perm filename OUTGO.MSG[1,JMC]21 blob
sn#847137 filedate 1987-10-16 generic text, type C, neo UTF8
COMMENT ⊗ VALID 00031 PAGES
C REC PAGE DESCRIPTION
C00001 00001
C00004 00002 ∂06-Oct-87 1227 JMC
C00093 00003 ∂01-Oct-87 1404 JMC re: Committees
C00094 00004 ∂01-Oct-87 1410 JMC recursivity
C00095 00005 ∂01-Oct-87 1418 JMC visiting professors
C00096 00006 ∂01-Oct-87 1433 JMC Message of 1-Oct-87 14:23:53
C00098 00007 ∂01-Oct-87 1438 JMC INS
C00099 00008 ∂01-Oct-87 1455 JMC re: Reminder
C00100 00009 ∂01-Oct-87 1459 JMC re: Various Topics
C00101 00010 ∂06-Oct-87 1444 JMC re: aids
C00105 00011 ∂06-Oct-87 2230 JMC re: Talk at Stanford
C00107 00012 ∂06-Oct-87 2338 JMC re: Talk at Stanford
C00108 00013 ∂06-Oct-87 2352 JMC AIDS
C00110 00014 ∂07-Oct-87 1005 JMC book the Green Library should get
C00111 00015 ∂07-Oct-87 1014 JMC re: gorbachev and circumscription
C00112 00016 ∂07-Oct-87 1318 JMC re: DAI Workshop Funding Request
C00113 00017 ∂07-Oct-87 1430 JMC knowledge
C00114 00018 ∂07-Oct-87 1442 JMC same message, better address
C00115 00019 ∂07-Oct-87 1550 JMC wrong address
C00117 00020 ∂07-Oct-87 1556 JMC re: more on gorbachev
C00118 00021 ∂07-Oct-87 1602 JMC receipt of paper
C00119 00022 ∂07-Oct-87 2232 JMC supreme court
C00122 00023 ∂08-Oct-87 1625 JMC re: supreme court
C00124 00024 ∂08-Oct-87 1632 JMC re: NSF Centers
C00125 00025 ∂08-Oct-87 2044 JMC
C00126 00026 ∂08-Oct-87 2117 JMC report on ebos
C00127 00027 ∂09-Oct-87 1106 JMC reference
C00128 00028 ∂09-Oct-87 1110 JMC re: Sept Auditron Readings
C00129 00029 ∂09-Oct-87 1203 JMC please send paper
C00130 00030 ∂09-Oct-87 1225 JMC re:aids
C00134 00031 ∂09-Oct-87 1236 JMC re: One More About Your Lecture
C00136 ENDMK
C⊗;
∂06-Oct-87 1227 JMC
To: alyson%amcad.uucp@HUSC6.HARVARD.EDU
\input memo.tex[let,jmc]
\title{MATHEMATICAL LOGIC IN ARTIFICIAL INTELLIGENCE}
This article concerns computer programs that represent information
about their problem domains in mathematical logical languages and use
logical inference to decide what actions are appropriate to achieve their
goals.
Mathematical logic isn't a single language. There are many
kinds of mathematical logic, and even choosing a kind doesn't specify
the language. This is done by declaring what {\it non-logical} symbols
will be used and what sentences will be taken as axioms. The non-logical
symbols are those that concern the concrete subject matter, e.g. objects
and their locations and motions.
All kinds of mathematical logic share two ideas. First it must
be mathematically definite what strings of symbols are considered
formulas of the logic. Second it must be mathematically definite what
inferences of new formulas from old ones are allowed. This permits
writing computer programs that decide what combinations of symbols are
sentences and what inferences are allowed in a particular logical
language.
Mathematical logic has become an important branch of
mathematics, and most logicians work on problems arising from the internal
development of the subject. It has also been applied to studying the
foundations of mathematics, and this has been its greatest success. Its
founders, Aristotle, Leibniz, Boole and Frege also wished to apply it to
making reasoning about human affairs more rigorous. Indeed Leibniz was
explicit about his goal of replacing argument by calculation. However,
expressing knowledge and reasoning about the common sense world in
mathematical logic has encountered difficulties that seem to require
extensions of the basic concepts of logic, and these extensions are only
beginning to develop.
If a computer is to store facts about the world and reason with
them, it needs a precise language, and the program has to embody a
precise idea of what reasoning is allowed, i.e. of how new formulas may be
derived from old. Therefore, it was natural to try to use mathematical
logical languages to express what an intelligent computer program knows
that is relevant to the problems we want it to solve and to make the
program use logical inference in order to decide what to do. (McCarthy
1960) contains the first proposals to use logic in AI for expressing
what a program knows and and how it should reason. (Proving logical
formulas as a domain for AI had already been studied).
The 1960 paper said:
\begingroup\narrower\narrower
% COMMON.TEX[E80,JMC] TeX version Programs with Common Sense
The {\it advice taker} is a proposed program for solving problems by
manipulating sentences in formal languages. The main difference between
it and other programs or proposed programs for manipulating formal
languages (the {\it Logic Theory Machine} of Newell, Simon and Shaw and
the Geometry Program of Gelernter) is that in the previous programs the
formal system was the subject matter but the heuristics were all embodied
in the program. In this program the procedures will be described as much
as possible in the language itself and, in particular, the heuristics are
all so described.
The main advantages we expect the {\it advice taker} to have is that
its behavior will be improvable merely by making statements to it,
telling it about its symbolic environment and what is wanted from it. To
make these statements will require little if any knowledge of the program
or the previous knowledge of the {\it advice taker}. One will be able to
assume that the {\it advice taker} will have available to it a fairly wide
class of immediate logical consequence of anything it is told and its
previous knowledge. This property is expected to have much in common with
what makes us describe certain humans as having {\it common sense}. We
shall therefore say that {\it a program has common sense if it automatically
deduces for itself a sufficiently wide class of immediate consequences of
anything it is told and what it already knows.}
\par\endgroup
The main reasons for using logical sentences extensively in AI
are better understood by researchers today than in 1960. Expressing
information in declarative sentences is far more flexible than
expressing it in segments of computer program or in tables. Sentences
can be true in much wider contexts than specific programs can be useful.
The supplier of a fact does not have to understand much about how the
receiver functions or how or whether the receiver will use it. The same
fact can be used for many purposes, because the logical consequences of
collections of facts can be available.
