5P41 RROO785-11 Annual Report
I. Narrative Description

This third year of the 5-year renewal of the SUMEX resource grant has been an
active year not only for the SUMEX staff, but for the SUMEX-AIM community involved
in developing expert systems. Successes in developing such systems, many of them
stemming from projects in the SUMEX-AIM community, continue to stimulate strong and
growing interest in AI research on many educational, governmental, and industrial fronts.

This is an annual report for the Stanford University Medical EXperimental
computer resource for applications of Artificial Intelligence in Medicine (SUMEX-AIM).
It covers the period between May i, 1983 and April 30, 1984.

This third year of the 5-year renewal of the SUMEX resource grant has been an
active year not only for the SUMEX staff, but for the SUMEX-AIM community involved
in developing expert systems. Successes in developing such systems, many of them
stemming from projects in the SUMEX-AIM community, continue to stimulate strong and
growing interest in AI research on many educational, governmental, and industrial fronts.

In addition, this past year has seen concurrent development of SUMEX-AIM as a
distributed scientific resource. Our approved project goals focus principally on the
merging of state-of-the-art community research in biomedical AI applications with new
computing tools and on the challenges they will bring to the SUMEX-AIM community and
resource. The SUMEX staff continues to exploit these advances in professional
workstations and communication technology, while at the same time maintaining our high
standards for a computing resource.

This third year also saw the initiation of a number of SUMEX-AIM pilot projects.
These pilot projects provide new activities and research directions for the community to
replace existing projects which have matured and moved off the SUMEX-AIM resource.

The earlier phases of the SUMEX-AIM resource were characterized by the building
of a national community of biomedical AI collaborators around a central resource located
at Stanford University. Beginning with 5 projects in 1973, the AIM community grew to
11 major projects at our renewal in 1978. This past year saw the completion of two long
term and successful projects on SUMEX-AIM; DENDRAL and PUFF/VM. There
currently are 13 fully-authorized projects plus seven pilot efforts.

Many of the computer programs under development by these groups are maturing
into tools increasingly useful to the respective research or clinical communities. We
continue to seek out new AI applications in our community of biomedical and computer
scientists who interact through electronic media. The SUMEX-AIM community is
beginning to evolve as a highly distributed resource, with the SUMEX staff and computer
facility serving as the backbone to electronic communication and systems support. The
community is becoming more and more involved in personal computers and professional
workstations, and more heavily dependent on network communication facilities for
interactions, collaborations, and sharing.

The following sections cover the activities of the SUMEX-AIM resource this past
year, including brief summaries of the our objectives, a characterization of biomedical Al
research, resource organization and operating procedures, recent core progress in system
development and basic AI research, and progress in the collaborative projects.

1 E. A. Feigenbaum
Summary of Research Progress 5P41 RRGO785-11

I.A. Summary of Research Progress

1.A.1. Overview of Objectives and Rationale

SUMEX-AIM ("SUMEX") is a national computer resource with a dual mission: 1)
promoting applications of computer science research in artificial intelligence (AI) to
biological and medical problems, and 2) demonstrating computer resource sharing within a
national community of health research projects. The central SUMEX-AIM facility is
located physically in the Stanford University Medical School and serves as a nucleus for a
community of medical AI projects at universities around the country. SUMENX provides
computing facilities tuned to the needs of AI research and communication tools to
facilitate remote access, inter- and intra-group contacts, and the demonstration of
developing computer programs to biomedical research collaborators.

I.A.1.1. What is Artificial Intelligence

The subfield of computer science known as Artificial Intelligence, or AI, deals with
symbolic reasoning using large amounts of heuristic knowledge. Many of the world’s
difficult problems are symbolic, such as troubleshooting electronic or mechanical
equipment, medical diagnosis and therapy planning, and configuring elemental parts into a
whole system. For these kinds of problems, AI offers new opportunities for developing
computer-based solutions.

In addition to using symbolic representations of knowledge, AI also uses heuristic
methods for processing information. Heuristics are rules of thumb, judgmental rules that
aid in finding plausible solutions. AI is distinguished from other areas of computing in its
attention to both symbolic (non-numeric) information and heuristic (non-algorithmic)
methods for solving problems.

Placing Al in Computer Sctence

The major focus of AI is understanding intelligence through construction (or
programming) of machines that behave intelligently. That is a grand goal. In the short-
term, AI research focuses on non-numerical problem solving in order to build experience
with problem solving methods, techniques for representing various kinds of knowledge,
interfaces with users, and numerous other issues.

One of the distinguishing features of problems for which AI methods have been
developed is that the problems are not well-structured. That is, one does not already
know in advance (from the problem description alone} what the best method is for solving
the problem. In short, there are no algorithms. Broadly speaking, AI substitutes
exploratory search for precise, algorithmic solution methods.

Expert Systems and Applications

The national SUMEX-AIM resource is an outgrowth of a long, interdisciplinary line
of artificial intelligence research at Stanford and elsewhere concerned with the
development of concepts and techniques for building “expert systems" [1]. An “expert
system" is an intelligent computer program that uses knowledge and inference procedures
to solve problems that are difficult enough to require significant human expertise for their
solution. For some fields of work, the knowledge necessary to perform at such a level,

bo

E. A. Feigenbaum
5P41 RROO785-11 Overview of Objectives and Rationale

plus the inference procedures used, can be thought of as a model of the expertise of the
expert practitioners of that field.

Two important features that distinguish expert systems from conventional
programs are flexibility and understandability. Expert systems are flezible in the sense
that they can be changed and extended easily, and they are understandable in the sense
that they can explain the contents of their own knowledge bases and their own lines of
reasoning [10]. These features are especially important in medicine, where knowledge is
changing rapidly and where practitioners have to understand the reasons for a program’s
decisions because they have to accept responsibility for following (or not following) those
decisions.

The application areas range from medicine to electronics, from machinery to
software. The problems range from diagnosis and troubleshooting (analysis) problems to
planning and configuration (synthesis) problems. Knowledge bases for expert systems are
built iteratively -- usually through long interactions over many months between a human
Specialist who understands the details of the domain and a knowledge engineer who
understands the programming details of the system.

