perm filename NEWS75.PRO[CUR,JMC] blob
sn#148441 filedate 1975-03-04 generic text, type C, neo UTF8
COMMENT ā VALID 00008 PAGES
C REC PAGE DESCRIPTION
C00001 00001
C00002 00002 \\M0BASL30\M1BDR40\M2FIX25\M3BASL30\M4BASI30\M5BASB30\.
C00004 00003 \F4
C00009 00004 \C\F5BEHIND THIS PROPOSAL
C00043 00005 \CPRINCIPAL INVESTIGATORS' DIVISION OF LABOR
C00045 00006 \CWORK SCHEDULE
C00047 00007 \F2\CEXPERIMENTAL ELECTRONIC NEWSPAPER
C00050 00008
C00052 ENDMK
Cā;
\\M0BASL30;\M1BDR40;\M2FIX25;\M3BASL30;\M4BASI30;\M5BASB30;\.
\F1\CResearch Proposal Submitted By
\CStanford University
\CStanford, California
\Cto
\CTHE NATIONAL SCIENCE FOUNDATION
\Cfor
\CEXPERIMENTAL ELECTRONIC NEWSPAPER
Proposed starting date: 1 July 1975
Two-Year period ending
30 June 1977
Amount Requested (by fiscal year):
FY 76 $138,831
FY 77 $122,254
\F0_________________________________________
John McCarthy - Co-Principal Investigator
Professor of Computer Science
___________________________________________
Edwin B. Parker - Co-Principal Investigator
Professor of Communication
____________________________________
Dan Drew - Co-Principal Investigator
Professor of Communication
____________________________________
For the University
file: news75.pro[cur,jmc]
\C\F1February 1975
\F4
\CAbstract
\F3\J Economic, environmental, and technological forces are pushing toward the
day when computer terminals in the home will provide easy access to many information
sources. A major impetus toward widespread use of the home information utility is
likely to be the development of electronic newspapers. Before the electronic
newspaper and multi-use home terminals become a reality, however, research will
be needed in the areas of man-machine interaction, file handling and retrieval
techniques, computer-to-computer interfaces, and readership patterns.
To perform this research, support is requested over a two-year period so
that an experimental electronic newspaper can be established, made operational,
and evaluated. The newspaper would be offered as a campus service at Stanford
University, with segments of the student and faculty populations as its readership.
We see the eventual conversion of newspapers to electronic form
as a potential solution to problems of information monopoly.
\F5First\F0, the cost of entry to the newspaper business will be reduced since
the newspaper will not require a production or distribution organization
and will be able to live in publicly available time-sharing systems.
\F5Second\F0, if there is no production imposed limit to the size of the
newspaper, it will be easier for a newspaper to cater to the interests
of special groups. \F5Third\F0, since all "letters to the editor" are automatically
and instantly accepted,
individuals or groups considering themselves unfairly treated
in stories will be able to make their replies available within minutes
of noticing the alleged unfairness. Since this will be before most
readers have read the individual story, and since a reader will be able
to ask "and what does the Governor have to say about that" immediately
after reading the story, the danger of hit-and-run journalism will be
reduced. We will experiment with this
facility locally at Stanford, but the experiment will become realistic
only when there are enough terminals so that the subjects of news stories
find it worthwhile to keep track of what is said about them and reply.
In the above respects, \F4The Stanford Electronic Times\F0 differs
decisively from other experiments with new ways of providing information
to the home. In these systems, the user
is passive in two respects. First, his capacity to select information is
limited, and second, he doesn't have an effective low-overhead right to
reply or originate information.\.
\C\F5BEHIND THIS PROPOSAL
\F0\J Experts in the computer science and communication fields envision a day
when most American homes will have computer terminals providing speedy access to
a multitude of information sources. Using a home information terminal, a person
could call up books or reference files from the library, receive information
about goods and services available in the community, or read stories produced by
the local newspaper.
Benefits of such a system would be tremendous. Individuals would be able
to obtain information more quickly and easily than ever before. They would also
have more control over the information they received. The system would be
receiver-oriented, not sender-oriented as is the present system. A reader, for
example, could bypass news summaries prepared by newspaper editors and search their
files for information about subjects of particular interest. Newspapers could
provide greater quantities of more varied information because they would not be
trapped by the strict space limitations presently imposed by economic constraints.
Also, publishing opportunities would expand because a large organization would not
be necessary to produce a newspaper or magazine. The environment also would
benefit: Fuel presently used to deliver newspapers, books, and magazines would be
conserved and trees now destined to become newsprint would be saved.
