COMMENT ⊗   VALID 00002 PAGES
C REC  PAGE   DESCRIPTION
C00001 00001
C00002 00002	Reaction to reaction:
C00006 ENDMK
C⊗;
Reaction to reaction:

	I find myself increasingly serving as an administrator
presiding over getting support for and providing facilities to
a conglomerate of separate projects with no opportunity to
exercise any judgment over what lines of research are most
promising.  This is not what I started the project for, and
if it becomes clear that this is the only option, I will not
continue as P.I.  I don't know what Stanford would do in that
case - find another or drop it.

	1. Even if we are not forbidden to reallocate money, this
will already be quite difficult given the expectations that the
group leaders will develop if there is a formal allocation in the
final version of the proposal.

	2. I believe that formal reasoning is one of the fundamental
scientific problems of AI, and it is only being pursued at Stanford.
If I cannot maintain a viable effort in this area, I will have
to give up ARPA support rather than the research area and confine
myself to paper studies.

	I realize that in the present mood of ARPA Headquarters,
this might suit them quite well, but I think that it would be wrong
to continue the project without a basic research core.  Therefore, I
hope that you will try to work something out that we can live with.

	If there is doubt about whether formal reasoning is scientifically
important, I suggest that you get some consultants.  I am even
willing to defend it to the Jasons.  If there is doubt that the
applied work in AI supported by DoD needs a basic research core,
I suggest this very question be put to some such group.  In fact,
they may have already dealt with it.

	I realize that you have probably taken a pretty hard buffeting
on these matters, but keeping a basic research component alive is
very important, and I hope you will succeed.
Can we talk about it on the phone.  I have to be at Project MAC on
Thursday, and I am leaving for the AI conference on Saturday.

	Let me personally apologize for making matters more tense just
before you go back to M.I.T.

					John