The knowledge of an expert system consists of facts and heuristics. The "facts"
constitute a body of information that is widely shared, publicly available, and generally
agreed upon by experts in a field. The “heuristics" are the mostly-private, little-discussed
rules of good judgment (rules of plausible reasoning, rules of good guessing) that
characterize expert-level decision making in the field. The performance level of an expert
system is primarily a function of the size and quality of the knowledge base that it
possesses. One of the key ideas in maintaining flexibility and understandability is the
clean separation of elements of the knowledge base from elements of the program that
interpret the knowledge base.

The major issues in building expert systems, at the moment, are:

e selecting an appropriate problem (in terms of size, difficulty, importance,
decomposability, risk) ,

@ selecting a representation and control structure (or framework system that

supplies both),

settling on an appropriate vocabulary and conceptualization for the problem,

finding an available expert,

transferring the expert’s knowledge into the program (knowledge engineering),

refining the knowledge base with feedback from test cases,

packaging the system in a form that is acceptable to end-users,

validating the quality of the program’s advice.

One of the best known expert systems is MYCIN [3], a program in which the
separation of knowledge (of medicine) from the rest of the program was carefully
engineered. (The abstracted case of an arbitrary knowledge base and a framework
interpreter, plus auxiliary programs, was achieved in the EMYCIN system [16], to which
knowledge of other domains can be added to build a diagnostic system in those domains.)

Currently authorized projects in the SUMEX community are concerned in some

way with the application of AI to biomedical research’. The tangible objective of this
approach is the development of computer programs that will be more general and effective

2
Brief abstracts of the various projects can be found in Appendix B on page 209 and more detailed progress
summaries in Section ! on page 69.

3 E. A. Feigenbaum
Overview of Objectives and Rationale 5P41 RROO785-11

consultative tools for the clinician and medical scientist. There already have been
promising results in areas such as chemical structure elucidation and synthesis, diagnostic
consultation, molecular biology, and modeling of psychological processes.

Needless to say, much is yet to be learned in the process of fashioning a coherent
scientific discipline out of the assemblage of personal intuitions, mathematical procedures,
and emerging theoretical structure comprising artificial intelligence research. State-of-the-
art programs are far more narrowly-specialized and inflexible than the corresponding
aspects of human intelligence they emulate; however, in special domains they may be of
comparable or greater power, e.g., in the solution of formal problems in organic chemistry.

I.A.1.2. Impact of AI in Biomedicine

There is a certain inevitability to the field of Artificial Intelligence and its
applications, in particular, to medicine and biosciences. The cost of computers will
continue to fall drastically during the coming two decades. As it does, many more of the
practitioners of the world’s professions will be persuaded to turn to economical automatic
information processing for assistance in managing the increasing complexity of their daily
tasks. They will find, from most of computer science, help only for those problems that
have a mathematical or statistical core, or are of a routine data-processing nature. But
such problems will be relatively rare, except in engineering and physical science. In
medicine, biology, management, indeed in most of the world’s work, the daily tasks are
those requiring symbolic reasoning with detailed professional knowledge. The computers
that will act as intelligent assistants for these professionals must be endowed with
symbolic reasoning capabilities and knowledge.

The growth in medical knowledge has far surpassed the ability of a single
practitioner to master it all, and the computer’s superior information processing capacity
thereby offers a natural appeal. Furthermore, the reasoning processes of medical experts
are poorly understood; attempts to model expert decision-making necessarily require a
degree of introspection and a structured experimentation that may, in turn, improve the
quality of the physician’s own clinical decisions, making them more reproducible and
defensible. New insights that result may also allow us more adequately to teach medical
students and house staff the techniques for reaching good decisions, rather than merely to
offer a collection of facts which they must independently learn to utilize coherently.

The knowledge that must be used is a combination of factual knowledge and
heuristic knowledge. The latter is especially hard to obtain and represent since the
experts providing it are mostly unaware of the heuristic knowledge they are using.
Medical and scientific communities currently face many widely-recognized problems
relating to the rapid accumulation of knowledge, for example:

e codifying theoretical and heuristic knowledge

e effectively using the wealth of information implicitly available from textbooks,
journal articles and other practitioners

e disseminating that knowledge beyond the intellectual centers where it is
collected

® customizing the presentation of that knowledge to individual practitioners as
well as customizing the application of the information to individual cases

We believe that computers are an inevitable technology for helping to overcome

E. A. Feigenbaum 4
5P41 RROO785-11 Overview of Objectives and Rationale

these problems. While recognizing the value of mathematical modeling, statistical
classification, decision theory and other techniques, we believe that effective use of such
methods depends on using them in conjunction with less formal knowledge, including
contextual and strategic knowledge.

Artificial intelligence offers advantages for representing and using information that
will allow physicians and scientists to use computers as intelligent assistants. In this way
we envision a significant extension to the decision-making powers of specific practitioners
without reducing the importance of those individuals in that process.

Knowledge is power, in the profession and in the intelligent agent. As we proceed
to model expertise in medicine and its related sciences, we find that the power of our
programs derives mainly from the knowledge that we are able to obtain from our
collaborating practitioners, not from the sophistication of the inference processes we
observe them using. Crucially, the knowledge that gives power is not merely the
knowledge of the textbook, the lecture and the journal, but the knowledge of good
practice--the experiential knowledge of good judgment and good guessing, the knowledge
of the practitioner’s art that is often used in lieu of facts and rigor. This heuristic
knowledge is mostly private, even in the very public practice of science. It is almost never
taught explicitly, is almost never discussed and critiqued among peers, and most often is
not even in the moment-by-moment awareness of the practitioner.

Perhaps the the most expansive view of the significance of the work of the SUMEX-
AIM community is that a methodology is emerging for the systematic explication, testing,
dissemination, and teaching of the heuristic knowledge of medical practice and scientific
performance. It may be less important that computer programs can be organized to use
this knowledge than that the knowledge itself can be organized for the use of the human
practitioners of today and tomorrow.