A major impetus toward widespread use of home information terminals is
likely to be the development of electronic newspapers. Publishers are turning to
computer technology in an effort to increase the speed and efficiency of their
production departments. In some operations, reporters and editors ply their
craft on computer-operated video display terminals. Optical character recognition
scanners convert copy to machine language for processing in a phototypesetter.
Newspapers are also using the computer for information storage on a limited scale.
The \F4New York Times \F0 has established an information bank which is available
to subscribers and which shows promise for future growth.
A number of forces are being felt by the newspaper industry, however, that
might push the use of computer technology one step further -- into the newspaper
reader's living room. Foremost among these is a newsprint shortage that is
expected to worsen. This shortage -- and the probable cost increases associated
with it -- could make the video terminal located in the subscriber's home an
attractive alternative to newsprint.
Despite gains in production efficiency through the use of computer
technology, the newspaper distribution system, for the most part, remains as it
has for years. Men load bundles of bulky newspapers on trucks that take the
papers to newspaper carriers, who in turn, deliver it to your front door. An
electronic distribution system would reduce the time between the lockup of the
paper and its arrival at the home. In fact, lockup of the newspaper, in the
current sense of the term will be unnecessary. Cable television, which is
expected to cover most American homes in the next two decades, may be the eventual
mode of transmission for the electronic newspaper, but telephone transmission
is also quite feasible.
Before the electronic newspaper and the multi-use home information terminal
become reality, however, a number of problems will need thorough research. This
proposal seeks to develop an experimental electronic newspaper on the Stanford
University campus which would serve as the basis for research in the following
areas:
1. \F4Man-machine interaction. \F0An electronic newspaper must be designed
so the casual user, who knows little or nothing about a computer, can use it
easily without reading anything longer than a placard on the terminal. The
"reader" also should be able to use the system after a long absence without having
to relearn procedures. This requires the development of a uniform and easily
remembered style of interaction.
These same problems apply to the communicator's use of the new technology.
The system must be designed so reporters and editors can insert and update
information in a way that would not hamper their journalistic functions.
2. \F4File handling and retrieval techniques. \F0The Associated Press
A-wire alone produces some 375,000 characters per day. Because a newspaper
collection would be kept over long periods of time from several wire services
and its own local sources, news data indexing and handling techniques must be
developed to handle the tremendous volume of information.
3. \F4Computer interface. \F0Because of the large amounts of information
from various sources that would have to be stored on computer, it is necessary
to develop ways of making disparate computers interface smoothly and divide a work load.
The long range goal of making all of the information in the country accessible from
any one location is not being solved by creating a national information system.
Instead, many information systems with limited goals are being developed and will
need to be interlinked.
4. \F4Reader Feedback. \F0With an electronic newspaper, it would be
possible to monitor reader choices electronically and provide editors with this
information. Readers also would be able to use their terminals to respond to the
newspaper via a "letter to the editors" system. A serious concern of communication
researchers is the effect of this feedback on the attitudes and behavior of
journalists.
5. \F4Readership Patterns. \F0Newspaper readers would gain more control
over information they received with the electronic distribution system. As mentioned
above, they would be able to bypass editorial decisions and search for information
of particular interest. How this capability affects the reader's behavior and
attitudes is still another matter of interest.
6. \F4Reader-interest Profiles. \F0A selective dissemination-of-information
type system would be developed through which readers could describe their personal
interests and request the system automatically to find stories conforming to these
interests. The goal would be to maximize the efficiency and the flexibility of
the system for the reader.
The Stanford University Campus provides an ideal setting for conducting such
a project. The Artificial Intelligence Laboratory and Communication Department have
the computer expertise and research skills necessary for daily operation of the
newspaper.
Most of the necessary hardware and software to support the electronic
newspaper are already available. The Stanford Artificial Intelligence Laboratory
has a time-sharing system with a PDP-10 computer, while the Stanford Center for
Information Processing operates an IBM360/67; the /67, which serves the main
campus, supports an estimated 150 electric typewriter and CRT terminals at a
variety of locations. A maximum of 126 of these can be on-line at any given
time. In addition, the Artificial Intelligence Laboratory,
with 70 terminals and located approximately
three miles from the main campus, currently operates a system which provides for
automatic input of the Associated Press A-wire
and the \F4New York Times\F0 News Service into the PDP-10. Users can have
displayed stories about any topic in which they are interested. Persons using
one of the terminals may request that the computer interrupt their work when
bulletins or new information about a topic come over the wire.
The production and distribution of local news and information would be
handled within the context of SPIRES II, an information retrieval system.
