuous speech.
 
 	[Let me list a second item here.]
 
 	Achieve good-quality, natural, recognizable speech with
 500 to 3000 bits per second (through commpression) so it can be
 made secure for DoD communication --  and also master the handling of
 speech in packet communication networks.
 
 Integrated C↑3 Systems
 
 	Provide an integrated, coherent, secure, effective computer-
 communication base of Command and Control -- i.e., an ARPANET-like
 system with additional media (satellites, ground radio,
 aircraft radio), security, message services, database services, and so
 on, with the emphasis on integration/coherence.
 
 [end of list]
 
 	What are we going to do about all this?  Here in IPTO we are
 going to continue our interaction with George and get off to a strong
 start with the new Deputy Director who will be coming on board in a
 couple of weeks.  We are going to try to sell them our view of the 
 world, but we are going to be moving closer to their view, surely,
 in the process.  We will be pressing on you in ways not natural to
 my philosophy -- not to get you to do research you do not want to do
 or to make compromises you do not want to make, but to get you to see 
 the picture clearly as it is seen from here so you can make wise and
 correct decisions.  And we will be asking you, as soon as we can
 arrange it, to come in and meet and try to influence the new Director
 and Deputy Director.  Meanwhile, we are open to advice and counsel --
 indeed, need it and will appreciate it greatly.
 
 	On the positive side, let me say that a lot of the offices in 
 DoD and elswhere in the government have by now heard of IPTO and
 are impppressed with the technology the IPTO Community 
 has created.  There is some real support out there in the technical
 offices of DoD.  Also on the positive side, let me say that the IPTO
 Program Managers have been doing a marvelous job.  Their performance
 on the second day of the Apportionment Review was superb, and George
 Heilmeier realized he had really been in a session (and he acknowledged
 it).
 
 	On the negative side, the fact is that we are not making as much
 progress as I think necessary, and the timing in relation to my own
 plans is very poor.  As most of you know, I have been thinking
 in terms of going back to MIT in September, and, unless, I can
 get a strong candidate-successor before the new Director,
 I am afraid he might take the occasion of my leaving to put
 a strongly applications-oriented person into the job.
 
 	My reporting on the problem at this time is not intended to
 sound a general quarters alarm; it is to make sure that you are aware
 that a serious problem exists in ARPA-IPTO and demands profound
 consideration by all of you.  Please share this information with those
 in your organizations who should ponder the matter and will respect its
 sensitivity, and please call me to discuss any or all aspects of it.
 I'll keep you informed as the situation develops.
 
 	
 [Time Lapse.  Now it is Tuesday Morning.]
 
 	Yesterday afternoon, Dave Russell and I spent another hour
 and a half with Heilmeier and Tachmindji.  As a result, my perception
 of the situation is a bit more definite.  I'll add a few paragraphs
 to Sunday's message (which I decided to hold until after the Monday
 meeting) and send it off to you.
 
 	The 'directoral guidance' re Intelligent Systems is
 now quite specific in these respects:  ARPA does not want to continue
 to fund the field in an open-ended stream-of-research way; it
 wants to redirect most of the AI research it is funding in such a way
 as to test or measure the present capability of the field to
 serve real DoD application needs, and it would like to see the support
 of basic research in AI either taken over by an agency such as NSF
 or, at least, shared more equitably than it is now.  The transition
 will be handled insofar as possible to avoid damage to the field, but
 the transition will be made.  In some instances, contracts in the
 IS area, or largely in the IS area, will be extended on a short-term
 (e.g., 6 mo.) basis while redirection takes place.
 Dave and I are directed to discuss with other agencies the assumption,
 on their part, of a larger share of the over-all support of the
 field.  For the time being, the over-all IPTO IS budget is not to be
 reduced, but it is necessary that the same funding level now cover
 some application efforts.  Definite allocations of funds within the
 IS budget have not been directed, but it is clear that George is
 determined to bring about the shift to application and will direct
 definite allocations if he is not satisfied with IPTO's progress in
 effecting the desired change in direction.
 
 	The Intelligent Terminals sub-area of IS, which is a new
 sub-area, planned to be funded in FY 76, is not touched by the
 direction just mentioned, and its funds are not part of the basic-
 plus-applied total that is supposed to be the same in FY 76 as in FY 75.
 The IT program has been planned in such a way that its application and
 technology-transfer aspects are clear and definite, so it does not
 come under the pressure to shift from basic to applied.
 
 	In Image Understanding, the direction is less definite, but the
 pressure is strong to achieve actual applications of results in
 image coding, image enhancement, and image restoration, and the
 sense is that IPTO will have to determine and make the case for the
 plausibility of a program in image interpretation  (alias, extraction
 of information from images, alias 'image understanding' in analogy
 with 'speech understanding') before actually setting out on an
 Image Understanding program.  Again, the cost of application and
 technology-transfer work will have to come out of a total budget that
 is not greater than last year's.
 
