COMMENT ⊗   VALID 00004 PAGES
C REC  PAGE   DESCRIPTION
C00001 00001
C00002 00002	centre[w84,jmc]		3 notes on Centre Mondial library project
C00003 00003	From: John McCarthy
C00012 00004	These notes (made in September) are partly obsolete, but I
C00021 ENDMK
C⊗;
centre[w84,jmc]		3 notes on Centre Mondial library project

Comments on "Un projet pilote de bibliotheque electronique" by M. Griffiths


Task				estimated effort

list of texts			2 man weeks
definition of internal form	1 man month, maybe 2
initial specs of man machine
interaction			3 man month

revised spec after criticism	2 man month

preliminary reader's manual	2 man month

reading program			4 man month
Notes on the Electronic Library Project - 1984 January

From: John McCarthy

1. The meeting of January 4 including Michael Griffiths, John McCarthy,
Todd Simonds, Pierre Gravier and Francoise Derray recorded a substantial
meeting of minds which we believe shared by Raj Reddy and Jean-Jacques
Servan-Schreiber about the way to proceed with the electronic library
project.  The plan will be summarized in the minutes of the meeting.

2. The project will be located at LISH in Marseilles under the direction
of Michael Griffiths.  The Centre Mondial will provide two posts and a
VAX of still unspecified configuration as well as other expenses not
yet specified.  CNRS will provide the space.

3. The general plan is that of Michael Griffiths's memorandum:
"Un Projet Pilote de Bibliotheque Electronique".

The remaining points are supplementary to that memorandum.

4. The third world connection of the project should not be
forgotten while the pilot project is carried out in Marseilles.
Third world use of the library could begin when the following
conditions are met:

	a. The system works reliably in France.

	b. Its attractiveness and usefulness has been
demonstrated.

	c. The number of books and other documents has
reached an interesting level.  This may or may not require
prior solution of the problem of including and displaying
diagrams, mathematical formulas and tables.

	d. The costs of third world experiments are
determined and a budget provided.

We presently expect that the third world experiment would
involve a satellite connection and clusters of terminals.


5. I recommend postponing relying on information compression
techniques.  They require development work, and if the sole
storage format is a compressed one, other parts of the project
will be delayed until this format is decided and debugged.
The cost of storage is dominated by the cost of document
acquisition, so compression won't save much money.  Where
compression wins is in saving transmission cost, but this depends on
terminals with the computing capability and storage capability
required for decompression.  We will have no difficulty if
we begin with an uncompressed format and plan for a
supplementary compressed format.  We may keep both compressed
and uncompressed copies of a document in storage and use
one or the other according to the capabilities of the terminal
a particular reader is using.  When there is substantial long
distance use, compression will become important.

6. The most serious worry about whether the project will succeed
concerns whether people will find that the advantages of immediate
access to this library will overcome the disadvantages of reading
through a terminal and the limited content of the initial library.
Therefore, I suggest that we plan rather early to provide some
additional services.  Besides making the library more attractive
to users (especially students), they will make demonstrations
to government officials more interesting.  I have in mind the following.

	a. String search, e.g. for first mention of characters and
for quotations.

	b. The ability to make excerpts.  They could be printed at
the library and mailed to the reader requesting it.

	c. A literature course using the library would make use
of string search for quotations and also automatic jumping to
referenced material.

	d. The ability to automatically look up words in the dictionary.
This would help both students and people in the third world whose native
language isn't French.  It will make French literature more attractive
to all foreigners.

7. I have left an annotated copy of Griffiths's memorandum with
reactions to some of its details.

8. We should be able to take a terminal to the office of a government
official and demonstrate reading and some auxiliary services within
a year of approval of the budget.  Of course, this depends on prompt
acquisition of the hardware and when the two people can start work
in Marseilles.My continued role in the electronic library project.

1. I think the project could well survive if I dropped out
now, because Michael Griffiths knows what to do, and the
arrangements for his managing the project have been substantially
negotiated.  However, I think I can continue to be helpful for
at least until September 1984 as a consultant.  Perhaps my
continued association will also help maintain continuity in
the eyes of supporters of the project goals outside the
Centre Mondial.

2. I propose to come to Paris, and most likely Marseilles also,
three or four more times in 1984.  I would plan to make suggestions
and embody them in a memorandum like the one associated with this
trip.