The {\it advice taker} prospectus was ambitious in 1960, would
be considered ambitious today and is still far from being immediately
realizable. This is especially true of the goal of expressing the the
heuristics guiding the search for a way to achieve the goal in the
language itself. The rest of this paper is largely concerned with
describing what progress has been made, what the obstacles are, and how
the prospectus has been modified in the light of what has been
discovered.
The formalisms of logic can be used to differing extents, mostly
much less ambitious, and we'll begin by recounting some of them.
1. A machine may use no logical sentences --- all its ``beliefs''
being implicit in its state. Nevertheless, it is often appropriate to
ascribe beliefs and goals to the program, i.e. to remove the above
sanitary quotes, and to use a principle of rationality --- {\it It does
what it thinks will achieve its goals}. Such ascription is discussed in
(Dennett 1971), (McCarthy 1979) and (Newell 1980). The advantage is that
the intent of the machine's designers and the way it can be expected to
behave may be more readily described {\it intentionally}
than by a purely physical
description.
The relation between the physical and the {\it intentional}
descriptions is most readily understood in simple systems that admit
readily understood descriptions of both kinds, e.g. thermostats. Some
finicky philosophers object to this, contending that unless a system has a
full human mind, it shouldn't be regarded as having any mental qualities
at all. This is like omitting the numbers 0 and 1 from the number system
on the grounds that numbers aren't required to count sets with no elements
or one element. Much more that can be said about ascribing
mental qualities to machines, but that's not where the main action is in
AI.
2. The next level of use of logic involves computer programs that
use sentences in machine memory to represent their beliefs but use
other rules than ordinary logical inference to reach conclusions.
New sentences are often obtained from the old ones
by ad hoc programs. Moreover, the sentences that appear in memory are
from a program-dependent subset of the logical language being used.
Adding certain true sentences in the language may even spoil the functioning of
the program. The languages used are often rather unexpressive compared to
first order logic, for example they may not admit quantified sentences,
and they may represent ``rules'', e.g. the equivalent of universally
quantified sentences in a separate notation. Often rules cannot be
consequences of the program's action; they must have all been put in by the
``knowledge engineer''. Sometimes the reason programs have this form is
just ignorance, but the usual reason for the restriction is the practical
one of making the program run fast and deduce just the kinds of
conclusions its designer anticipates. Most often the rules are
implications used in just one direction, e.g. the contrapositive of an
implication is not used. We believe the need for such specialized inference
will turn out to be temporary and will be reduced or eliminated by
improved ways of controlling general inference, e.g. by allowing the
heuristic rules to be also expressed as sentences as advocated in
the above extract from the 1960 paper.
3. The third level uses first order logic and also logical
deduction. Typically the sentences are represented as clauses, and the
deduction methods are based on J. Allen Robinson's (1965) method of
resolution. It is common to use a theorem prover as a problem solver,
i.e. to determine an $x$ such that $P(x)$ as a byproduct of a proof of
the formula $∃x.P(x)$ that represents the proposition that there is an
$x$ satisfying the predicate $P$. This level is less used for practical
purposes than level two, because techniques for controlling the
reasoning are still insufficiently developed, and it is common for the
program to generate many useless conclusions before reaching the desired
solution. Indeed unsuccessful experience (Green 1969) with this method
led to more restricted uses of logic, e.g. the STRIPS system of (Nilsson
and Fikes 1971).
The commercial ``expert system shells'', e.g. ART, KEE and
OPS-5, use logical representation of facts, usually ground facts only,
and separate facts from rules. They provide elaborate but not always
adequate ways of controlling inference.
In this connection it is important to mention logic programming,
first introduced in Microplanner (Sussman et al., 1971)
and from different points of view by Robert Kowalski (1979) and Alain
Colmerauer in the early 1970s.
A recent text is (Sterling and Shapiro 1986). Microplanner
was a rather unsystematic collection of tools, whereas Prolog relies
almost entirely on one kind of logic programming, but the main idea
is the same. If one uses a restricted class of sentences, the so-called
Horn clauses, then it is possible to use a restricted form of logical
deduction, and the control problem are much eased, and it is possible
for the programmer to anticipate the course the deduction will take.
The price paid is that only certain kinds of facts are conveniently
expressed as Horn clauses, and the depth first search built into
Prolog is not always appropriate for the problem. Nevertheless,
expressibility in Horn clauses is an important property of a set
of facts and logic programming has been successfully used for many
applications, although it seems unlikely to dominate AI programming
as some of its advocates hope.
Although third level systems express both facts and rules
as logical sentences, they are still rather specialized. The axioms
with which the programs begin are not general truths about the world
but are sentences whose meaning and truth is limited to the narrow
domain in which the program has to act. For this reason, the ``facts''
of one program usually cannot be used in a database for other programs.
4. The fourth level is still a goal. It involves representing
general facts about the world as logical sentences. Once put in
a database, the facts can be used by any program. The facts would
have the neutrality of purpose characteristic of much human information.
The supplier of information would not have to understand
the goals of the potential user or how his mind works. The present
ways of ``teaching'' computer programs amount to ``education
by brain surgery''.
A major difficulty is that fourth level systems require extensions
to mathematical logic. One kind of extension is non-monotonic
reasoning, first proposed in the late 1970s (McCarthy 1977, 1980, 1986),
(Reiter 1980), (McDermott and Doyle 1980). Traditional logic is monotonic
in the following sense. If a sentence $p$ is inferred from a collection
$A$ of sentences, and $B$ is a more inclusive set of sentences (symbolically
$A ⊂ B$), then $p$ can be inferred from $B$.
If the inference is logical deduction, then exactly the same
proof that proves $p$ from $A$ will serve as a proof from $B$. If the
inference is model-theoretic, i.e. $p$ is true in all models of $A$,
then $p$ will be true in all models of $B$, because the models of $B$
will be a subset of the models of $A$. So we see that the monotonic
character of traditional logic doesn't depend on the details of the
logical system but is quite fundamental.
While much human reasoning corresponds to that of traditional
logic, some important human common sense reasoning is not monotonic. We
reach conclusions from certain premisses that we would not reach if
certain other sentences were included in our premisses. For example,
learning that I own a car, you conclude that it is appropriate on a
certain occasion to ask me for a ride, but when you learn the further
fact that the car is in the garage being fixed you no longer draw that
conclusion. Some people think it is possible to try to save
monotonicity by saying that what was in your mind was not a general rule
about asking for a ride from car owners but a probabilistic rule. So
far it has not proved possible to try to work out the detailed
epistemology of this approach, i.e. exactly what probabilistic
sentences should be used. Instead AI has moved to directly formalizing
non-monotonic logical reasoning.