 	The third basic-research area (the third of the three
 IPTO Computer and Communication Sciences subelements) is Advanced
 Memory Technology.  It is almost wholly new in FY 76 and so does not
 come under the same kind of redirectional pressure as IS and IU.  The
 Very Large Database Systems part of AMT is set up with strong attention
 to application and technology transfer.  The Advanced Memory Concepts
 part has been planned as a quite-far-out program and therefore will
 have philosophical problems that may translate themselves into
 funding problems, but it is too early to tell just what will happen.
 The AMC Program Planning Committee (Berlekamp Commitee) is going
 to brief Heilmeier and others on (tentatively) April 17, and the
 situation will begin to clarify itself then.
 
 	Finally (in this report of what I learned yesterday and how
 it shaped my perception of what is going on), the feedback from our
 Apportionment Review is going to slip a bit, and I will not know today
 (as I think I indicated I would) exactly what the 'reapportionment
 guidance' is.
 
 	From what I have said, you can see clearly, I think, that
 we are at a watershed in the history of ARPA-IPTO.  Although the
 redirection is not wholly, or even to any large extent, in accord with
 my own philosophy of research support, it will at least remove or
 reduce the dissonance that has long characterized
 the relation of the IPTO program to the DoD organizations that are
 supposed to use the results of IPTO-supported R&D.  At the same time,
 it will introduce dissonance into the relation between IPTO and its
 basic-research contractors.  I am deeply concerned about my own role
 in the redirection -- whether to fight it, try to contain it, or
 join it wholeheartedly and try to steer it in such a way as to
 wind up with a larger, stronger, more productive enterprise.  I
 have been about half way between the first two alternatives, but 
 neither the half-way-between point nor either of the first two is
 really a workable position within ARPA.  It has to be either leave
 and fight or stay and join -- and it is clear that to adopt the former
 course precipitously would have a very bad effect on the program.
 And it is such an important -- in many ways, absolutely crucial --
 program!
 
 	One of the next steps is to get the IS PIs to come here to talk
 with George Heilmeier and Don Looft (who is the new Deputy Director,
 just now coming on board).  I'll be contacting some of you about that
 in the near future.
 
 	Meanwhile, please let me hear from you.  I need your counsel
 and help.
 
 [Another time lapse.  Now it is Wednesday morning.]
 
 	Let me end this message with a proposal: what the response of 
 IPTO and the contractor community should be to the situation I have
 described.  I am sure the situation is real.  I am not just in a
 temporarily gloomy state.  I have smoothed quite a bit, held back on
 composing such a piece as this for some time, not wanting to be an
 alarmist.  Indeed, I would much rather talk with each of you
 personally about the situation -- face to face or on the phone, so I
 could react to your individual responses in real time.  (I realize
 that the written word is no proper medium for this kind of 
 communication, but there is not time for so many individual
 interactions, and I think I should get on with this.)  Here is
 the proposal to which I invite your reaction:
 
 	IPTO should proceed promptly but deliberatlyy to construct
 a new modus operandi that will provide a new basis for very significant
 advances in computer and communications sciences, together with their
 application to improve U.S. defense, during the next ten years.
 The new game will wholeheartedly embrace the goal of bringing about
 applications of new technology.  It will continue the present
 devotion to major scientific and technological advances (and not
 devote much time or money to merely incremental improvements), but it
 will devote a much larger fraction of its resources to moving the
 advances into use.  It will not, over a long period, be a zero-sum game.
 (For a time, until it proves itself, it will have to operate with a
 level or declining budget.)  It will take
 advantage of every demonstrated success to increase both the basic
 research and the application budgets.  But most of the growth will be in
 the area of present lack, in computer and communication engineering and
 applications, and there will be a significant shift in the center of
 gravity of the contractor community.  The shift will give the
 university research groups an engineering arm , a marketplace,
 customers, users.  Several 'industrial' contractors and several
 Service laboratories will be brought into the community, and maybe
 one or two more of the FCRCs.  (The Lincoln Laboratory is the only
 'Federally Controlled Research Center' in the IPTO contractor
 community in a major way.)  Interaction
 between the university research groups and the engineering and
 application parts of the community will be real and strong.  The
 university people will learn more about possible DoD applications,
 and there will be less basis for the belief (which exists in some
 quarters) that there is an active shunning of research vehicles
 that might appear to be related to DoD applications.
 The interaction will strengthen the basic work because there
 will be more feeback from real tests of the new ideas and because
 every star needs an audience and performs best before a big and
 enthusiastic one.  Moreover, ideas will really start to move into
 use.  The presently vast gulf between how software is created in
 some of the IPTO-sponsored laboratories and how it is created
 in the shops that produce DoD's software will narrow.  It will take
 three years instead of thirty for spaghetti stacks
 to  get from LISP to COBOL.
 