Evidence of the impact of SUMEX-AIM in promoting ideas such as these, and
developing the pertinent specific techniques, has been the explosion of interest in medical
artificial intelligence and the specific research efforts of the SUMEX community. As
SUMEX has entered its second decade, we have found that the small community of
researchers that characterized the AIM field in the early 1970’s has now grown to a large,
accomplished, and respected research community. The American Association for Artificial
Intelligence (AAAI), the principal scientific membership organization for the AI field, has
4000 members, over 1000 of whom are members of the medical special interest group
known as the AAAI-M. This subgroup was founded by members of the SUMEX-AIM
community who were active in AAAI and is the only active subgroup in the Association.
The organization distributes semiannual newsletters on medical AI and provides a focus
for co-sponsoring relevant medical computing meetings with other societies (such as the
American Association for Medical Systems and Informatics -- AAMSI). Medical AI papers
are prominently featured at both medical computing and artificial intelligence meetings,
and artificial intelligence is now routinely featured as a specific subtopic for specialized
sessions at medical computing and other medical professional meetings. For example,
members of the AIM community have represented the field to physicians at the American
College of Pathology and American College of Physicians meetings for the last several
years. A mere decade ago, the words "artificial intelligence” were never uttered at such
conferences. The growing interest and recognition are largely due to the activities of the
SUMEX-AIM community.

Another indication of the growing impact of the SUMEX-AIM community is its
effect on medical education. For reasons such as those outlined above, there is an
increasing recognition of the need for a revolution in the way medicine is taught and

5 E. A. Feigenbaum
Overview of Objectives and Rationale 5P41 RROO785-11

medical students organize and access information. Computing technology is routinely
cited as part of this revolution, and artificial intelligence (and SUMEX-AIM research)
generally figures prominently in such discussions. Such diverse organizations as the
National Library of Medicine, the American College of Physicians, the Association of
American Medical Colleges, and the Medical Library Association have all called for
sweeping changes in medical education, increased educational use of computing
technology, enhanced research in medical computer science, and career development for
people working at the interface between medicine and computing; reports of all four
organizations have specifically cited the role of artificial intelligence techniques in future
medical practice and have used SUMEX-AIM programs as examples of where the
technology is gradually heading.

In summary, the logic which mandates that artificial intelligence play a key role in
enhancing knowledge management and access for biomedicine -- a logic in which we have
long believed -- has gradually become evident to much of the biomedical community. We
are encouraged by this increased recognition, but realistic about the significant research
challenges that remain. Our goals are accordingly both scientific and educational. We
continue to pursue the research objectives that have always guided SUMEX-AIM, but
must also undertake educational efforts designed to inform the biomedical community of
our results while cautioning it about the challenges remaining.

E. A. Feigenbaum 6
5P41 RROO785-11 Details of Technical Progress

1.A.2. Details of Technical Progress

I.A.2.1. Facility Management and Operation

The following material covers the SUMEX-AIM resource activities over the past
year in greater detail. Individual sections cover progress in ;

e Facility Management and Operation
e Timesharing Systems
e Professional Workstations

e Networking and Communications

These sections outline accomplishments in the context of the resource staff and resource
management. Details of the progress and plans for our external collaborative projects are
presented in Section II beginning on page 69.

I.A.2.2. Facility Management and Operation

SUMEX-AIM continues to manage and operate it’s computing resources in a
effective and efficient manner conducive to providing a reliable and robust computing
environment.

While the previous year (Year 10) involved a major move from the KI10 Tenex
system to a new DECsystem 2060, this year saw more emphasis to our gradual move to
distributed processing, while continuing to improve our excellent timesharing environment
on the 2060. This development is covered in full in section I1.A.2.2 starting on page 15.

Our continued movement to professional workstations has taken on several forms.
We have continued to acquire Lisp machines for use by the SUMEX community while at
the same time investigating the use of remote virtual graphics and new lower cost
workstations such as the Apple Macintosh, Sun workstations, and others that are
appearing on the market. The development of professional workstations is covered in
more detail in section I.A.2.3 starting on page 21.

SUMEX continues to expend a great deal of effort in the support and development
of our networking and communications facilities. Key to our ability to provide the
maximum computing power available to the greatest number of users is a mechanism for
making it irrelevant where that user is physically located. By having a robust networking
and communications environment, we are able to extend our facility to any user or group
of users, thereby making available to them the power and convenience of SUMEX.
Further information on the progress made in networking and communications can be
found in section I.A.2.4 starting on page 23.

In the area of facility management and operation, several noteworthy events
occurred over the past year which will be explained in more detail here.

7 E. A. Feigenbaum
Details of Technical Progress 5P41 RROO7&5-11

SUMEX/HPP Welch Road Computing Facility

A major development this past year at SUMEX was the move of the Heuristic
Programming Project to their new location at 701 Welch Road, adjacent to the Stanford
Medical Center. Since this group is a major user of the SUMEX-AIM resource and the
focus for most of the core AI research, a good deal of effort was expended to provide a
robust computing environment at their new location. This development involved several
stages and levels of technical deveiscpment, ranging from construction of the machine
room, new cable and wiring installation, procurement and setup of networking hardware,
to major new developments in the networking software and a twisted pair Ethernet
communication link between this site and the main SUMEX Computer room. All of the
hardware and facilities purchases were funded from sources other than SUMEX.

We setup the general communications capabilities for the two buildings occupied by
the HPP. This involved wiring up local terminals, installing local Ethernets (both 3 and 10
megabit capability), and acquiring and installing networking hardware such as terminal
interface processors (TIP) and gateways, as well as extending the current SUMEX TIP
and GATEWAY software to handle both 3 and 10 megabit network traffic.

But the most important and most interesting development in this process was the
“twisted pair’ ethernet developed by the SUMEX engineering staff to allow high speed
reliable communications between this Welch Road facility and the SUMEX machine room.
Further information on this new ethernet can be found in Section I.A.2.4 on page 23.

HPP researchers are routinely using this link to communicate with SUMEX and the
central university network. In addition, various Lisp machines and printers located in the
HPP facility and connected to a local network are able to communicate with the
university network.

The end result is that we have successfully been able to extend the SUMEX
computing environment to a remote site, providing a high speed link to the facilities of
SUMEX while also allowing for local distributed processing. We see this experience has
being most valuable in the future as we move further into a distributed environment,
while still needing the sharing of resources and communication links provided by large
timesharing systems and local] area networks.

