COMMENT ⊗   VALID 00002 PAGES
C REC  PAGE   DESCRIPTION
C00001 00001
C00002 00002	\\M0BASL30\M1BASI30\M2NGR40\M3NGR25\M4NGR20\MFSTA200\
C00012 ENDMK
C⊗;
\\M0BASL30;\M1BASI30;\M2NGR40;\M3NGR25;\M4NGR20;\MFSTA200;\;
\'3;↓↓\FFS\F2
\'3;↓Q\CSTANFORD UNIVERSITY
\F3\CSTANFORD, CALIFORNIA 94305
\F4



ARTIFICIAL INTELLIGENCE LABORATORY\←L\-R\/'7;\+R\→.\→S   Telephone:
COMPUTER SCIENCE DEPARTMENT\←S\→.415-497-4430
\F0\C5 November 1974




Dear Colleagues:

\J	At 3IJCAI I was asked to prepare a plan for converting
the committee into an international AI organization for consideration
at the next meeting which will be held in Tbilisi.  There was no
solid commitment to formalize an organization, but it seems to
me that the consensus included the following points:

	1. It would be an organization of individuals and not
representatives of societies or governments.

	2. If formed it would take over the IJCAIs and probably
the AI journal assuming the editorial board was agreeable.

	3. It would do what else it could to advance the field, but
there were no strongly expressed proposals of what this might
be.

	It seems to me that the following additional considerations
apply:

	1. An organization of this kind could probably have an
income of about $10,000 per year from membership dues and
leftovers from conferences.  This is not enough for even one
employee, and there is no present reason why it would be desirable
to have a headquarters and an employee.

	2. International expenditures on AI may be about $10,000,000
per year depending on definitions, so the AI Society's money should
not be spent on research.  It might be spent on prizes or fellowships,
but there has not recently been any single piece of research that
merits a large prize, and we shouldn't do it just to get publicity,
because publicity \F1per se\F0 is not of obvious value.  I would
favor giving a prize on an \F1ad hoc\F0 basis if something sufficiently
admirable came up before a meeting and there was no conflict of
interest situation.  Therefore, I think the governing committee should
have advance authorization to award a prize but no commitment to
do so and no stipulation as to the amount.

	3. An obviously good function of the organization would be
to maintain a mailing list.  This could be used both as a reference
for finding individuals and by organizations with something to sell.
(I don't think we will be flooded with insurance brochures, but there
might be useful book ads.)

	4. A journal publishing submitted abstracts of reports
and publications might be maintained preferably in co-operation
with an institution with a computer suitable for maintaining
the necessary files and allowing on-line access.  It might also
take orders for reports, although for U.S. users of U.S. reports,
the National Technical Information Service does the job well
enough.

	5. The organization might publish a bulletin, but it
seems more reasonable to encourage international membership
in SIGART at present.

	6. If an organization is formed, it should meet rarely,
and no-one should be trapped into making it his main professional
activity.  Still worse would be having several people whose
main activity was jockeying for power in it.

	7. At present the IJCAIs are being handled by ad hoc
committees formed by volunteers present at the previous
IJCAIs.  Co-ordination is achieved by having a general chairman
of the conference with full authority to set all arrangements
and even to change the location of the next meeting.  As long
as this system works at all, it does so with a minimum of total
effort.  Up to now, the arrangements for the conference have
been adequate and the papers have been reasonably well chosen.
Under these conditions, what the participants get from the
conference is limited by the state of AI research which
determines how good the papers will be rather than by the
conference arrangements.

	There are two conditions under which this scheme should
be replaced by a more formal one.  First, at some time it may
become difficult to get a suitable general chairman and committee.
An organization might help especially if it had a staff - but
we are unlikely to be able to afford a staff.  Second, a dispute
about something might arise which challenges the legitimacy of
the ad hoc meeting that determines the next conference.  Someone
might be tempted to try to pack the meeting, etc.  In that case,
it will become necessary to have an executive committee deriving
its authority from a vote of the members.

	8. The problem of legitimacy of the control of the surplus
from IJCAIs has arisen.  I propose that it be solved as follows:
(i) the trustees constitute themselves into an organization called
finance committee for IJCAIs. (ii) The ad hoc organizing group
meeting at each IJCAI formally give the money from that conference
to the finance committee.  (iii) Changes in the composition of the
finance committee are made with the concurrence of the old committee
and the ad hoc committee. (iv) If these bodies are unable to agree,
then a new three member finance committee shall be elected by a
mail ballot of the attendees at the last two IJCAIs counting the
one at which the dispute occurs.  Nominations are to be submitted
by the old finance committee and by the ad hoc committee.

	In this way we can avoid fuss unless we can't agree.

	In view of the above considerations, I propose postponing
the formation of an international AI organization for yet one
more conference.\.

\←L\→S\←R\-L\/'2;\+L\→L
Sincerely yours,





John McCarthy
Director, Artificial Intelligence Laboratory
Computer Science Department
\←S\→L
JMC:pw
aiorg.pln[cur,jmc]:SU-AI