 	In short, IPTO should correct the present imbalance, should
 build up an engineering and applications wing to make the contractor
 community strong and capable of meeting the challenges that are not
 now being met well.  This will remove the dissonance that is causing
 the present trouble and will turn grumbling into appreciation.  In the
 process, it will do a very good thing for the basic science, which
 cannot really get along playing so much to itself as it now
 (allegedly and probably actually) is.
 
 	That is just a rough first cut and deals with objectives more
 than with how to achieve them, but it seems best to propose
 something for discussion rather than just to ask you to ponder
 the situation.  Please let me have your reactions soon.
 
 				Regards
 
 				Lick
 
 -------


∂01-APR-75  1142		1,OH
 MR. BLUE WOULD LIKE YOU TO FORMALLY SUBMIT THE REVISED BUDGET
 WHICH WAS CONTAINED IN YOUR MESSAGE TO LICKLIDER OF 3-28  (ARPA)
 IF YOU ARE AT ALL UNCLEAR, PLEASE CALL HIM.


∂01-APR-75  1217		1,OH
 FIELDS, 202 694-5922.  PLEASE CALL AS SOON AS YOU GET IN.  NO EMERGENCY,
 BUT HE DOES NEED TO CONTACT YOU.


∂28-MAR-75  0519		S,LES
 Subject: Revised Budget
 Here is a new budget for the AI Project, revised in accordance with
 your guidelines of yesterday afternoon.  If and when this becomes firm,
 we would like to know as soon as possible so that we can give maximum
 notice to the 25 people who must be laid off.
 
 A copy of the following text has been FTPed to your OFFICE-1 area
 in file SAIL.BUDGET
                                                                     98
 
 
            Appendix G                    Research Comp. Sci.
 
              BUDGET                     Manna, Zohar            12,480
                                          Research Comp. Sci.
 
 Six  months  beginning   1  July        Panofsky, Edward  F.     8,112
 1974.    Current   salaries  are         Comp. Sys. Engineer
 projected  4% upward  to account
 for expected salary increases.          Rubin, Jeff              7,887
                                          Systems Programmer
       Salaries (6 months)
                                         Weyrauch, Richard        8,886
 Faculty                                  Research Associate
 
   McCarthy, John         $13,254      TOTAL Res. Staff Sal.    $96,932
    Professor
    50% Acad. Yr., 100% Summer         Student Research Assistants
                                           50% time unless
   Green, Cordell           8,106          noted otherwise
    Assist. Professor
    50% Acad. Yr., 100% Summer             Arnold, Reginald      $2,059
 
   Winograd, Terry          5,017          Barstow, David         2,246
    Asst. Professor, 50%
                                           Bolles, Robert         2,246
 TOTAL Faculty Salaries   $26,377
                                           Cartwright, Robert S.  2,059
 
 Research Staff                            Elschlager, Robert     2,059
 
   Allen, John             $8,736          Farmwald, Mike; 9%       300
    Research Associate
                                           Gennery, Don           2,246
   Cohn, Avra J.            5,928
    Research Programmer                    Harvey, Brian          2,246
 
   Earnest, Lester D.      12,879          Karp, Peggy            2,246
    Research Comp. Sci., 80%
                                           Lenat, Douglas         2,246
   Gafford, Thomas          3,744
    Design Engineer, 50%                   Moravec, Han P.        2,246
 
   Glassmire, William       7,638          Steinberg, Louis       2,246
    Research Associate
                                           Suzuki, Norihisa       2,246
   Gorin, Ralph             8,886
    Systems Programmer                     Taylor, Russell        2,246
 
   Luckham, David          11,756          Thomas, Arthur J.      2,246
 Appendix G                                                          99
 
 
     Wagner, Todd J.        2,246      TOTAL SALARIES          $177,779
 
     Wilkins, David E.      2,059      STAFF BENEFITS
                                           18% to 8-31-75        10,667
 TOTAL S.R.A. Salaries    $35,488          19% 9-1-75 on         22,519
 
                                       TRAVEL (past experience)  10,000
 Others
                                       EQUIPMENT RENTAL          21,000
   Allen, Hersche          $6,240          (IBM 3330 disc, 4 drives)
    Administrator
                                       EQUIPMENT MAINTENANCE     18,000
   Rabiroff, David          1,036          (past experience)
    Courier, 25%
                                       COMMUNICATIONS            11,000
   Stuart, Elbridge         4,767          (Telephones, etc.)
    Electronic Tech.
                                       PUBLICATIONS COST         12,000
   Wood, Patricia           4,505          (Past Experience)
    Secretary
                                       OTHER OPERATING COSTS     14,314
   Zingheim, Thomas J.      2,434          (e.g. office supplies,
    Electronic Tech., 33%                  postage, freight, utilities)
 
 TOTAL Others Salaries    $18,982      INDIRECT COSTS
                                           (47% of all above)   139,721
 
                                       CAPITAL EQUIPMENT
                                           mapping multiplexor   15,000
                                           hand/eye camera        6,000
 
                                       TOTAL BUDGET            $458,000
 Appendix G                                                         100
 
 
    Note on External Computing
           Requirements
 
 Two tasks (Program Understanding
 Systems  and   Natural  Language
 Understanding) require access to
 Interlisp  services  on  outside
 machines  until  more  computing
 resources    become    available
 locally  and  Interlisp  is made
 operable  on our  machine, which
 will probably be about mid-1976.
 The   total  level   of  support
 needed is about 50 CPU hours per
 week  on  a  Tenex  machine with
 256k words of main memory.
 