Digital Equipment Corporation stops development of 86-bit product line

Digital Equipment Corporation, a long time supplier of high speed 36-bit
timesharing computers to the Artificial Intelligence community, announced that it was
stopping all development of future 36-bit products, and instead starting a program to
provide a migration path to its line of VAX minicomputers.

Many DEC 20 customers had been anticipating a new yet unannounced machine
from DEC code named the 'Jupiter’, which had been reported to be a order of magnitude
faster than the current KL10 processor used in DEC20’s and DECI1O’s. However, DEC’s
announcement means this effort has stopped, and we can expect no more 36-bit products
from Digital Equipment Corporation.

The effect of this announcement to the AI community is disappointing, although
not totally unexpected. The DECsystem20 has been the predominant timesharing machine
used to support Artificial Intelligence based research, but yet researchers have been in
need of more processing power and larger address spaces for quite a few years. DEC has
clearly decided to devote their resources to VAX development. For those in need of
greater 36-bit processing power or address space, you must now look to newer less

E. A. Feigenbaum 8
5P41 RROO785-11 Details of Technical Progress

experienced companies such as Foonley and Systems Concepts for follow on 36-bit
products which those firms are preparing.

The impact of this decision on the SUMEX-AIM community must be examined in
conjunction with our development of AI systems on personal Lisp machines. We have
outlined very clearly our plans to move to distributed Lisp-based workstations for AI
Research, and this is clearly where we see the AI computing market heading. These
machines offer much better cost/performance ratios than timesharing machines, high
resolution bit-mapped screens, and powerful Lisp programming environments for the
development and eventual dissemination of AI based systems. However, this is not too say
we still do not see a role for the large timesharing machine in our environment. We still
believe in the use of a large central mainframe computer as the anchor for a large
community of users. The mainframe also functions as a central facility for communication
and collaboration, and provides fast Lisp cycles for program development where the
application is not in need of a specialized workstation.

Other SUMEX Computing Facilities

SUMEX continues to support other mainframe computers, file servers, professional
workstations, and assorted printers and terminals for use by the SUMEX-AIM community.

1. The SUMEX-AIM File server, based on a VAX 11/750 computer, continues to
serve the needs of the workstation users within SUMEX-AIM. The use of
SAFE by users of our 2060 is minimal. We plan to extend the use of SAFE in
the future by providing more convenient access by 2060 users than is currently
available.

to

. The VAX 11/780 computer system, originally purchased with DARPA funds
and previousivy located in Margaret Jacks Hall on campus, has become a
SUMEX-AIM resource this past year. The system was moved to a new location
on the Stanford campus which provides a better environment for a computer
of this size. This VAX is now shared between the Computer Science
Department and the SUMEX-AIM community.

3. SUMEX continues to support a wide range of professional workstations from
such vendors as Xerox, Symbolics, and Hewlett Packard for the development
and testing of AI applications. Additional work has been started to explore
the use of the Apple Macintosh and Apple Lisa within SUMEX. More
information on these developments can be found in section I.A.2.3.

9 E. A. Feigenbaum
Details of Technical Progress

a

5P41 RROO785-11

 

Centra! Processor

DEC KL10-E

2M words of memory, Cache

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

RH20 RH20 RH20 11/40 FRONT END DIB20
MassSus Mass8us MassBus UNIBUS 1/0 bus
Disk Controlier Disk Controller DEC AN20 La
ano Drive and Onve p—4  ARPAnet 50 Kbit
DEC RPO7 DEC RPOS intertace PS
MEIS La |
Disk Controlier MassGus DEC BA10
and Drive Ethemet 3 Mbit T
DEC RPO7 intertace
DEC LP10
Line Printer
Console TTY
2 DEC TU-78
Tape Drives
and Controller
KLINIK Line
Logging TTY
DEC LP-26
Line Printer
——| 6Line Scanners
—— DEC DH-11
96 Lines total
6 Lines Lae
TYMNET 4.8 Kbit
imtertace
Figure l: Current SUMEX-AIM Decsystem 2060 Computer Configuration

. Feigenbaum

10
5P41 RROO785-11 Details of Technical Progress

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Central Processor
DEC KS-10
512K words Memory
LA-36 Console
Unibus Adapter Unibus Adapter
| UNIBUS UNIBUS
ee
Massbus Adapter Ethernet Line Scanner
UNIBUS DEC DZ-11
Massbus Adapter Interface 16Lines [/--—
! Mass6Bus
| MassBus
ee
Disk Controlier Tape Controller
and Drive and Drive
DEC RP-06 DEC TU-45

 

 

 

 

 

 

Figure 2: Current SUMEX-AIM 2020 Computer Configuration

11 E. A. Feigenbaum
Details of Technical Progress 5P41 RROO785-11

 

Central Processor
DEC VAX 11/780

 

 

 

 

 

 

 

 

 

 

 

 

 

 

4 Mbytes Memory
Floating Pt Unit
LA-36 Console
MassBus UNIBUS
Disk Controller Line Scanner
and Drive DEC DZ-11
DEC RP-06 16 Lines
Disk Controller Ethernet
and Drive UNIBUS
DEC RP-07 Interface
Tape Controller Chaosnet
and Drive " UNIBUS
DEC TU-77 [ i interface

 

 

 

 

 

 

Figure 3: Current Shared VAX Computer Configuration

B. A. Feigenbaum 12
5P41 RROO785-11 Details of Technical Progress

 

 

Margaret Jacks Electrical Pine Halt
Heuristic Prog Proj Engineering
10 Mbit ethernet CSU

    
  
 

a a
SCORE 2060 Sumex VAX
Ether TIP CSLI 2060
Oandelions

Dover Printer
Other CSD Equip

NS file server

  

 

 

 

 

 

 

 

 

 

Oncocin Offices

 

 

Ether TIP
Dandelions

fr] HP 9836's

SUMEX Offices

 

 

 

Medical Center
SUMEX Mach Rm

 

 

Xerox Alto
R ater 2060 Dolphin
[a] epeate 2020 Dandelions
VAX - File Server SUN - Devel
[g] Gateway VAX - Devel Ether TIP
Ether TIP imprint-10 Printer

 

 

PDP-11 - Devel

 

 

 

 

 

 

 

 

10 Mbit ethernet

 

 

 

Whelan Building (Welch Road)
Heuristic Prog Proj

 