Formalized non-monotonic reasoning is under rapid development
and many kinds of systems have been proposed. I shall concentrate on
an approach called circumscription, because I know it, and because it
has met with wide acceptance and is perhaps the most actively pursued
at present. The idea is to single out among the models of the collection
of sentences being assumed some ``preferred'' or ``standard'' models.
The preferred models are those that satisfy a certain minimum principle.
What is to be minimized is not yet decided in complete generality,
but many domains that have been studied yield quite general theories
using minimizations of abnormality or of the set of some kind of entity.
The idea is not completely unfamiliar. For example, Ockham's razor ``Do
not multiply entities beyond necessity'' is such a minimum principle.
Minimization in logic is another example of an area of mathematics
being discovered in connection with applications rather than via
the normal internal development of mathematics. Of course, the reverse
is happening on an even larger scale; many logical concepts developed
for purely mathematical reasons turn out to have AI importance.
As a more concrete example of non-monotonic reasoning, consider
the conditions under which a boat may be used to cross a river. We all
know of certain things that might be wrong with a boat, e.g. a leak, no
oars or motor or sails depending on what kind of a boat it is. It would
be reasonably convenient to list some of them in a set of axioms.
However, besides those that we can expect to list in advance, human
reasoning will admit still others, should they arise, but cannot be
expected to think of them in advance, e.g. a fence down the middle of
the river. This is handled using circumscription by minimizing the set
of ``things that prevent the boat from crossing the river'', i.e. the
set of obstacles to be overcome. If the reasoner knows of none in a
particular case, he or it will conjecture that the boat can be used, but
if he learns of one, he will get a different result when he minimizes.
This illustrates the fact that non-monotonic reasoning
is conjectural rather than rigorous. Indeed it has been shown
that certain mathematical logical systems cannot be rigorously extended,
i.e. that they have a certain kind of completeness.
It is as misleading to conduct a discussion of this kind
entirely without formulas as it would be to discuss the foundations of
physics without formulas. Unfortunately, many people are unaware of this
fact. Therefore, we present a formalization by Vladimir Lifschitz
(1987) of a simple example called ``The Yale shooting problem''. Drew
McDermott (1987), who has become discouraged about the use of logic in
AI and especially about the non-monotonic formalisms, invented it as a
challenge. (The formal part is only one page, however). Some earlier
styles of axiomatizing facts about change don't work right on this
problem. Lifschitz's method works well here, but I think it
will require further modification.
In an initial situation there is an unloaded gun and a person
Fred. The gun is loaded, then there is a wait, and then the gun
is pointed at Fred and fired. The desired conclusion is the death
of Fred. Informally, the rules are (1) that a living person remains alive until
something happens to him, (2) that loading causes a gun to become loaded,
(3) that a loaded gun remains loaded until something unloads it, (4) that
shooting unloads a gun and (5) that shooting a loaded gun at a person
kills him. We are intended to reason as follows. Fred will remain alive
until the gun is fired, because nothing can be inferred to happen to
him. The gun will remain loaded until it is fired, because nothing
can be inferred to happen to it. Fred will then die when the gun is
fired. The non-monotonic part of the reasoning is minimizing ``the
things that happen'' or assuming that ``nothing happens without a reason''.
The logical sentences are intended to express the above 5 premisses,
but they don't explicitly say that no other phenomenon occurs. For example,
it isn't asserted that Fred isn't wearing a bullet proof vest, nor are
any properties of bullet proof vests mentioned. Nevertheless, a human
will conclude that unless some unmentioned aspect of the situation
is present, Fred will die. The difficulty is that the sentences admit
an {\it unintended minimal model}, to use the terminology of mathematical logic.
Namely, it might happen that for some unspecified reason the gun becomes
unloaded during the wait, so that Fred remains alive. The way non-monotonic
formalisms, e.g. circumscription and Reiter's logic of defaults,
were previously used to formulate
the problem, minimizing ``abnormality'' results in two possibilities,
not one. The unintended possibility is that the gun mysteriously
becomes unloaded.
By popular demand Lifschitz's actual axioms are relegated to the Appendix.
It seems likely that introducing non-monotonic reasoning
will not be the only modification to logic that will be required in order
to give machines human capability for common sense reasoning.
In order to make programs that reason about their own
knowledge and belief, i.e. have even rudimentary consciousness,
it is necessary to formalize many {\it intensional} notions, e.g.
knowledge and belief. (McCarthy 1979a) formalizes some of them
in first order logic by introducing propositions and individual
concepts as individuals. Complicating such efforts are the
paradoxes discovered by Montague (1963). It will be necessary
to weaken the axioms suitably to avoid them, but a good way of
doing it hasn't yet been found.
It seems also to be necessary to formalize the notion of context,
but this is in a very preliminary state of investigation.
!\medskip
\noindent{\bf AI and Philosophy}
Artificial intelligence cannot avoid philosophy. If a computer
program is to behave intelligently in the real world, then it must be
provided with some kind of framework into which to fit particular facts
it is told or discovers. This amounts to at least a fragment of some
kind of philosophy, however naive. Here we are agreeing with
philosophers advocating the study of philosophy who claim that someone
who purports to ignore philosophy is merely condemning himself to a
naive philosophy.
AI could probably make do with a naive philosophy for a long
time, since AI is still far behind the intellectual performance of
people who are philosophically naive. Unfortunately, it hasn't been
possible to say what a naive philosophy is, and the philosophers offer
little guidance.
The next plausible alternative might be to take one of the
philosophies that have been proposed by philosophers and build our
programs to seek knowledge and represent it in accordance with its
tenets. This also hasn't been possible. Either no-one in AI,
including retreaded philosophers, understands philosophical theories
well enough to program a computer in accordance with their tenets,
or the philosophers haven't even come close to the required precision.
Actually it seems that some of the empiricist philosophies may be
precise enough but turn out to be inadequate when one attempts to use
them in the most modest computer programs.
Therefore, we AI researchers have found ourselves on our
own when it comes to providing the program with a basic intellectual
structure. Here is some of what we think we require.
1. Ontology. Taking Quine's idea that our ontology is defined
by the range of the bound variables, we need to specify what kinds of
entities are to be assumed, i.e. what are the robot's beliefs to be
about. Nominalism suggests that variables take material objects as
values. This promptly proves inadequate; for example, because it
doesn't permit the robot's designer to inform it about
what properties of objects are preserved when certain kinds of events
take place.
Quine (1987) tells us that ``But there is no place in science
for ideas'', arguing for this view with examples of the difficulty
in defining what it means for two people to have the same idea.