 Since  we  are  operating  under
 very       tight       financial
 constraints,  the  above  budget
 assumes      that     sufficient
 resources  can be  found  on the
 ARPAnet  and  elsewhere  to meet
 these  needs  at  no  additional
 cost.   It   is  our   hope  and
 expectation  that  we  can  find
 sufficient         computational
 support,  mostly  at  night,  in
 such places as SRI-AI and ISI.
 
 If  these expectations  prove to
 be ill-founded, we will  have to
 modify the  milestone schedules,
 or  seek   additional  financial
 support, or both.
 Appendix G                                                         101
 
 
                              Cost by Task
 
 The  costs of  the tasks  described in  preceding sections  are listed
 here,  with  laboratory  support  costs  (secretaries, administrators,
 travel,  etc.)  allocated  by  population.   Computer   service  costs
 (maintenance, rental, and system programming) are similarly allocated,
 except  that  Program  Understanding  Systems  and   Natural  Language
 Understanding are assessed only a half share each.  This allocation is
 made on the basis that much of their computational support is  to come
 from external sources during the period of this proposal (see "Note on
 External Computing Requirements", just above).
 
 All figures below include University indirect costs.
 
                              BASIC      SUPPORT   COMPUTER
                              TASK        COST      COST        TOTAL
         PROJECT              COSTS       ALLOC.    ALLOC.      COST
 
 Formal Reasoning             73,255     29,791     37,683    140,729
 
 Automatic Deduction          43,779     24,826     31,403    100,008
 
 Program Understanding        39,826     29,791     18,842     88,459
 
 Natural Language             12,669      9,930      6,281     28,880
 
 Hand-eye Systems             32,450     29,791     37,683     99,924
 
           TOTALS            201,979    124,129    131,892    458,000
 CC: Licklider%ISI


∂25-MAR-75  0556		network site ISI
 Date: 25 MAR 1975 0554-PDT
 From: FIELDS at USC-ISI
 Subject: ACCOUNTS
 To:   LES at SU-AI
 cc:   STUBBS
 
 TO THE BEST OF MY KNOWLEDGE GENE STUBBS HANDLES REQUESTS
 FOR NEW ISI ACCOUNTS.
 BEST
 CRAIG
 -------

∂25-MAR-75  0623		S,LES
 Subject: Request for Computer Account
 Dear Gene:
 
 I wish to request that one of our Student Research Assistants, namely
 Louis Steinberg, be given access to one of the Tenex machines at ISI.
 He is working for Cordell Green on Program Understanding Systems,
 under ARPA sponsorship, and needs some Interlisp service in support
 of that work.
 
 Our computer system does not yet support Interlisp and probably won't
 for a year or so.  If it would help for load leveling, Steinberg
 could restrict his work to certain specified hours of the day or
 night. 
 
 I realize that those machines are fairly heavily loaded already, but
 if you can reasonably squeeze on another one, we would appreciate it.
 
 		Regards,
 		Les
 CC: Stubbs%ISI


∂24-MAR-75  1223		S,LES
 My apologies for sending a dunning courier to your office.  I had
 arranged for them to be paid at this end (I thought), but I suppose
 that they make more money if they collect twice.
 
 The formal submission is in the mill and should arrive within a few
 days.
 CC: Licklider%ISI


∂24-MAR-75  1559		S,LES
 We would like to request an ISI account for one of our people
 working on Program Understanding Systems (under Cordell Green).
 My question is, to whom should the request be addressed?
 CC: Fields%ISI


∂20-MAR-75  0434		S,LES
 We're in a bit of trouble, with an XGP that has been broken for
 several days and a malfunctioning McCarthy (nothing serious).  As
 things stand, we hope to get to hard copy by Thursday night and ship
 you a copy by courier. 
 
 Most of the proposal is available for perusal: for example, to see
 Cordell's part, get to our system and say
  TYPE PU[R,LES]
 where "PU" stands for program understanding (really).  Similarly, to
 see Luckham's part, use AD (for Automatic Deduction) in place of PU. 
 Winograd's part is in NL and Binford's in HE.  The formal reasoning
 part should appear there shortly as "FR".  Part of the Introduction
 is in "FRONT". 
 
 To type out particular pages, use a parenthetical expression.  For
 example, if you say "TYPE PU[R,LES](3)" then it types just page 3.
 More generally, you can use expressions of the form "(4:6,7,8:14)"
 and it will do all that.
 