SPIRES was developed by the Institute for Communication Research
under a grant from the National Science Foundation.
The Stanford Center for Information
Processing (SCIP) now operates SPIRES II which became operational in 1972. With
SPIRES II, a user can create, update, maintain, search, and display stored
information. The user may work with either a private data base of just a few
records or with a public data base that includes information ranging from a
restaurant guide to a Library of Congress "card" catalog file
(which contains over 170,000 records).\.
\F5\CPLAN OF ACTION
\F0\J An experimental electronic newspaper would be established on the Stanford
University campus. Its operation would provide the reader with news about important
national and iternational events and information about developments in the Stanford
area. It would contain some local news, science information (news about important
national and local developments in science) and updated information about scheduled
events, movies, restaurants, concerts, etc.
The base for the paper would be the Associated Press and the \F4New York
Times \F0 News Service which would be stored on a PDP-10 computer at the Artificial
Intelligence Laboratory. Efforts would be made to expand the file by adding other
wire services. Some of the information including local news and
science information, would be stored on the
IBM 360/67. Readers would be able to call up information stored on either computer
at any of the terminals located on the campus. They could either request a
"front page" containing a summary of the day's major events or an index in
headline form that summarized important information not included on the front page.
Establishment and operation of the interface between the two computers would
produce the research about machine interface that must be solved if persons are to
be able to retrieve information from disparate systems.
A graduate student with professional journalism training would serve as editor
of the newspaper. Each day, the editor would prepare a front page that included what
he perceived to be the major local, national and international events. He also would
write a news summary in headline form of events not included on the front page.
Depending on the amount of time available, the editor might also prepare a sports
page, a financial page, etc.
The editor would be assisted by a clerk who would double as a reporter.
The clerk would update information about local events, such as movie and concert
schedules, and would occasionally write stories about important local news
events. The clerk would also gather and assemble science information of interest
to the Stanford community. This would include important national science developments
reported by the wire services; campus developments such as details about conferences,
grants, important research resulls, available papers, etc.
Two full-time computer programmers and one working part-time computer
programmer would deal with the computer problems. They would develop data
indexing and handling techniques necessary to handle the voluminous amounts of
information that would have to be stored for the newspaper, and would continue to
refine these techniques so that operation of the system, for both communicators
and receivers, was as simple as possible. Additionally, they would deal with the
interface between the two computers used in the project.
A graduate student would serve as research assistant to help the
investigators collect and evaluate data obtained during the project.
Graduate and undergraduate students enrolled in Communication 175,
Public Affairs Reporting, which is taught each spring quarter by Professor
Dan Drew, would participate in actively producing the electronic newspaper.
Some of the stories they covered during their class exercises would be placed in
computer storage to provide readers with news about important local events.
Because of the availability of terminals on the campus, the
electronic newspaper can be developed and reader acceptance tested in a
preliminary way without buying more terminals. A larger scale test will
require more terminals, but perhaps this can be done with multi-purpose
terminals used for other computing and communication purposes.
It will, however, be necessary to
purchase a multiplexer in order to increase the capabilities of the PDP-10, and
develop an interface between the PDP-10 and the IBM 360/67.\.
\F5\CRESEARCH AREAS
\F4COMPUTER SCIENCE RESEARCH
\F0\JThis project involves research in three areas of computer science -
the man-machine interface, the creation, maintenance and retrieval of
data bases, and the interaction of programs in different computers.
1. \F4The man-machine interface. \F0
The \F4electronic newspaper\F0 provides an excellent opportunity for
research in man-machine interaction, because the users will differ widely
in their motivation to overcome difficulties and because many features
can be provided that differ widely in their intrinsic difficulty of use.
Our goal is that the \F4newspaper\F0 will prove congenial to users
instructed only by a placard on the terminal and the prompt features of
the program. Moreover, a user who is experienced with some of the features
should not be frustrated by excessive dialog in familiar areas but should
be able to get help with less familiar features. It should also be
convenient to return to using the program after months of absence.
The Stanford Artificial
Intelligence Laboratory has accumulated much experience in designing
and using programs that interact with users at terminals. Programs
from which we have learned include: the time-sharing system itself,
text editors, message sending and receiving systems, on-line desk
calculator simulators, interpreters like LISP, several variants of
our news service program, spelling checkers, file directory editors,
a program for charging purchases in our commissary and making the
vending machine give out food, and many others. The users of these
programs include experienced programmers, beginning programmers,
secretaries experienced in operating our editors etc., and temporary
secretaries who have to operate them with only 15 minutes
instruction. One may use a program constantly or only a few times a
year. Many people do by hand jobs for which there is a readily
available program simply because they find it inconvenient to learn
or relearn how to use the relevant program (or even an unfamiliar
feature of a program they use constantly).