However, if a program is to search for a good idea by generating
lots of them and testing, then it needs some criteria for deciding
when it has already tested a certain idea. Thus ideas as objects
seem to be required, but how to avoid the difficulties Quine cites
hasn't yet been discovered. Present AI systems can't enumerate
ideas.
2. Free will. The robots we plan to build are entirely
deterministic systems. However, a sophisticated robot must decide
what to do by considering the various things it {\it can} do and
choosing which has the best consequences in view of the goals
it has been given. In order to do this, it must be able to represent
``I can do A and I can do B, but B seems better, so while I can
do A, I won't''. What does it mean for a robot to believe, ``I
can, but I won't''? It's a determinist system, so either it will
do A or it won't. (McCarthy and Hayes 1969) contains some proposals
for resolving the problem of free will for robots.
3. Non-monotonic reasoning. AI programs require ways of
jumping to conclusions on the basis of insufficient evidence.
AI researchers' attempts to determine an intellectual framework
precise enough for programming AI systems have already led to a
certain philosophical views --- both to taking sides in some ancient
philosophical controversies and to proposals that we regard as new.
Here are some points, some of which are controversial within AI.
1. Incrementalism or modesty. The facts about the effects of
actions and other events that have been put into the databases of AI
programs are not very general. They are not even as general as what
questioning would elicit from naive people, let alone general enough to
satisfy people familiar with the philosophical literature. However,
they suffice in limited cases to determine the appropriate action to
achieve a goal. Observing their limitations in practice leads to
further advance.
We argue that this is a useful methodology even when
the objectives are philosophical. One can design formalisms that can be
used in working systems and improve them when their defects become
apparent.
The philosopher might claim that the working systems are
too trivial to be of interest to him. He would be wrong, because
it turns out that the philosophical investigations of action have
missed important phenomena that arise as soon as one tries to
design systems that plan actions. Here are two examples. First the ideas
of association dating at least from Mill and going through the
behaviorists are too vague to be programmed at all. Second,
philosophers have missed most of the non-monotonic character of
the reasoning involved in everyday decision making.
For AI it is not only important that the researcher be
able to revise his ideas. It is also important that the program
be able to improve its behavior incrementally, either by accepting
advice from the user or by learning from experience, and this requires
new languages for expressing knowledge. For example, a baby first
considers the word ``mother'' as a proper name, then as a general
name for adult women, and still later as a designating a relation.
Languages with appropriate {\it elaboration tolerance} can interpret certain
sentences as changing a predicate symbol from taking one argument
to taking two. They don't exist yet.
2. Objectivity. Regardless of one's ultimate view of reality,
in designing robots we need to make the robot view the world as an
external reality about which it has and can obtain only partial
knowledge. We will not be successful if we design the robot to
regard the world as merely a structure built on its sensory information.
There needs to be a theory (it could be called meta-epistemology)
relating the structure of a world, a knowledge-seeker in that world,
the interaction channel between the knowledge-seeker and the rest
of the world, the knowledge-seeker's rules for deciding what assertions
about the world are meaningful and its rules for accepting evidence
about the world and what the knowledge-seeker can discover. If the
rules are too restrictive, perhaps as in some operationalist
philosophies of science, the knowledge-seeker will be unable
to discover basic facts about the world, regarding the assertions
as insufficiently operational to be even meaningful.
!\noindent {\bf Remarks}
Many of these remarks involve stating a position on issues
that are controversial even within AI.
1. Artificial intelligence is best regarded as a branch of
computer science rather than as a branch of psychology. Namely,it
is concerned with methods of achieving goals in situations in which
the information available has a certain complex character. The methods
that have to be used are related to the problem presented by the situation
and are similar whether the problem solver is human, a Martian or a
computer program.
Initially some people were over-optimistic about how long it
would take to achieve human level intelligence. Optimism was natural,
because only a few of the difficulties had been identified. Enough
difficulties have been identified by now to establish AI as one of the
more difficult sciences. Maybe it will take five years to achieve
human level intelligence, and maybe it will take five hundred.
2. It is still not clear how to characterize situations in which
intelligence is required. Evidently they have a certain open-ended
character. Even in a game like chess where the rules are fixed, the
methods for deciding on a move have an open ended character --- new
ways of thinking about chess positions are invented all the time.
3. AI has so far identified certain methods of pattern matching,
heuristic search of trees of possibilities, representation of information
by rules and learning. Other methods are still to be characterized,
especially methods of representing problems as collections of subproblems
that can be examined separately to get certain results that can then
be used in studying their interactions.
4. The logic approach to AI is not the only one that may lead
to success. For example, approaches more closely tied to biology
may succeed eventually, even though most of the biology motivated
approaches that have been tried since the 1950s have dried up.
5. Much of AI's controversial character comes from its
implications for philosophy, a subject about which there are strong
views. AI tends to support rationalist and realistic views of
philosophical problems rather than empiricist, phenomenological or
idealist views. It encourages a piecemeal approach to the philosophy
of mind in which mental qualities are considered separately rather
than as part of a grand package. This is because some systems have
important but rather limited mental qualities.
6. There are many open problems in formalizing common sense and
many approaches to solving them awaiting exploration. 2,000 years of
philosophy has only limited relevance. In my opinion, their proper
discussion is unavoidably mostly technical, involving the actual logical
formalisms being used.
7. The situation calculus used above has important known
limitations. The $result(e,s)$ formalism has to be modified to
handle continuous time, and a quite different formalism is needed
for expressing facts about concurrent events. Kowalski's (1986)
{\it event calculus} is a candidate for meeting both of these
requirements.
8. The study of AI may lead to a mathematical metaepistemology
analogous to metamathematics. Namely, one can study the relation between
a knower's rules for accepting evidence and a world in which he is
embedded. There can then be actual mathematical theorems about whether
certain intellectual strategies will discover certain facts about the
world. I think this will eventually revolutionize philosophy.
9. The best general text on the logic approach to AI is
(Genesereth and Nilsson 1987).
\medskip
!\noindent{\bf References}
\noindent
{\bf Dennett, D.C. (1971)}: ``Intentional Systems'', {\it Journal of Philosophy}
vol. 68, No. 4, Feb. 25.
\noindent
{\bf Fikes, R, and Nils Nilsson, (1971)}:
``STRIPS: A New Approach to the Application of
Theorem Proving to Problem Solving,'' {\it Artificial Intelligence}, Volume 2,
Numbers 3,4, January,
pp. 189-208.
\noindent
{\bf Genesereth, Michael and Nils Nilsson (1987)}: {\it The Logical
Foundations of Artificial Intelligence}, Morgan-Kaufman.