 To halt the typout at any time, type <CONTROL>C twice.
 
 I am embarassed that this is dragging on so.  It is not as if we were
 prolonging the activity because we enjoy it so much.
 CC: Licklider%ISI


∂31-JAN-75  1354		network site ISIA
 Date: 31 JAN 1975 1334-PST
 From: LICKLIDER at USC-ISIA
 Subject: Hieronymus
 To:   McCarthy at SU-AI
 cc:   Licklider
 
 	John, I do not know him well enough to be helpful.
 He seemed bright and interesting when he was here for a short visit,
 but I did not get enough of a fix on him to be sure whether I'd
 want to hire him, myself, or urge you to keep him
 hired.
 
 	My consideration of your proposal was interrupted by
 paperwork requirements levied by the new Director and by
 the need to get out several Memos Requesting ARPA Orders before
 COB today.  I'll get back at it shortly.  I'll have to ask for
 considerable sharpening of the objective structure; it now reads
 as though there are a lot of miscellaneous objectives to be reached
 in a year or two, but that it is not clear what roads they are on,
 what major goals they are subordinate to.  Also, I'll have to ask for
 a major reconsideration of the equipment section, which sounds as though
 it is heading for another ideosyncratic system.  (I have an idea
 about how to handle that, but I can't describe it now.)  Sorry to 
 have to noodle about on the proposal.  Al Blue got us an extension
 to cover the time required for noodling without losing the money.
 But we'll have to work pretty fast.  I assume I should communicate
 with LES -- with carbons to you?
 
 				Regards
 
 				Lick
 -------
 


∂23-JAN-75  2355		S,LES
 Craig Fields @ARPA noticed that the proposal promised "initial
 specifications for a knowledge representation of language" in
 January 1975.  Guess what?  He wants a copy.
 
 If you actually have something like that, I can ship it with
 some other stuff I'm sending.  If not, I can put him off.
 CC: tw


∂24-JAN-75  0952		1,TW
 The file KRL[1,tw] is a current collection of stuff which I am having
 the grad students read.  I would not be happy about it going out as
 a product, since it is at least as unfinished as the various warning
 notes within it insist.  If the goal is to show him that work is indeed
 going on, then it might be waved in front of him.  If you want to
 show that milestones are being carefully met, put him off for a few weeks
 and there will be a real version of it..
  --terry


∂23-JAN-75  1342		ESS,JMC
 ∂23-JAN-75  1251		network site BBN
  Date: 23 JAN 1975 1552-EST
  From: FIELDS at BBN-TENEX
  Subject: STANFORD PROPOSAL
  To:   JMC at SU-AI
  cc:   LICKLIDER
  
  JOHN,
  	I READ YOUR PROPOSAL. I AM VERY INTERESTED IN 
  TWO ITEMS. WOULD IT BE POSSIBLE TO GET A COPY OF
  LOW'S THESIS (AIM 242) BY US MAIL? IT WOULD BE TOO
  LONG TO PRINT ON MY TERMINAL. ALSO, ON PAGE 20
  YOU INDICATE THAT IN JANUARY, 9175 WINOGRAD SHOULD
  HAVE A COPY OF INITIAL SPECIFICATIONS FOR A KNOWLEDGE
  REPRESENTATION LANGUAGE. COULD I SEE A COPY OF THAT
  AS WELL? THANKS FOR BOTHIN  ADVANCE.
  
  	A WORD OF FRIENDLY ADVICE.  ON PAGE 41 YOU
  INDICATE THE NEED FOR A SUBSTANTIAL AMOUNT OF TENEX
  TIME AND ARPANET ACCESS. BOTH COST REAL DOLLARS
  AND I SUGGEST THAT IN YOUR NEGOCIATIONS WITH
  LICK THESE ITEMS ARE NOT OVERLOOKED. THE TENEX
  TIME WILL POSE THE GREATEST PROBLEM. ALTHOUGH
  NEW TENEXES ARE EXPECTED, THE DEMAND WILL EXCEED
  THE SUPPLY, AND LICK WILL HAVE TO MAKE A DECISION ON
  ALLOCATION SO AS NOT TO GIVE AWAY MORE THAN 100%
  OF TENEX AVAILABLE.  THIS WAS JUST TO HIGHLIGHT
  A POTENTIAL PROBLEM TO ASSIST YOU.
  
  BEST
  CRAIG
  
  P.S. HOW ARE THINGS IN THE 3850 CLUB?
  -------
 

∂6-JAN-75  1952		network site ISIA
 Date:  6 JAN 1975 1952-PST
 From: LICKLIDER at USC-ISI
 Subject: Proposal Received
 To:   LES at SU-AI, McCarthy at SU-AI, Feigenbaum at SU-AI
 cc:   Blue, Russell, Licklider
 
 	This evening I pulled in a copy of your draft proposal,
 ARPA[R,LES].  It looks complete.  Since tomorrow is Midyear
 Review Day, I won't get at it tonight, but I am very happy to have it
 and shall study it just as soon as possible.  Please send any
 additional parts that become available in draft as they become avail-
 able.
 