1.5 Mbit

phone line 10 Mbit ethernet

 

Dandelions
LM-3600’s

Ether TIP
Dolphins

Dorado
Raven Printer

 

 

q

3 Mbit ethernet

 

 

 

Figure 4: SUMEX-AIM Ethernet Configuration

13 E. A. Feigenbaum
Details of Technical Progress 5P41 RROO785-11

 

 

Central Processor
DEC VAX 11/750 Console TTY

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

2 Mbytes Memo
yt ry Tusa
MassBus UNIBUS
Kennedy
CDC 256 Mbyte ane oe Bet
Removable Fujitsu Eagle & Emulex
‘ 414 Mbyte
Disk Drive Disk Oe Controlier
System industries Disk Controller
9900 Disk Controtier — DEC AKo?
Fujitsu Eagle Fujitsu Eagle Ethernet
414 Mbyte 414 Mbyte UNIBUS
Disk Drive Disk Drive Intertace
OZ-11
Line Scanner
8 Lines

 

 

 

Figure 5: SUMEX-AIM File Server {SAFE}

R.A. Feigenbaum 14
5P41 RROO785-11 Details of Technical Progress

 

Central Processor
DEC VAX 11/750

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Console TTY
2 Mbytes Memory
TU 58
UNIBUS
Disk Controller Ethernet
and Drive UNIBUS
DEC RKO7 Interface
Disk Controller DEC DZ-11
and Drive Line Scanner
DEC RKO07 i 8 Lines

 

 

 

 

 

 

Figure 6: SUMEX-AIM Development Vax {ARDVAX}

15 E. A. Feigenbaum
Progress - Timesharing Systems 5P41 RROO785-11

I.A.2.3. Timesharing Systems

Continued support and development of our timesharing systems this past year has
concentrated on several areas, including improvement of user services such as printing
spoolers and archiving support, implementation of features from our KI10 Tenex system,
enhancing network interface service, correcting encountered system bugs, and
implementing new features for better user community support. In addition, we have
invested further effort in supporting the VAX/UNIX system in conjunction with the
SUMEX-AIM file server installation.

DEC system 2060/TOPS-20 System

Support of our main timesharing machine, the DECsystem 2060, has continued
during grant year 11.

e Hardware development

1. The DECsystem 2060 system now in operation at SUMEX differs greatly
from our previous KI10 Tenex system. Whereas before, the KI10 system
and TENEX software required much in-house development and support,
life is easier with the 2060. Being at Stanford University, where there
are at least 7 other DECsystem 20’s with similar hardware and software
is a great advantage. We are able to share our experiences with other
sites, and have become an integral part of the Stanford DEC community.
In addition, the DEC2060 hardware has been more reliable and easier to
maintain than the KI10 system.

te

. Additional modems were added to the 2060 to provide support for BELL
212A 1200 baud protocols. This adds an alternative to Vadic 1200 baud
service. Modems which use the Bell 212A standard are more widely
available for much less cost than Vadic modems.

© TOPS-20 Monttor Software Enhancements

1. A significant enhancement to our TOPS-20 monitor occurred this year
when we implemented the software from our Tenex system which allowed
extended support for the '?’ feature of TOPS-20 when parsing filenames.
This feature allows a user at any time to get a list of possible choices
when needing to input a file name by just typing ’?’. This returns an
actual list of file names, whereas in the standard TOPS-20 monitor, just
the string ‘input filespec’ was returned. This is a very significant and
useful change to our TOPS-20 system.

2. We continued to keep up-to-date on the various bug fixes and monitor
improvements that we received from DEC and other TOPS-20 sites.
These included several fixes and rewrites to the Internet IP/TCP code
which went under a major revision this past year.

3. We installed the capability for users to access their subdirectories as if
they were the owners of such. While this may seem to be the logical way
to implement subdirectories to begin with, DEC’s models of
subdirectories was a bit different. Our changes have since been installed
on other DEC20’s at Stanford and elsewhere.

4. We installed the capability to vary the allocation of windfall cycles in

E. A. Feigenbaum 16
5P41 RROO785-11 Progress - Timesharing Systems

accordance with user classes. This will allow us more flexibility in
assigning jobs enough cycles to run comfortably while limiting their
usage to a strict percentage of the machine.

5. We installed several new features in our TOPS-20 EXEC to facilitate the
use of the system by the users. Among these features was the ability to
edit any previous command you had entered, and then re-execute that
command. This saves extra keystrokes and has proven to be very useful.
The code to do this came from the University of Texas at Austin.

6. We switched our system this year to using encrypted passwords. This
means that passwords are not stored in any readable form on the
computer system, and if an illegal user should gain access to the system,
he/she would not be able to find out the passwords of any other users.
We feel this feature is quite important as the frequency of computer
break-ins/attempts increases.

7. Software was added to our monitor in order to record the last reader of a
file. Previously, only the date of the last read was recorded, while both
the writer and date were recorded for creating and writing a file. This
gives users the ability to determine which other persons may have been
reading files.

e Printer Support

The support on the 2060 for various printers in the SUMEX community has
been greatly enhanced this past year.

1. Support was added for the Xerox Raven printer at the Welch Road
facility to allow spooling and direct output to the printer. In addition,
code was added to the spooler to print out a header page identifying the
user, filename, and date for each job.

bo

. Similar spooler support was added for the Xerox Dover printer in
Margaret Jacks Hall.

3. SUMEX installed a Printronix line printer at Welch Road to allow users
to print out files remotely from the 2060. The Frintronix is connected to
SUMEX via a twisted pair serial line.

4. We transformed TENEX software to the normal TOPS-20 line printer
spooler program to look out for users who had accidentally printed an
"unprintable file’, meaning a binary file of some sort which does not
contain legible characters. We do this both by counting the number of
binary characters in the first page of the file, and by not printing the file
if the count exceeds a certain threshold. A similiar scheme is also used
taking into account the vertical motion of the first page.

5. Additional modifications were made to the LP10 line printer driver
software in the TOPS-20 monitor to improve the reliability of using this
line printer, which came from our KI10 system.

