COMMENT ⊗   VALID 00002 PAGES
C REC  PAGE   DESCRIPTION
C00001 00001
C00002 00002	This is a  reaction to the  Newell plan for the  distributed national
C00011 ENDMK
C⊗;
This is a  reaction to the  Newell plan for the  distributed national
laboratory and to Newell's message of 19 December.

	With  regard  to  the latter,  I  would  take very  seriously
remarks by Atkinson to the effect that now is not a  propitious time.
If some  Democrat wins in Nowvember,  there may be a  renewed wave of
scientific administrative activism, and we should be ready to get our
licks in with the next scientific adviser.  However, that's too iffy,
and fundamentally  support of science is not  a very partisan matter.
By the way, does anybody  know whether Senator Proxmire would  attack
NSF for supporting AI or whether he might like  it.  My impression is
that he is a  real Yahoo, and we wouldn't get to first base with him.
However, I would like to know the source of Atkinson's opinion that a
proposal  for NSF  support of  AI would  turn ARPA  off.   At present
writing,  we are still waiting  for our ARPA  renewal, and Russell is
optimistic, but it hasn't been presented to Heilmeyer for signing.

	As to authorship of the plan, the small  group proposed seems
ok, but we  in the universities have to face  the question of whether
and how hard we want  to push for SRI  participation.  My opinion  is
that SRI may  be mentioned in  the proposal, but the  proposal should
not  have an SRI  author unless the  issue is  dealt with explicitly.
Otherwise, someone in NSF might just sit on the proposal, not wishing
to say that one of the participants offers difficulty.

	I don't  like the  idea of  changing the  name simultaneously
with  changing the source of  support.  It makes it  too easy for the
thing to look  like starting  a new thing  which can be  indefinitely
delayed, rather  than continuing an  on-going activity that  will die
without support.  I do favor an eventual change of name to  something
like "cognology"  after the  field's next  major success.   I  prefer
"cognology"   to   "cognitive  science",   because   the   former  is
linguistically similar to existing well established sciences,  and we
should have such a name as soon as we can be considered established -
which  might  be  now.    "Cognitive  science"  suggests a  temporary
interdisciplinary  field.     I  also   don't  like  something   like
"informatology" which  can be and  often is expanded  to cover fields
like business data processing which while intellectually smaller than
AI are financially  and organizationally much  larger.  To  state the
matter generally, I would not like a name that covers activities that
are now or might be  carried out on a  scale as large or larger  than
that of AI.

	I  agree about  avoiding "now  is  the time  for X".   In  my
opinion,  the  committee  organized  research  effort  involving many
laboratories is about  to get a bad  name, because I don't  think the
speech effort will be regarded as sufficiently successful.

	I think  the initial presentation is too  much a proposal for
starting a grandiose new activity rather than reorienting an existing
successful activity.  The former is subject to the question, "Why not
start on a much smaller scale?"

	The   analogy  between   DNLAI  and  the   existing  national
laboratories in physics and meteorology and chemistry should  be made
at the beginning, and the  possibility of it working well distributed
should  be   stated  as  an  advantage  over  the  existing  kind  of
laboratory.   It  should  be made  clear  that it  is  no more  of  a
sub-granting agency than are these laboratories.

	I have some doubts about the personal LISP computer and would
prefer a LISP processor on a shared large memory.  I would like a new
LISP having  finally concluded  that Interlisp  is a Roccoco  monster
full of  unnecessary ornaments that detract  from the clean structure
of the basic LISP idea.

	DNLAI as a consortium of universities seems the best scheme.

	Newell, are you willing to be head of DNLAI?

	I  would soften  the  remark  about  the  basic  AI  research
facilities being no  longer supportable by ARPA.  I  don't want it to
be a self-promoting prophecy.

	It seems  to me that as in physics, most projects would still
apply to the funding agencies for grants for salaries.   Whether they
would get computing  funds that could be used at  DNL or elsewhere at
the investigator's discretion or whether they would get grants of DNL
resources is an issue that would arise.   I favor the former, because
I think the centers would get these dollars and should have to remain
competitive.  The centers should  be able to give away small  amounts
of computing resources without specific approval.

	The rest of the plan seems ok to me, though I imagine I might
have  some suggestion for  a different organization  of the proposal,
but maybe they aren't worth pursuing.

	I want to express my gratitude to Allen for  the work put in,
and I  hope he will  explore the matter  at least another  round with
Atkinson.  I  am also willing  to put in  a day at  a meeting on  the
subject when Allen considers it appropriate.

	For the information of those other than Allen, I made about 6
telephone  calls to  people not connected  with the  existing centers
outlining the proposed  DNL and asking whether  they would favor  the
idea  and would  want to  participate.   They all  said yes  and much
preferred  it  to trying  to  get  their local  computing  centers to
provide suitable facilities.

	After looking this  over, it doesn't  seem like it  addresses
all the  important issues, but I  think I'll fire it  off rather than
delay another month.