 				Regards
 
 				Lick
 -------


∂13-JAN-75  1357		network site OFF
 Date: 13 JAN 1975 1357-PDT
 From: LICKLIDER at OFFICE-1
 Subject: Formal Submission of Proposal
 To:   LES at SU-AI
 cc:   Licklider
 
 	By all means plunge on with the formal submission asap, but
 be sure to associate costs with individual items.  It is
 not certain that we shall be able to fund the work at the proposed
 level, and it would be convenient if the costs could be grouped
 in such a way that we could negotiate them downward without causing
 a retyping of the whole proposal.
 
 	I am off now for a four-day trip.  When I get back, I'll get
 together with the PMs concerned, come to a decision about funding
 level, and communicate with you right away.  If there is anything in the
 final version of the proposal (final except for funding level) that is
 not in the draft we have, please send it asap to Dave Russell 
 (Russell at ISI), not to me.  (He'll be here~.)
 
 				Regards
 
 				Lick
 -------

∂13-JAN-75  1044		S,LES
 Budget Modifications to Stanford A.I. Proposal
 
 We have cut a few people out of the A.I. Lab. budget, (people known
 to be leaving soon, with no replacement planned), so the two-year
 total for everything is now $2,899,320.
 
 There is also a need for some Interlisp time on outside machines for
 Winograd's and Green's groups, as we have discussed.  Both the amount
 of time needed and what it would cost if we had to pay hard money
 remain a bit squishy.  My current estimate is that each project will
 need about 25 CPU hours/week on a Tenex machine with 256k of core.
 If we are unable to scrounge that much time, the cost of buying it
 for a year or so (after which our system should support Interlisp)
 would be something like $130k for each project. 
 
 Allocating costs to projects, we get the following breakdown:
 				 Budget		Interlisp
   Formal reasoning		$826,525
   Theorem proving		 457,970
   Program understanding systems	 305,965	$130,000
   Natural language understanding 355,795	 130,000
   Hand-eye systems		 953,065
 				 -------	 -------
 			      $2,899,320	$260,000
 
 This information will be incorporated in the formal proposal submission,
 as you requested.  I hope that this covers all the data you need for
 preliminary review.  We will press on with the formal submission
 whenever you say the word.
 
 Best Regards,
 Les
 CC: licklider%ISIA;licklider%OFFICE-1


∂10-JAN-75  0553		network site ISIA
 Date: 10 JAN 1975 0552-PST
 From: STUBBS at USC-ISI
 To:   LES at SU-AI
 cc:   STUBBS
 
 
 
 LES-
 
 
 FANTASTIC!!! THANX A "HEAP".
 
 ITS ONE THING TO GET UP AT 6 BUTQUITE ANOTHER TO STAY ALL NIGHT! I'M NOT
 SURE I 'LL PAY THE OVERTIME FOR AN ALL NITE SESSION!!
 
 REGARDS
 GENE
 -------

∂10-JAN-75  0217		S,LES
 Memory Justification
 First, I must confess that I'm not capable of getting up by 6 AM,
 except to go skiing.  I was here from the night before.
 
 Here is a stab at the requested introduction.
 ----------------------------------------------------------------
 		Background and Technical Need
 
 The PDP-10 computer facility provides primary support to ARPA-IPT
 research projects in computer vision, industrial automation,
 automatic programming, natural language understanding, and formal
 reasoning, among others.  These projects place an extremely heavy
 load on both the main memory and central processor subsystems.
 For example:
 
   1) Research in computer vision and automatic assembly employs
 single programs of 300k words and more, but maximum available user
 memory is only about 150k words.  The system complexity introduced by
 the need to circumvent this limitation imposes a severe restraint on
 research progress. 
 
   2) Similarly, certain experimental LISP programs for the logic of
 computable functions do not fit in the available memory.  This places
 a bound on what can be accomplished in certain important directions.
 
   3) Research in formal reasoning involves first order logic programs
 exceeding 80,000 words each and there are often as many as five
 people needing to run these programs at once.  With the available
 memory, running just two of these jobs severly degrades system
 efficiency and performance.
 
 Overall, a number of projects that are important to established
 research objectives cannot be run at all during the daytime, and
 run much too slowly at night.  The research staff has attempted
 to adapt to this situation by spreading their work over night and
 weekend periods and by utilizing spare capacity at other installations
 on the ARPA Network, mostly at night.  There is rarely a time when
 there are less than a dozen persons utilizing the Stanford facility
 -- even at 3 AM on weekends.
 
 On top of existing needs, certain of the research projects (notably
 automatic programming and natural language understanding) are just
 beginning to mature technically, so their computational requirements
 may be expected to increase greatly in the near future.
 