\noindent
{\bf Green, C., (1969)}:
``Application of Theorem Proving to Problem Solving. In IJCAI-1, pp. 219-239.
\noindent
{\bf Kowalski, Robert (1979)}: {\it Logic for Problem Solving},
North-Holland, Amsterdam.
\noindent
{\bf Kowalski, Robert and Marek Sergot (1985)}: {\it A Logic-based Calculus of
Events}, Dept. of Computing, Imperial College, London.
\noindent
{\bf Lifschitz, Vladimir (1987)}:
``Formal Theories of Action'', to be published.
\noindent
{\bf McCarthy, John (1960)}: ``Programs with Common Sense'', in Proceedings of the
Teddington Conference on the Mechanization of Thought Processes, Her Majesty's
Stationery Office, London.
% common[e80,jmc],
% common.tex[e80,jmc]
\noindent
{\bf McCarthy, John and P.J. Hayes (1969)}: ``Some Philosophical Problems from
the Standpoint of Artificial Intelligence'', in D. Michie (ed), {\it Machine
Intelligence 4}, American Elsevier, New York, NY.
% phil[ess,jmc] with slight modifications
\noindent
{\bf McCarthy, John (1977)}:
``Epistemological Problems of Artificial Intelligence'', {\it Proceedings
of the Fifth International Joint Conference on Artificial
Intelligence}, M.I.T., Cambridge, Mass.
% ijcai.c[e77,jmc]
\noindent
{\bf McCarthy, John (1979)}:
``Ascribing Mental Qualities to Machines'' in {\it Philosophical Perspectives
in Artificial Intelligence}, Ringle, Martin (ed.), Harvester Press, July 1979.
% .<<aim 326, MENTAL[F76,JMC],mental.tex[f76,jmc]>>
\noindent
{\bf McCarthy, John (1979a)}:
``First Order Theories of Individual Concepts and Propositions'',
in Michie, Donald (ed.) {\it Machine Intelligence 9}, (University of
Edinburgh Press, Edinburgh).
% .<<aim 325, concep.tex[e76,jmc]>>
\noindent
{\bf McCarthy, John (1980)}:
``Circumscription - A Form of Non-Monotonic Reasoning'', {\it Artificial
Intelligence}, Volume 13, Numbers 1,2, April.
% .<<aim 334, circum.new[s79,jmc], cirnew.tex[s79,jmc]>>
\noindent
{\bf McCarthy, John (1986)}:
``Applications of Circumscription to Formalizing Common Sense Knowledge''
{\it Artificial Intelligence}, April 1986
% circum.tex[f83,jmc]
\noindent
{\bf McDermott, D. and J. Doyle, (1980)}:
``Non-Monotonic Logic I,'' {\it Artificial Intelligence\/},
Vol. 13, N. 1
\noindent
{\bf McDermott, D. (1987)}:
``A Critique of Pure Reason'', {\it Computational Intelligence}, with
peer commentaries, forthcoming, 1987.
\noindent
{\bf Montague, Richard (1963)}: ``Syntactical Treatments of Modality, with
Corollaries on Reflexion Principles and Finite Axiomatizability,''
{\it Acta Philosophica Fennica\/} {\bf 16}:153--167. Reprinted in (Montague 1974).
\noindent
{\bf Montague, Richard (1974)}: {\it Formal Philosophy}, Yale University Press
\noindent
{\bf Newell, Allen (1981)}: ``The Knowledge Level,'' {\it AI Magazine\/},
Vol. 2, No. 2
\noindent
{\bf Quine, W. V. (1987)}: {\it Quiddities}, Harvard University Press.
\noindent
{\bf Reiter, R.A. (1980)}: ``A Logic for default reasoning,'' {\it Artificial
Intelligence\/}, 13 (1,2), 81-132.
\noindent
{\bf Robinson, J. Allen (1965)}: ``A Machine-oriented Logic Based
on the Resolution Principle''. {\it JACM}, 12(1), 23-41.
\noindent
{\bf Sterling, Leon and Ehud Shapiro (1986)}: {\it The Art of Prolog}, MIT Press.
\noindent
{\bf Sussman, Gerald J., Terry Winograd, and
Eugene Charniak (1971)}: ``Micro-planner Reference Manual,'' Report AIM-203A,
Artificial Intelligence Laboratory, Massachusetts Institute of Technology,
Cambridge.
\medskip
!\noindent Appendix: Lifschitz's Cauality Axioms for the Yale Shooting Problem
Lifschitz's axioms use the situation calculus of (McCarthy and
Hayes 1969) but introduce a predicate $causes$ as an undefined notion.
We quote from (Lifschitz 1987).
``Our axioms for the shooting problem can be classified into three groups.
The first group describes the initial situation:
%
$$holds(alive,S0),\eqno(Y1.1)$$
$$¬ holds(loaded,S0).\eqno(Y1.2)$$
%
The second group tells us how the fluents are affected by actions:
%
$$causes(load,loaded,true),\eqno(Y2.1)$$
$$causes(shoot,loaded,false),\eqno(Y2.2)$$
$$causes(shoot,alive,false).\eqno(Y2.3)$$
%
These axioms describe the effects of
{\it successfully performed} actions;
they do not say {\it when} an action can be successful.
This information is supplied separately:
%
$$precond(loaded,shoot).\eqno(Y2.4)$$
%
The last group consists of two axioms of a more general nature. We use
the abbreviations:
%
$$success(a,s) ≡ ∀f(precond(f,a) ⊃ holds(f,s)),$$
$$affects(a,f,s) ≡ success(a,s) ∧ ∃v\;causes(a,f,v).$$
%
One axiom describes how the value of a fluent changes after an action affecting
this fluent is carried out:
%
$$success(a,s) ∧ causes(a,f,v) ⊃ (holds(f,result(a,s)) ≡ v=true).\eqno(Y3.1)$$
%
(Recall that $v$ can take on two values here, $true$ and $false$; the
equivalence in $Y3.1$ reduces to $holds(f,result(a,s))$ in the first case and
to the negation of this formula in the second.)
If the fluent is not affected then its value remains the same:
%
$$¬affects(a,f,s) ⊃ (holds(f,result(a,s)) ≡ holds(f,s)).\eqno(Y3.2)\hbox{''}$$
Minimizing $causes$ and $precond$ makes the right thing happen.
While these axioms and {\it circumscription policy} solve this problem, it
remains to be seen whether we can write a large body of common sense
knowledge in the formalism without getting other unpleasant surprises.