These problems will never go away completely, but the problems they
occasion can be minimized by a proper style of man-machine
interaction. We have made a number of discoveries about this some of
which are perhaps counter-intuitive and don't yet get wide assent
among designers of interactive programs. \F5First\F0,
in so far as possible, interaction should be on one level rather than
hierarchical; i.e., in a program with modes and submodes, the user often
forgets what mode he is in, what mode the various facilities require,
and even when he remembers, he often finds himself in the wrong mode.
This is a special case of the general principle that the user will
have difficulty whenever the designer of the program introduces a
new concept rather than a notation for a concept that the user already
has in his mind.
\F5Second\F0, there should always be readily available help from the
program itself in case the user forgets some facet of its use. It is
convenient for the user always to be able to get help by typing "?" instead
of whatever command the program might expect next. In general, we are
talking about adapting the program to ease of use for the average person,
not for experienced programmers. Of course, even experienced programmers
benefit from improved clarity of program use.
When displays are available, a part of the screen can be used to
tell what options are available to the user.
2. \F4File handling and retrieval techniques. \F0In the course of
the last decade many computer-based data bases have been created by
organizations all over the country, and many more will be created in
the next ten years. Each of them has been designed by itself and has
been designed to interact with a specific kind of terminal with a
specific style of dialog. A user of information may want to refer to
many different data bases, and this will not be feasible if he needs
a different terminal and a different style of dialog for each one.
It would be convenient for the user if a single standard style of
information storage and interaction would be developed and used
nationally, but this is quite unlikely to happen - for bad reasons
like unwillingness to look at other people's work, and for good
reasons like the availability of different hardware and software to
different people at different times. We wish to explore the
following potential way out of the difficulty:
Instead of trying (probably vainly) to get people to standardize their
data bases, \F2we shall develop a standard way of describing data bases\F0 - what
information is stored therein and how it is represented. Instead of a standard
interaction style, \F2we shall develop a standard way of describing interaction
styles. \F0Our users will make their inquiries in our style to which they have
become accustomed, and a program will use the data base description and
interaction description to interface our user and the foreign data base. A
possible initial example is the \F4New York Times\F0 Data Bank.
Another possibility
is to interface with ticket reservation systems so that an \F4Electronic Times\F0
reader who has seen an event mentioned in the "paper" can make a reservation.
The work of the Electronic Systems Lab at MIT in developing a common interface
to multiple data base systems will be taken into account.
We are also aware of the considerations advanced in \F4Design of Man-Computer
Dialogues\F0 by James Martin (Prentice-Hall 1973), but the electronic
newspaper situation will permit more continuous optimization of the
interactive system than did any of the programs described in that book.
We also want to see how far we can go in allowing more sophisticated
criteria for information selection than categories determined by human
editors or the presence of words and phrases in the text. For example,
we would like to the system to classify nuclear reactors with energy and
nuclear weapons with disarmament, but as the reader no doubt sees, even
this has its problems.
3. \F4Inter-computer interfaces. \F0It will become more and more common
in the future to want to interconnect computer systems so that users of
one system (or perhaps the system itself) can have access to
the resources (such as information, services, and computation power) of
other systems. For \F4The Stanford Electronic Times,\F0
a computer link between the 360/67 located
on the main campus and the PDP-10 located at the Artificial Intelligence Laboratory
would be established. The PDP-10 interface would be developed by the AI Lab.
The /67 interface would be developed by SCIP.
These are technical problems that have been solved before. The interesting
problem is how to arrange matters so that the user of one computer can use
facilities of the other without having to learn the details of its system.\.
\F4BEHAVIORAL-JOURNALISTIC RESEARCH
\F0\JBesides the computer science research descibed above,
this proposal would permit research in the following
behavioral-journalistic areas.
1. \F4Reader Feedback. \F0An electronic monitoring system would be
developed which would tally and record information about reader preferences.
This would include data about the types of stories readers were choosing, the
amount of time users were spending with the system, the number of times they
bypassed the editor's decisions, and the number of persons using the electronic
newspaper. Readers also would be able to use their terminals to insert "letters
to the editor" into the system. Because a "letter to the editor"
could be written and sent in a fraction of the time it takes to write
and mail a letter to the editor to a conventional newspaper, and
because the "publication" of all "letters" is guaranteed, we would
expect much more vigorous response than a conventional newspaper
gets. These data and the letters would provide valuable descriptive
information about reader response to the electronic newspaper.