 Clearly, the addition of substantial main memory to the computer
 facility is an essential step toward meeting the computational
 requirements of this research program. 
 
 ----------------------------------------------------------------
 
 Gene:  I hope that isn't too far off target.  Give me a blast if it is.
 
 Cheers,		Les
 CC: stubbs%ISIA


∂9-JAN-75  0746		network site ISIA
 Date:  9 JAN 1975 0747-PST
 From: STUBBS at USC-ISI
 Subject: MEMORY PROPOSAL
 To:   LES at SU-AI
 cc:   STUBBS
 
 HI LES - YEP,SAME OLD GENE STUBBS! JUST HERE IN ARPA AFTER SIX YEARS AT
 DSS-W. (NOT ONE REGRET ABOUT HAVING LEFT DSS-W EITHER!!!)
 
 SORRY THATTHE SYSTEM LET YOU DOWN SO EARLY IN THE MORNING. YOURE RIGHT
 THAT TO HAVE GOTTEN SUCH AN EARLY RESPONSE WOULD HAVE BEEN SURPRISING!
 EARLY IN THE 'AM' I MEAN. DON'T TELL ME THAT YOU "UNIV. TYPES" NORMALLY
 GET TO YOUR DESK AT 0600 !!!WE SURE DON'T!!!
 
 WELL - BACK TO BUSINESS. 
 WHAT I NEED FOR THE "BACK. & TECH.NEED" PARA. IS A BRIEF COMMENTARY
 ON THE HISTORY ETC TALKING ABOUT THE PEDIGREE OF THE SYSTEM,ITS
 APPLICATION TO OTHER PROGRAMS,LIGHT NOTE ON ACCOMPLISHMENTS COUCHED
 SOMEWHAT TOWARD WHAT WOULD HAVE TO HAPPEN IF YOUR SYSTEM WEREN'T
 AVAILABLE ETC. I'M SORRY I DON'T HAVE AN EXAMPLE ON THE SYSTEM
 BUT I'LL TRY TO RELAY THE GIST OF ONE WE JUST USED FOR AN ADDIT.
 MACHINE BUY AT USC:
 
 "THE PDP-10/TENEX COMPUTING RESOURCE CURRENTLY FORMS THE BACKGROUND
 FOR THE IPTO RESEARCH PROGRAM. PICTURE PROCESSING,SPEECH UNDERSTANDING,
 ARTIFICIAL INTELLIGENCE,AUTOMATIC PROGRAMMING,SOFTWARE PRODUCTION 
 TECHNOLOGY,ETC,ETC - TO SITE A FEW EXAMPLES - DEPEND ALMOST EXCLUSIVELY
 ON PDP-10 COMPUTING POWER. THERE ARE NUMEROUS RASONS FOR THIS - 
 (1) - - - - - - - - - - -
 (2) - - - - - - - -- - - 
 (3) - - - - - - - - - --
 
 IN THE IMMEDIATE FUTURE, IPTO NEEDS FOR COMPUTING POWER WILL
 DRAMATICALLY INCREASE.NEW PROGRAMS ARE BEGINNING OR OLDER PROGRAMS
 ARE INCREASING, AND THESE WILL NEED ADDITIONAL RESOURCES. "
 
 THAT KIND OF A COMMENT LES, APPROPRIATELY ALTERED TO FIT YOUR SITUATION,
 IS THE KIND OF THING I HAD IN MIND. ACTUALLY, THE PAPER I TOOK
 THE ABOVE FROM IS MUCH LONGER THAN WHAT I'VE SAID HERE AND THIS LEAVES
 SOME HOLES OF COURSE, BUT MAYBE THIS GIVES YOU THE IDEA.
 
 SURE DO APPRECIATE YOUR HELP - SEE IF YOU CAN'T GET SOME OVERTIME FOR
 THE 0600 ARRIVAL TIME !!! I CERTAINLY WOULDN'T OBJECT (MUCH).
 
 THANX AGAIN
 REGARDS
 GENE
 -------


∂08-JAN-75  1721		S,LES
 Subject: Computer Memory Proposal
 Hi Gene.  I believe that we've chatted a few times in the past,
 if you are the Gene Stubbs who was at DSS-W.
 
 Incidentally, I happened to be at my desk when your messagee came in
 at 6am (our time) and thought I'd surprise you with an instant
 response.  Unfortunately, part way through the message our computer
 hiccuped and died -- a victim of a power surge caused by a storm
 moving through -- so I had to wait awhile. 
 