Another current question is whether we can get by with axioms about the
external world only or whether the axioms must contain information about
the purposes of the reasoning in order to determine the preferred models.
Moreover, there are many more possibilities to explore for the formal
minimum principle required for common sense reasoning.
!\smallskip
\centerline{Copyright \copyright\ \number\year\ by John McCarthy}
\smallskip
\noindent This draft of logic.2[w87,jmc] TEXed on \jmcdate\ at \theTime
\vfill\eject\end
%logic.2[w87,jmc] Yet another try
∂06-Oct-87 0847 JMC reply to message
To: PHY
[In reply to message rcvd 06-Oct-87 08:37-PT.]
Please file all the above.
∂05-Oct-87 2354 JMC aids
To: su-etc@SAIL.Stanford.EDU
1. The following article confirms what some of us have
suspected about the attitude of the homosexual activists.
2. Vigorous enforced sexual quarantines at the beginning might
have slowed the spread of AIDS. But remember it was a while before
any tests were available.
3. Compared to Alzheimer's disease, AIDS is very well funded.
The claim that it has not been well funded is an aspect of the
politics of claiming victimization. Considering the exotic
nature of the problem, progress has been rapid. Compared to
progress on Alzheimer's disease, which affects 2.5 million
Americans, progress on AIDS has been extremely rapid.
a009 2248 05 Oct 87
PM-AIDS Book, Bjt,0723
Gay Community Was Slow To React Adequately To AIDS Epidemic
By RICH CARTIERE
Associated Press Writer
SAN FRANCISCO (AP) - A homosexual author charges gay activists
nationwide at first failed to understand the AIDS epidemic as a
medical problem rooted in promiscuity, treating it instead as a
''public relations problem.''
That mistake, coupled with the political expediency and scientific
rivalry that wracked response to AIDS between 1980 and 1985, helped
the disease rage out of control, says Randy Shilts in his book, ''And
the Band Played On: People, Politics and the AIDS Epidemic.''
When some gay community leaders tried to caution against profligate
sex during the epidemic's initial years, they were attacked as
''sexual Nazis'' and ''anti-sex brownshirts,'' says Shilts.
''Self-criticism was not the strong point of a community that was
only beginning to define itself affirmatively after centuries of
repression,'' Shilts writes. His book is scheduled to be released
Oct. 30.
Shilts, who has covered AIDS full-time since 1983 for the San
Francisco Chronicle, says his book is the first attempt to document
''how completely and consistently the Reagan administration turned
its back and ignored its own health officials'' in funding the fight
against the disease.
He says it also is the first time that the scientific community's
initial reluctance to deal with the disease and the subsequent
rivalry between leading researchers has been comprehensively studied.
''In the first two years, some people were warned by superiors that
research into AIDS would hurt their careers,'' Shilts said in an
interview Monday. ''And then the bickering in the following years
between scientists delayed progress.''
He writes that by the time America ''paid attention to the disease,
it was too late to do anything about it.'' The epidemic ''was allowed
to happen by an array of institutions, all of which failed to perform
their appropriate tasks to safeguard the public health.''
Gastrointestinal diseases that swept through homosexual communities
nationwide just before the AIDS epidemic should have been early
warning signs about the health dangers of promiscuity, but Shilts
says they were ignored by gay leaders.
Tempering his criticism, Shilts says, ''The gay community was slow
to react adequately, but what then happened in response to the crisis
happened mainly because of gay people's involvement.''
Shilts says doctors with homosexual patients delayed putting
together risk-reduction guidelines, and were reluctant to join
community panels.
In Vancouver, British Columbia, gay bathhouse owners became angry at
the local homosexual newspaper for ''running a health page; this
obsession with a handful of sick people in the United States was bad
for business,'' according to Shilts.
The book also indentified the man referred to by researchers as
''Patient Zero'' because he may have brought AIDS to North America.
At least 17 percent of the first 248 AIDS cases reported in the
United States by 1982 are linked to Gaetan Dugas, who remained
defiantly sexually active after his diagnosis and died in 1984, says
Shilts.
Of the first 19 AIDS cases reported in Los Angeles, four of the
victims had sex with the airline steward from Montreal, and another
four had sex with one of his sexual partners. New York's first two
cases, in 1979, were Dugas' ex-sexual partners.
From just one sexual encounter with Dugas, researchers were able to
link 11 other cases of acquired immune deficiency syndrome, Shilts
writes.
Shilts charges the Reagan administration ignored pleas from
government scientists and did not allocate adequate money for AIDS
research.
Citing documents turned up using the Freedom of Information Act,
Shilts alleges then-Health and Human Services Secretary Margaret
Heckler misled Congress when she testified on April 12, 1983, that
her department was using ''every dollar necessary to try to find an
answer.''
Shilts quotes from an internal memo written that day by Dr. Don
Francis, chief of AIDS lab research at the federal Centers for
Disease Control, which says: ''The inadequate funding to date has
seriously restricted our work and has presumably deepened the
invasion of this disease into the American population.''
AIDS, which damages the body's immune system, leaving victims
susceptible to infections and cancer, is spread most often through
sexual contact, needles or syringes shared by drug abusers, infected
blood or blood products, and from pregnant women to their offspring.
As of Sept. 28, the CDC had received reports of 42,354 Americans
with AIDS, 24,412 of whom had died.
AP-NY-10-06-87 0134EDT
***************
∂05-Oct-87 1448 JMC afore-mentioned demand for final report
To: LES
I had phoned her and grumbled that they send out these demands
for reports without saying what it is other than the
number. Clearly some of the buck needs to be passed to Jussi.
∂05-Oct-87 1443 AS.JAL@forsythe.stanford.edu NSF Final Report
Received: from LINDY.STANFORD.EDU by SAIL.STANFORD.EDU with TCP; 5 Oct 87 14:43:38 PDT
Received: by lindy.stanford.edu; Mon, 5 Oct 87 14:43:19 PDT
Date: Mon, 5 Oct 87 14:41:57 PDT
From: Judy Leasher <AS.JAL@forsythe.stanford.edu>
To: jmc@sail.stanford.edu
Subject: NSF Final Report
October 5, 1987
TO: Professor John McCarthy
Computer Science
FROM: Judith Leasher
Sponsored Projects Office
Encina #40, Mail Code 6060
Telephone: 723-2907
ITS Account: as.jal
SUBJECT: NSF Final Report for Grant No. DCR 8206565
As requested in our telephone conversation of this morning, the
title of the above grant is: "Mechanical Theorem Proving and
Development of EKL", for the period September 1, 1982 - June 30,
1987.