6. We greatly enhanced our support of the IMAGEN Imprint-10 laser
printer this past year. A new IP/TCP Ethernet interface was installed on
the printer (discussed further in Section I.A.2.4 replacing the existing
serial interface. This new interface allows for more efficient printer

17 E. A. Feigenbaum
Progress - Timesharing Systems 5P41 RROO785-11

operation, and greater flexibility in choosing output modes, number of
copies, header pages, and other features. We have also implemented a
TOPS-20 spooler for the Imagen as well.

e User System Software

1. We have continued to assemble and maintain a broad range of utilities
and user support software on the 2060. These include operational aids,
statistical packages, DEC-supplied programs, text editors, text search
programs, file space management programs, graphics support, text-
formatting and justification assistance, magnetic tape conversion aids,
and many more. We also are importing software tools and packages
wherever necessary to avoid reinventing the wheel and wasting our own
efforts. Packages have been imported from Texas Instruments, Columbia
University, the University of Texas at Austin, Yale University, and other
Stanford sites.

2. SUMEX has continued to provide to its users the latest releases of
various Lisp dialects that run under TOPS-20. This past year we agreed
to provide disk space to store the ’official’ version of Interlisp-20 from
XEROX due to the fact that the machine at Xerox used to support
Interlisp before was being removed. Interlisp-10 is now officially
maintained at SUMEX by our staff and XEROX personnel. In addition,
we continue to support the full variety of LISPUSERS packages.
Portable Standard Lisp (PSL) developed at the University of Utah has
also been installed on SUMEX.

3. We continue to use MM, a very powerful and flexible mail system, on the
2060. Electronic bulletin boards are also extensively-supported at
SUMEX. These provide a rather informal mechanism for community
discussions and debates. Other bulletin boards, read and contributed to
throughout the INTERNET community, are available for perusal at
SUMEX. These bulletin boards cover such topics as AI Discussions, Micro
Computers, Terminals, and Workstations.

4. SUMEX participates with other Stanford sites in a general license for
access to the SCRIBE text-formatting system from UNILOGIC, including
versions to run under TENEX, TOPS-20, and UNIX. SCRIBE is the
preferred tool for text preparation at SUMEX.

5. Versions of various user utilities and system utilities were updated
throughout the year. These programs included network server processes,
statistical packages, system daemon programs, and several programs for
processing electronic mail.

6. Various system network tables and networking software were updated to
accommodate the ARPANET split that occurred this year. The network
change effectively split the ARPANET into two networks, the MILNET,
which is a secure private part, and the rest of the ARPANET, which
operates as the original ARPANET did before.

e Documentation and Education

We have expended considerable effort to develop, maintain, and facilitate
access to our documentation so to accurately reflect available software. The

E. A. Feigenbaum 18
5P41 RROO785-11 Progress - Timesharing Systems

HELP and Bulletin Board subsystems have been important in this effort. As
subsystems are updated, we generally publish a bulletin or small document
describing the changes. As more and more changes occur, it becomes more
difficult for users to track down all of the change pointers. Within manpower
limits, we are in a continual process of reviewing the existing documentation
system for compatibility with the programs now on line and to integrate
changes into the main documents. This also will be done with a view toward
developing better tools for maintaining up-to-date documentation.

e Software Sharing

1. As stated previously, we firmly believe in importing rather than
reinventing software where possible. As noted above, a number of the
packages we have brought up are from outside groups. Many avenues
exist. for sharing between the system staff, various user projects, other
facilities, and vendors. The advent of fast and convenient
communication facilities coupling communities of computer facilities has
made possible effective intergroup cooperation and decentralized
maintenance of software packages.

to

. The TENEX, TOPS-20, and UNIX sites on the ARPANET have been a
good model for this kind of exchange based on a functional division of
labor and expertise. The other major advantage is that as a by-product
of the constant communication about particular software, personal
relationships between staff members of the various sites develop. These
collegial interactions serve to pass general information about software
tools and to encourage the exchange of ideas among the sites. Certain
common problems are now regularly discussed on a multi-site level.

3. We continue to draw significant amounts of system software from other
ARPANET sites, reciprocating with our own local developments.
Interactions have included mutual backup support, experience with
various hardware configurations, experience with new types of computers
and operating systems, designs for local networks, operating system
enhancements, utility or language software, and user. project
collaborations. We have been able to import many new pieces of
software and improvements to existing ones in this way. Examples of
imported software include the message manipulation program MM, SAIL,
PASCAL, SOS, INTERLISP, the C cornpiler, VAX Ethernet code, the
PHOTO program, ARPANET host tables, various user utilities, and
many others.

4. Finally, we also have assisted groups that have interacted with SUMEX
user projects in acquiring access to software available in our community.
We are repeatedly providing tape preparation and copy service to many
SUMEX-AIM projects to aid in sharing their software with outside
requestors.

19 E. A. Feigenbaum
Progress - Timesharing Systems 5P41 RROO785-11

DECsystem 2020/TOPS-20 System

1. Monitor Upgrade -- Our 2020 system has continued to run very reliably this
past year. We have updated the 2020 monitor with bug fixes and performance
improvements regularly. There will likely be few further monitor releases for
the 2020 since it does not support extended addressing and there are no plans
to add this feature.

bo

. Demo Controls -- We continue to use the 2020 system for demos of AI systems
developed at SUMEX. This demo system takes advantage of the "pie-slice"
scheduler in the TOPS-20 release 4 monitor. We now guarantee dedicated
users a large fraction of the machine but also allow others to do useful work
when the demo demand is low. This system has nicely met the needs of both
groups.

VAN /UNIX Systems

We continued to provide systems support for the VAX/UNIX 11/780 system
(named "AIMVAX’) shared by the SUMEX-AIM community and Stanford Computer
Science Department. Various efforts included supporting the UNIX monitor, installing
new network software, and in bringing up various user subsystems.

Further development has continued in support of the SUMEX-AIM File Server
(SAFE) based on a VAX 11/750. We successfully converted SAFE to Berkeley Unix 4.2
server, and with the help of the Computer Science Department, converted the Ethernet
Pup software to run under UNIX 4.2

B. A. Feigenbaum 20
5P41 RROO785-11 Progress - Professional Workstations

1.A.2.4. Professional Workstations

Our ongoing movement to professional workstations is taking on several forms. We
continue to carry out our acquisition plans for acquiring Lisp machines for use by the
SUMEX community, as well as investigating the use of remote virtual graphics and new
lower cost workstations in our environment. This work is prototypical of what other
groups will face and we hope will serve to find effective solutions to common problems.