The data also would be given to the editor, and his behavior would be
observed to determine the effects of such feedback on journalistic
decision making. Interviews and semantic differential scales would
be used to measure any changes in the editor's perceptions of his
audience, as a result of the feedback.
2. \F4Reader Interest Profile. \F0A selective dissemination-of-information
system would be developed. On the basis of information the reader provided at the
terminal about his or her interests, the system would search out stories relevant
to these interests. Electronic monitoring would be developed to provide data
about the extent to which such a system was used and the types of information
readers requested. Interviews and direct observation would determine reader
response to the system and would further determine what kinds of problems, if
any, were encountered by the readers.\.
\CPRINCIPAL INVESTIGATORS' DIVISION OF LABOR
\JProfessor John McCarthy will be in charge of the computer science
research component of the project, the PDP-10 software, and the
interconnection between the PDP-10 and the IBM 360/67.
He will also take part in
using the \F4Stanford Electronic Times\F0 to
evaluate the social, economic, cultural
and political effects of electronic newspapers.
Professor Edwin Parker will act as administrative coordinator of the
entire project. He also will be in charge of the 360/67 software and
will act as consultant for the behavioral research component.
Professor Dan Drew will supervise the journalism component and the
behavioral research outlined in the proposal.\.
\CWORK SCHEDULE
\F3July - August, 1975
\F0* Establish link between PDP-10 and 360/67.
* SPIRES file defined for experimental electronic newspaper.
* Visit East Coast newspapers to study computerized production
systems currently in use.
\F3September - November 1975
\F0* Prototype Experimental newspaper begins operation with participation
of Communication 175 students.
* Collection of data about software problems.
\F3December 1975 - February 1976
\F0* Reprogramming and reworking software on the basis of September through
November experience.
\F3March - November 1976
\F0* Resume operation of experimental electronic newspaper with
Communication 175 students participating again during the Fall quarter.
* Collection of data in behavioral areas.
* Collection of data about software problems.
\F3December 1976 - February 1977
\F0* Reprogramming and reworking software on the basis of experience with the
experimental electronic newspaper.
\F3March - May 1977
\F0* Resume electronic newspaper operation.
* Continue data collection.
\F3May - June 1977
* Analysis of data in behavioral areas.
* Final report written.
\F2\CEXPERIMENTAL ELECTRONIC NEWSPAPER
\CEstimated Cost Breakdown
7/1/75 - 6/30/76 7/1/76 - 6/30/77
I. Direct Salaries
Professor J. McCarthy -- no remuneration --
Principal Investigator
10% time (2.4 mm)
Professor E. Parker $ 2,830 $ 3,028
Principal Investigator
10% time (2.4 mm)
Professor D. Drew 4,417 4,726
Principal Investigator
1/3 time (summers) (2.0 mm)
Computer Programmers
2 full-time @ $13,000 per year (48.0 mm) 26,000 27,820
1 part-time @ $13,000 per year (12.0 mm) 6,500 6,955
Secretary
1 quarter-time @ $10,000 per year (6.0 mm) 2,500 2,675
Research Assistants
2 half-time @ $325 per month x 9 (18.0 mm) 5,850 6,260
1 half-time @ $325 per month x 12 (12.0 mm) 3,900 4,173
------ ------
TOTAL SALARIES 51,997 55,637
II. Staff Benefits
18% through Aug. 31, 1975 (6 mo)
19% through Aug. 31, 1976 (12 mo) 9,619 10,849
20% through Aug. 31, 1977 (6 mo)
III. Expendable Material
(Includes Xerox, postage, telephone)
12 months @ $150 per month 1,800 1,980
IV. Report Production - 0 - 1,500
7/1/75 - 6/30/76 7/1/76 - 6/30/77
V. Travel
3 RT East Coast @ 500 each 1,500 - 0 -
VI. Services
Computer services
12 months @ $1,000 per month 12,000 12,000
Newswire services
12 months @ $100 per month 1,200 1,200
------ ------
TOTAL DIRECT COSTS (I through VI) 78,116 83,166
VII. University Overhead
47% of Total Direct Costs 36,714 39,088
VIII. Capital Equipment
Terminal Multiplexer 20,000 - 0 -
Computer Interconnection 4,000 - 0 -
------ -------
------ -------
TOTAL PROJECT COST $138,831 $122,254
Notes:
Salary escalation @ 7% per annum
S & E escalation @ 10% per annum