If I can be of further assistance, please let me know.
Thank you.
To: JMC@SAIL
cc: AS.JAL
∂05-Oct-87 1446 JMC re: NSF Final Report
To: AS.JAL@FORSYTHE.STANFORD.EDU
[In reply to message sent Mon, 5 Oct 87 14:41:57 PDT.]
Thanks, I wouldn't have guessed it was that grant.
∂05-Oct-87 1428 JMC
To: PHY
Sorry, that was daedal.dvi[f87,jmc].
∂05-Oct-87 1428 JMC
To: PHY
OK, please print daedal.dvi[w87,jmc] again and ship it.
∂05-Oct-87 0918 JMC re: visiting professors and industrial lecturers
To: FEIGENBAUM@SUMEX-AIM.STANFORD.EDU
[In reply to message sent Sun, 4 Oct 87 18:35:22 PDT.]
We're fine, thanks.
∂05-Oct-87 0856 JMC
To: PHY
Thanks, will do.
∂04-Oct-87 1503 JMC paper for Daedalus
To: PHY
Please print daedal.dvi[f87,jmc] and Federal Express it to
Daedalus
Norton's Woods
136 Irving Street
Cambridge, MA 02138
for arrival Tuesday.
∂04-Oct-87 1433 JMC re: transfer to stanford
To: sharma%uicsrd.CSRD.UIUC.EDU@A.CS.UIUC.EDU
[In reply to message sent Sun, 4 Oct 87 14:43:16 cdt.]
You should talk further with Joe Weening. A
faculty member you should talk to is Anoop Gupta,
415 725-3716; AG@Amadeus.stanford.edu, who has quite general interests
in parallel symbolic computing. I don't plan to be
sufficiently active in parallel computing in general
to take on new students.
∂04-Oct-87 1028 JMC re: transfer to stanford
To: sharma%uicsrd.CSRD.UIUC.EDU@A.CS.UIUC.EDU
[In reply to message sent Sat, 3 Oct 87 21:16:32 cdt.]
Your project seems interesting, but the current and projected future
activities of my project (QLISP) at Stanford are limited to software.
However, should you come to Stanford, I and others in my group would
like to hear about your work and perhaps discuss the relation between
your hardware ideas and the experience we are getting in using a
parallel computer for LISP.
∂03-Oct-87 1656 JMC re: railroad conspiracy
To: WEISE@SIERRA.Stanford.EDU, su-etc@SAIL.Stanford.EDU
[In reply to message from WEISE@sierra.stanford.edu sent Sat 3 Oct 87 14:27:34-PDT.]
Well documented doesn't mean proved. Books promoting conspiracy theories
have abounded since, I believe, the time of ancient Greeks. The
"capitalist plot" genre has been popular for more than a century.
Consider what the lawyers for the bankrupt railroads could have done if
General Motors and Exxon could have been proved to have done them in.
Triple anti-trust damages on some billions of dollars would have saved the
railroads and enriched the lawyers almost up to their wildest dreams of
avarice. You subpoena the lobbyists and lawyers for GM, etc. and take
their testimony with threats of perjury indictments. It would have been a
far better racket than merely getting a few million dollars apiece from
each accident involving Pinto gas tanks exploding. Oh well, maybe it just
didn't occur to them.
∂03-Oct-87 1640 JMC
To: *
Congratulations Marty on getting domains in.
∂02-Oct-87 2154 JMC apology
To: danny@THINK.COM
I did not make the deadline on Grace Hopper.
Next year for sure.
∂02-Oct-87 2044 JMC
To: nilsson@TENAYA.STANFORD.EDU
Thanks for the file
∂02-Oct-87 1958 JMC
To: TK@SAIL.STANFORD.EDU
Can you MAIL me phone numbers? I'd like to talk about Lisp machines.
∂02-Oct-87 1902 JMC re: Railroad Conspiracy Theories, anyone?
To: masha@JUNE.CS.WASHINGTON.EDU, su-etc@SAIL.STANFORD.EDU
[In reply to message from masha@june.cs.washington.edu sent Fri, 2 Oct 87 16:05:53 PDT.]
I believe the accusations relate not to long distance railways but
to street railways. I remember in 1955 asking my landlord in New Paltz,
New York (what the devil was I doing in New Paltz) about some bits
of tracks peaking out through the asphalt in the middle of the main
street. He said that there had formerly been a railway between
Poughkeepsie and New Paltz, but it had gone out of business before
the War. It turned out he meant before World War I.
I suspect the accusation is exaggerated. However, there is one legal
anomaly that worked against the railroads. In the 19th century all
good liberals were enemies of the wicked greedy capitalist railroads,
and part of all liberal platforms was to tax them heavily. These
taxes remained, and some still remain, high on railroad rights-of-way.
On the other hand the roads the truckers use are supported by taxes,
and the taxes on trucks themselves aren't very high. Very likely
the car companies support the highway lobbies which resist and
general lowering of taxes on railroads. However, the railroad
lobbies are not entirely impotent. They have succeeded in preventing
laws that would give coal slurry piplines the right to cross railroads.
Not all wicked lobbies are corporate. Ideologically based lobbies
can be even worse, because they often can't be bought off or
compromised with. Environmentalist lobbies have, in their short
existences, harmed the public welfare at least as much as business
lobbies.
∂02-Oct-87 1847 JMC reply to message
To: PHY
[In reply to message rcvd 02-Oct-87 15:51-PT.]
Please chuck the MAA booklet and file the rest.
∂02-Oct-87 1530 JMC letter
To: PHY
Could you texify halste.re1[f87,jmc] and send it? You
can sign it PHY for John McCarthy.
∂02-Oct-87 1456 JMC re: visiting professors and industrial lecturers
To: FEIGENBAUM@SUMEX-AIM.STANFORD.EDU
[In reply to message sent Fri, 2 Oct 87 14:43:46 PDT.]
Forbus sounds like a good idea. I think I'll wait couple weeks for
additional suggestions before listing candidates to the faculty and
inquiring about his availability.
∂02-Oct-87 1426 JMC re: Research Mentor Information
To: VAL
[In reply to message rcvd 02-Oct-87 14:14-PT.]
It looks ok. As to slots, we might say "2 unpaid slots now, but funding
is pending". I guess you get to tell Gorbis what to do for now, but
I would prefer to focus him as much as possible on common sense itself
rather than pure logic. We should go easy on him at first, because he
may have some difficulties getting oriented. Please see that he learns
what he has to do to pass the comprehensives, etc. There is a mechanism
for orienting new students, but it doesn't always succeed with foreign
students.