Lisp-based Scients fie Workstations

SUMEX carefully developed and implemented our equipment acquisition plan for
year eleven by buying seven Xerox 1108 Lisp machines for use by SUMEX-AIM projects.
Two of these machines were purchased with special upgrade packages to provide floating
point capability, expanded microcode, and expanded memory. Our experiences with these
machines will be reported in next years report.

The XEROX Dolphin on loan to Rutgers University was returned to SUMEX this
year. This Dolphin had effectively served the Rutgers-AIM community in setting up their
Ethernet network and provided initial exposure to the Lisp machine technology. Now
that that experiment is successfully completed, the Dolphins will be used for AI system
development at Stanford.

SUMEN installed two SYMBOLICS 3600 Lisp machines, purchased with DARPA
funding, for use within the Heuristic Programming Project (HPP) at their new location at
Welch Road. We are currently awaiting a new release of the Symbolics Operating System
software before we can provide Ethernet access to our file server from these machines.
The 3600’s are used regularly by members of the HPP.

We still are using 4 preproduction models of the Dolphin workstations. One
preproduction model has been exchanged for a production system, and we are on schedule
with XEROX to exchange the remaining 4 machines for production models at no extra
cost. This process is hampered by the rate at which XEROX themselves can get
production machines. ,

We studied the benefits of buying the extended memory and microstore upgrades
to the Xerox 1108 Dandelion announced at AAAI-83 as being “under development." We
concluded that some users would benefit greatly from these enhancements and others not
at all. The most marked improvements came from system which were extremely memory
limited, such as NEOMYCIN. SUMEX will be acquiring two 1108’s with the upgrades for
full time use and testing.

A close relationship between SUMEX and the newly-formed Center for the Study of
Language and Information (CSLI) at Stanford was established. This has already benefited
SUMEX (and the ONCOCIN project in particular) in the loan, by CSLI to SUMEX, of
two Xerox 1108's which have been in constant use by researchers since January 12th. The
SUMEX staff assisted CSLI in bringing their DecSystem20 and network environment on-
line. CSLI has informally expressed an interest in working on the problem of distributed
Al computation with SUMEX researchers. CSLI will have 110 1108’s on the Ethernet
within the year. This resource suggests some exciting solutions to former cormpute-bound
problems. The ONCOCIN group has already implemented a preliminary network-based
Interactor which permits elements of ONCOCIN to run concurrently on different
machines. As of this writing, the Reasoner and its Debugger have been made to run
transparently in this mode, and to make good use of both processors.

21 E. A. Feigenbaum
Progress - Professional Workstations 5P41 RROO785-11

Virtual Graphics

SUMEX continued the development of a Virtual Graphics system written in
Interlisp-10 and running on our 2060. Any user running the V system on a workstation
can then use the package on the 2060 to drive the graphics display on the workstation. A
current application is to take nuclear magnetic resonance data on the 2060 and display
the atoms and their bonds on 2 SUN workstation by using splines. This development is in
its infancy, but is opened ended and has great potential with the price of workstations
capable of decent graphics reaching the two to three thousand dollar range. It allows those
users who cannot afford expensive lisp machines to have full graphics power available to
them by doing the actual graphics applications on a large time shared system, and then
doing the graphics itself remotely on a less expensive workstation. This development can
help users take advantage of the computing power of the DECsystem 20, while providing
many of the high speed graphics advantages of the Lisp Machines.

Apple Workstation Development

SUMEX-AIM has initiated a development project to pursue the effectiveness and
possible use of low ccst personal workstations within our environment. After examing a
number of new personal computers and workstations on the market, such as the Hewlett
Packard 150, IBM PC, Sun workstations, and others, we chose the Apple Macintosh and
Apple Lisa on which to begin our work in this area. These machines were chosen
technically due to their built in graphics, networking, mouse, windows, and menu support.
We also considered the very beneficial relationship formed between Apple and Stanford
University which provides us direct access to Macintosh hardware and software
documentation which is a necessity for the type of work we plan to do.

Our Macintosh development encompasses several areas ;

1. INFO-MAC Discussion List

An electronic discussion list was originated at SUMEX, and is currently
maintained here, to foster sharing and communication among research groups
and universities that are interested is pursuing the serious use of the Macintosh
within their respective environments. This list has been highly successful in
collaborating on Macintosh development and the sharing of ideas. The
discussion list currently contains over 50 sites, and well over 1000 participants.

to

. C Development Environment

A vital link in our development of Macintosh software is creating a C based
development system on our VAX computers for the coding and downloading of
software. Utilizing existing MC68000 C cross compilers on our VAX, we are
developing the necessary linkages in order to make the appropriate system
calls to routines in the Macintosh ROM’s for sophisticated graphics and system
related functions.

3. Macintosh print servers

In order to effectively use the Macintosh as a stand-alone workstation, we will
provide the ability to print out Macintosh developed files on our IMAGEN
laser printers.

4. Applebus to Ethernet Inter face

This development involves the hardware and software necessary to be able to

E. A. Feigenbaum 22
5P41 RROO785-11 Progress - Professional] Workstations

access various file and print servers on our Ethernet from a Macintosh. The
Macintosh will be connected to the Apple network called Applebus. Our
hardware provides an interface between Applebus and 1OMB Ethernet. The
software necessary for this project involves formulating Macintosh file level
and block level I/O requests into properly formatted Internet packets.

5. Virtual Graphics on a Lisa

The Virtual Graphics System, as previously reported, is in great need of a low
cost workstation on which it can run. We have started a project to port the
Virtual Graphics system to a Apple Lisa in hopes of providing to our users
high speed graphics at remote locations. We will report further on this project
in next year’s report.

Anticipating the popularity of our Macintosh developments, we are fully prepared
to make our efforts available to other research sites, Universities, and non-profit
institutions on a royalty-free basis in hopes of fostering continued development and
communal sharing.

{In addition to Lisp-based scientific workstations, we believe the use of low cost
workstations, which offer suitable local processing power, high resolution screens with easy
to use user interfaces, and networking and communications abilities, are vital to the
future of our resource. Our Macintosh and Lisa development efforts will allow us to use
and experiment with these workstations in our environment.