3. I suggest making my next visit around the last week of March
assuming that the budget has been approved by then.  It would be
best if there could then be a meeting including the leaders of the
project and the top management of the Centre Mondial.

4. The technical matters with which I would like to concern myself
include language provided to the user for interacting with the
system as well as the related documentation.  It would be well
to have drafts of these as early as possible.Notes on the electronic library project.
These notes (made in September) are partly obsolete, but I
include them, because some of the considerations still apply.

	1. The most straightforward collection of books would be
the 19th century French novels of Victor Hugo, Anatole France,
Jules Verne, Stendhal, etc.  They are out of copyright, and once
the process of selection and transcription has started, it can be
continued simply by continuing to pump in money.  I will bet that
many computer users who could actually read a book more conveniently
by getting it from the library will read it on-line instead.  For
others on-line availability will really win.

	2. The Centre Mondial seems to me to have very limited resources
in money, institutional stability and personnel.  Its advantage
is its willingness to plunge ahead.  Therefore, it is important to
design the Electronic Library Project to proceed in the most straightforward
possible way avoiding complications.  This means

		a. Avoid complicated negotiations - with publishers,
for institutional collaboration, etc.

		b. Avoid engineering complication and innovation.
Refrain from designing an ideal reading terminal.  Refrain at first
from including pictures.  Settle for standard telephone communication.
Do not get in line for Teletext or cable TV or digital telephone
service.  Offer ordinary telephone lines at speeds commonly used
in France.  Buy all equipment from a single supplier unless there
is an extremely strong advantage in doing otherwise.

		c. Books should be transcribed by companies taking
fixed price contracts for this kind of work.  Every effort should
be made to get low prices by adapting our requirements to the
current capabilities of the suppliers.

	3. In other words, "The realization of the electronic
library must not be allowed to depend on anything not yet
invented.  This includes both technology and social arrangements".
No-one's would-be inventions should be allowed to hitchhike on
the project.

	4. The issue of storing information as images rather than
as characters has been raised.  The advantage is that diagrams and
pictures can be stored and that photography is easier than transcription.
Nevertheless, we plan to use character strings for the following reasons:

	a. Characters are much cheaper to transmit.  Ordinary telephone
lines are adequate for reading if the information is stored as characters.
(Information compression techniques that might be proposed are equally
applicable to characters and images).

	b. Present terminals available throughout France work with
characters, so any presently realizable project must work this way.

	c. Most potential user services require computational access
to the text itself.  These include string search, indexing, automatic
transfer to referenced material and the ability to transfer excerpts to
user created documents either on the library computer or (more likely) on
the user's own computer.


TASKS FOR THE ELECTRONIC LIBRARY PROJECT


1. Find a resident manager.  Ideally he or she should be a French resident,
with experience in managing computer projects, the ability to formulate
requests for proposals and negotiate contracts, the ability to program,
familiarity with configuring computer and communication facilities and
familiarity with previous work in the use of computers in libraries and
the existing databases.  We aren't likely to get someone combining all
these virtues, so compromise is necessary.

It may be easier to get someone on a temporary basis from a computer
company or French government organization than to persuade such a
person to take a regular job with the Centre.  In this respect, the
organization of the Japanese Fifth Generation Project is of interest.

The remaining tasks listed in this memo really ought to be developed
in collaboration with the manager, but since we don't have one, here
goes.


2. Determine the configuration of a suitable VAX and negotiate with
D.E.C. for its purchase.  Pierre Gravier can help with this; indeed
maybe he can do it.  The main requirement is that it have a few
gigabytes of disk (RA81 looks good) and that it have the requisite
20 or so telephone lines.

3. Find out what is included in the Tresor de la Langue Francaise
and negotiate to include some of it in the library.  Michael Griffiths
has agreed to help with this.

4. Formulate and send RFPs (request for proposal) for reading
documents.  Raj Reddy's Indian contact should certainly get the
RFP.  Their 2000 francs per book projected price looks good.
Choose a supplier and begin having books transcribed.

5. Write a program for reading.  It should have the following features:

	a. Similar commands to EMACS but without requiring control
characters for moving around in the text.  There should be emphasis
on moving by pages and chapters.

	b. String search is required, and various improvements such
as search for a disjunction of strings may be useful.

	c. The program should be adjustable to the line length and
number of lines of the major terminals extensively used in France.


6. Finalize the choice of books for the initial system.

-------