 On the "Background and Technical Need" section, are you seeking
 something more concise than what we had under "Problem", or
 something more quantitative (e.g. performance figures), or
 something else?   I am not too familiar with the rules of that
 game.  One way to get the idea would be an example, if you have
 one on your computer somewhere.
 CC: stubbs%ISIA


∂8-JAN-75  0653		network site ISIA
 Date:  8 JAN 1975 0553-PST
 From: STUBBS at USC-ISI
 Subject: EQUIP.PROP.FOR 73C0435
 To:   LES at SU-AI
 cc:   STUBBS
 
 HI LES- GENE STUBBS,ARPA IPTO HERE.
 I'M A NEW BOY ON THE BLOCK IN IPTO FUNCTIONING AS  RESOURCE MANAGER.
 AL BLUE SUGGESTED I GET IN TOUCH WITH YOU FOR SOME HELP IN PREPARING
 THE ARPA ORDER REQUEST FOR YOUR PROPOSAL OF 8 DEC FOR THE ADDITIONAL
 MEMORYON YOUR PDP-10 TIMESHARING SYSTEM.
 WHAT I NEED IS A BRIEF INTRODUCTORY PARAGRAPH OR TWO TO COMPLETE
 A SECTION WE CALL "BACKGROUND AND TECHNICAL NEED".
 UNDER THE HEADING OF "PROBLEM" IN YOUR PROPOSAL THERE ARE SOME GOOD
 WORDS BUT I NEED SOMETHING A LITTLE MORE DEFINITIVE AND SURE WOULD 
 APPRECIATE YOUR HELP. I'M A "BEAN COUNTER" TYPE - FORMER PROCUREMENT
 OFFICER AND ACCOUNTANT SO AM NOT UP ON REQUIRED TECHNICAL LANGUAGE.
 I'M ON THE NET (OBVIOUSLY) AT ISI AS <STUBBS AT ISI>.
 THANX MUCH - WILL BE WAITING TO HEAR FROM YOU.
 REGARDS
 GENE
 -------


∂06-JAN-75  0755		S,LES
 I find that I cannot put our proposal into your area without knowing
 a password and, considering its size, I doubt that you would want to
 receive it as mail.  Consequently, I suggest that you try to suck it
 into your area.  The name of the file here is ARPA[R,LES].  Our
 system demands no passwords or account numbers for file transfers
 of this kind.
 CC: Licklider%ISIA;Licklider%OFFICE-1


∂6-JAN-75  1017		network site ISIA
 Date:  6 JAN 1975 1017-PST
 From: LICKLIDER at USC-ISI
 Subject: Additional PDP-10 Support
 To:   LES at SU-AI
 cc:   Fields, Licklider
 
 	
 	Fields gave me a copy of your message to him on 20 December 
 1974, and I think he wanted me to reply to you.  The reply has to be 
 that we shall do our best to help but are facing a situation of 
 serious overload during prime time.  Outside of prime time, there 
 appears to be plenty of accessible PDP-10 resource on the ARPANET>  I
 shall ask Gene Stubbs, as soon as he returns from leave, to make a map
 showing which ussers use which PDP-10's and what the average loads
 are as a function of time-of-day.   Then, if at all possible, we shall
 make some suggestions about where Cordell and Terry can get some
 computer time.
 	
 	For the middle term, three months to two years hence, we are
 purchasing additional PDP-10's (KA-10's).  For the longer term, we
 are going to have a serious study of how computer support should be pro-
 vided.
 	
 	It would be helpful to have your estimate of needed computer 
 support extended to indicatte how much processing and memory and file
 storage are needed.  Console time is not enough for planning purposes.
 	
 	Finally, let me remind you that computer support is a costly
 resource that has to be budgeted as part of the over-all proposing and
 contracting process.  ARPA cannot let research contracts and give a 
 promise, express or implied, to provide all the computer resource
 required.  So please be sure to include all needed computer support in 
 your next proposal to ARPA.  Perhaps I should take this opportunity
 to remind you that, in the proposal, you should provide a cost
 analysis that relates costs to objectives.  That is, each subproject
 should have one or more clearly defined objectives, and the projected
 cost should be associated with each such objective.  For that part of 
 the made up of computer support, it may be convenient to express it as
 a fraction of the cost of operating the laboratory computer facility
 (if it is the laboratory computer that is used), in which
 case the laboratory computer facility can be viewed as a cost
 center.  If the computer resource required is external, then there
 are three classes;  paid for at the suppliers rates, arranged
 through ARPA, and arranged for on an informal barter basis with
 another laboratory.  In your proposal, please specify the first,
 estimate the second on the basis of guidance soon to be provided, and
 specify the third.
 	
 	At the beginning of this message, I did not expect to get
 into it that deeply.  I shall, however, try to provide you some guidance
 on how to estimate the equivalent cost of TENEX service at an ARPA-
 supported service host.
 	
 				Regards,
 	
 				Lick
 JCRL/hcb
 -------

∂03-JAN-75  1556		S,LES
 Lick and Al Blue called this afternoon to say that they must have the
 proposal very soon.  They also laid on an additional requirement:
 that costs should be allocated to each project, including shares of
 local overhead items such as computer time.
 
 I said that I thought we could have a draft together, not including
 the cost allocations, by Sunday night, though it will clearly take
 some hustling.  Lick asked that we ship him a copy over the net.
 CC: jmc