Hewlett Packard Development

SUMEX assists the ONCOCIN Project is developing a computing environment for
developing AI applications based on HP 9836 workstations. These workstations were part
of a gift from Hewlett Packard to the Oncocin Project. Additional support peripherals for
the 9836’s included large capacity disk drives, color monitors, graphic tablets, and a laser
printer. Work is proceeding to network these machines onto the SUMEX Ethernet as soon
as suitablke networking hardware is available from HP. These machines will be used for
new and existing projects within Oncocin.

23 E. A. Feigenbaum
Networking and Communications 5P41 RR0OO785-11

1.A.2.5. Networking and Communications

A highly-important aspect of SUMEX-AIM is effective communication with remote
users and between the growing number of machines available within the SUMEX resource.
In addition to the economic arguments for terminal access, networking offers other
advantages for shared computing, including improved inter-user communications and
more effective software sharing.

Users accessing a remote computer will use a hardline connection to the computer
as a standard of comparison. Local networks stand up well in this comparison but remote
network facilities do not. Data loss is not a problem in most network communications; in
fact, with the more extensive error checking schemes, data integrity is higher than for a
long distance phone link. On the other hand, remote networking relies upon shared use of
communication lines for widespread geographical coverage at substantially reduced cost.
However, unless enough total line capacity is provided to meet peak loads, substantial
queueing and traffic jams result in the loss of terminal responsiveness. We continually
monitor the load statistics for our direct, dialup, and TYMNET lines to avert logjam
situations.

TYMNET

TYMNET provides broad geographic coverage for terminal access to SUMEX from
throughout the country and increasingly from foreign countries. With the installation of
our new DEC2060 computer system in January of 1983, we installed new TYMNET
equipment. After the initial debugging of the new equipment (called TYMCOM) the
equipment has been quite reliable. However, some months after the installation it was
discovered that the XON/XOFF protocol between the Tymcom and the 2060 had not
been properly specified in the Tymcom and was corrected. The number of user
complaints about connection problems have been greatly reduced. This is thought to be
the result of improved "backbone" lines within Tymnet and the installation of triple-duty
modems which simplify things for the users.

ARPANET

We retain our advantageous connection to the Department of Defense’s
ARPANET, now managed by the Defense Communications Agency (DCA). This
connection has facilitated close collaboration with the Rutgers-AIM facility and many
other computer science groups that are also on the net. We have maintained good
working relationships with other sites on the ARPANET for system backup and software
interchange. Such day-to-day working interactions with remote facilities would not be
possible without the integrated file transfer, communication, and terminal-handling
capabilities unique to the ARPANET. The ARPANET is also key to maintaining on-
going intellectual contacts between SUMEX projects such as the Stanford Heuristic
Programming Project authorized to use the net and other active AI research groups in the
ARPANET community.

This past year, SUMEX-AIM participated in the split of the ARPANET into two
networks; the MILNET, which is a highly secure strictly DOD-related part of the network,
and the ARPANET, which is the remainder of the ARPANET sites. This latter net
functions as we knew the ARPANET before. The MILNET can only be accessed via mail
gateways. No TELNET or FTP to MILNET sites is allowed. In addition, access to the
ARPANET TAC’s (Terminal Access Controllers) was restricted this past year to only
those users who were granted TAC access cards, which meant their username was
registered with the Network Information Center, and they were given a password with

E. A. Feigenbaum 24
5P41 RROO785-11 Networking and Communications

which they could dial into the ARPANET. SUMEX arranged for guest cards for those
users who needed such access.

We continue to be called upon to interact with outside organizations which are (or
wish to be) connected to our IMP. The line to Advanced Information and Decision
Systems occasionally causes trouble requiring diagnosis. The intended connection to
Perceptronics Inc. has evidently been canceled.

ETHERNET

A substantial portion of our system effort this past year went into continued
development of local Ethernet facilities which link the SUMEX resource hardware with
other parts of the campus, namely to 701 Welch Road, which is the new location of the
Heuristic Programming Project, and to the Computer Science Department building on
campus. We have also invested a great amount of effort this year to begin our transfer to
a 10 megabit Ethernet, while continuing support of our current 3 MB ethernet.

Specific areas of Ethernet development include:

1. Leaf server -- We continued support of the Sequin reliable packet protocol and
Leaf byte-level file transfer protocol to enable our Xerox D machines to access
files on our DEC20 systems. The Leaf server had to be modified on the DEC20
this past year when we switched to using encrypted passwords. The LEAF
server implementation for the 4.2 BSD release of Unix was also debugged and
installed at SUMEX. This allows us to access files stored on our VAX file
servers from either our 10MB or 3MB networks.

The Leaf protocol is built into the lowest levels of the Dolphin I/O system, and
allows any file on a remote file server to be accessed as easily as a disk file in
both paged or random access mode. The latest updates to the Sequin
transport level have made marked improvements in efficiency. The 2020 now
performs Leaf file transfers with a speed approaching that of XEROX’s
dedicated file server.

2. TOPS-20 Ethernet Server -- We continued to maintain and improve the
Ethernet service under the TOPS-20 operating system. This included updates
to the TELNET and FTP programs, as well as mail software, the previously
mentioned Leaf server, and network table maintanence programs.

3. Ethernet Gateway -- Our Ethernet gateway software has continued to run
reliably and effectively. The previous problems with lost packets and delayed
terminal response has been fixed, the cause of which was a bad memory board
and a software bug in the TOPS-20 operating system. Serious problems that
affected our net connectivity to other parts of campus were also discovered
and repaired this past year, thereby providing us with over 99° net
connectivity to the rest of campus. The changes involved board repair and
modifications to the topology of the campus Ethernet.

The gateway itself was generalized to handle 3 or more directly connected
networks where previously it had only dealt with 2 such networks. We
currently have two gateways, each handling the traffic between three local
networks, two of which are 3MB and one a 10MB network.

4. Ethernet TIP (EtherTIP) -- The EtherTIP provides multiple terminal access to
the Ethernet. A PUP ethernet operating system was written for MC68000-
based processors, and a MC68000-based EtherTIP was built based on this.

25 E. A. Feigenbaum