∂02-Oct-87 1408 JMC re: question
To: VAL
[In reply to message rcvd 02-Oct-87 13:55-PT.]
She came to Texas with us and will continue with us when we return.
Carolyn found her by way of an announcement on bboard.
∂02-Oct-87 1356 JMC logic in AI
To: nilsson@SCORE.STANFORD.EDU
Thanks for your latest draft. I may crib bits of it. I have
one quibble. I think we should adhere to the logicians'
terminology in the following respect. They don't use the
word "language" for first order logic, but reserve the
phrase "first order language" for a system that includes the
vocabulary of predicate and function symbols. I think this
usage is advantageous for AI as well as for logicians, because
it emphasizes that you get different languages according to
whether you write at(x,l), holds(at(x,l),s), loc(x,s) = l,
etc. or use a language permitting several of these.
∂02-Oct-87 1319 JMC re: Carolyn's phone number
To: VAL
[In reply to message rcvd 02-Oct-87 13:17-PT.]
Sorry, it is 3894.
∂02-Oct-87 1213 JMC re: Research Mentor Information
To: VAL
[In reply to message rcvd 02-Oct-87 11:58-PT.]
I think there should be two descriptions. However, why don't you
ask Igor to do the Qlisp one. Maybe there should be three - one
for program proving. For the latter, you should ask Carolyn whether
she wants to do it or whether Shankar should be asked. Carolyn's
phone number is 512 471-3874.
∂02-Oct-87 1204 JMC re: The Pope and Madonna
To: H.HUSSEIN@LEAR.STANFORD.EDU, su-etc@SAIL.STANFORD.EDU
[In reply to message from H.HUSSEIN@lear.stanford.edu sent Thu 1 Oct 87 22:39:38-PDT.]
Perhaps the Pope had Madonna confused with someone else.
∂02-Oct-87 1155 JMC re: visiting professors
To: BUCHANAN@SUMEX-AIM.STANFORD.EDU
[In reply to message sent Fri, 2 Oct 87 06:48:51 PDT.]
Please see my latest to the faculty. Is visiting professor better
for Reid Smith than industrial lecturer?
∂02-Oct-87 1153 JMC visiting professors and industrial lecturers
To: faculty@SCORE.STANFORD.EDU
In my previous message I should have said that visiting professors are not
the same as industrial lecturers. The latter are one per quarter and are
from local industry. The former, in my view, should primarily be
academics and will usually be from a distance. The industry lecturers
have always been appointed for a quarter, and the it seems to me that
visiting professors should normally be here for an academic year. I have
received two nominations for the visiting professor position, but both are
from local industry - Dick Duda and Reid Smith. Is there a reason why the
visiting professor position is more desirable for us and more attractive
to the potential invitee in these cases or would industrial lecturer be
more appropriate? We usually have settled the industrial lectureship
positions in January or February with negotiations opened in late fall and
this seems to suit both us and the candidates. However, the visiting
professor positions usually have to be settled in late fall so that people
can apply for leave, etc. I think we should try for some quite prominent
visiting professors.
∂01-Oct-87 1717 JMC re: visiting professors
To: FEIGENBAUM@SUMEX-AIM.STANFORD.EDU
CC: faculty@SCORE.STANFORD.EDU
[In reply to message sent Thu, 1 Oct 87 16:58:56 PDT.]
Ed asked to be reminded of the parameters regarding visiting
faculty, and I thought I'd better remind everyone.
The faculty agreed some time ago that we ought to have about a
visiting professor a year, for example, a senior person we might consider
trying to recruit. If we find a person the faculty considers suitable,
it's up to the Chairman to find the money, e.g. from the budget for
someone on unpaid leave. Shapiro visited in this way. I believe there's
no-one this year.
∂01-Oct-87 1643 JMC re: INS
To: SINGH@SIERRA.STANFORD.EDU
CC: su-etc@SAIL.STANFORD.EDU
[In reply to message from SINGH@Sierra.Stanford.EDU sent Thu 1 Oct 87 16:19:38-PDT.]
I'm sure that instances of bad behavior by INS can be multiplied
indefinitely. However, your reply was not fully responsive to
my question. Some things are Congress's fault. For example,
perhaps Congress has required consular officials (they are
part of the State Department rather than INS) to try to
guess whether a visa applicant will try to take a job
and has also imposed inconsistent requirements on INS.
However, beyond defects imposed by Congress, there seems
to be rudeness that could be cured purely administratively.
Who is the present head of INS? Is he a civil servant
or a political appointee? What interests attempt to
influence his policy? Are the immigration lawyers
a political force for good or evil? The need for
their services might partially depend on the system
being bad.
∂01-Oct-87 1524 JMC
To: CLT@SAIL.STANFORD.EDU
strange letter in telegram style from
Graphnet
329 Alfred Avenue
Teaneck, N.J. 07666
Lyngby
Dines Bjorner
Thanks for your positive reply. We are delighted. Do you have a title
for keynote speech. When will you arrive. Flight no. Day Time. Similar
for your departure. We will book double room in Copenhagen for shoulder
days/nights.
If not otherwise instructed, willyou be in Copenhagen Monday Oct 26
or culd you be so moved. Danish IFIP group is organising a 3 speaker
evening with Ershov. Somebody from Icot and would like to see you as one
of the three. Would you be available Danish IFIP group.
Would fund all additional expenses etc. Sorry to give you all this to work
on.
Sincerely yours
∂01-Oct-87 1519 JMC re: visiting professors
To: nilsson@TENAYA.STANFORD.EDU
[In reply to message sent Thu, 1 Oct 87 14:35:35 PDT.]
Do you think of him as a visiting professor or as an industrial lecturer?
Isn't there a rule that to be a visiting professor you have to be
visiting from somewhere where you have a professorial title. If
we want Duda as a visiting professor, I suppose we'll have to call
him something else.
∂01-Oct-87 1516 JMC mail
To: PHY
North-Holland computer publications catalogue 1987
discard
Artificial Intelligence journal
file
Newsweek
discard whenever it comes
`Some recent applications of knowledge' by Rohit Parikh - comments welcome
forward
An extremely lengthy letter from R. Thomas, Universite d'aix-Marseille
with xerox copies of another lengthy letter. Something to do with
`your speech in our University magazine when you taught here for a few
days, a few years ago.'
forward
∂01-Oct-87 1025 JMC re: non-monotonic reasoning
To: reiter%ai.toronto.edu@RELAY.CS.NET
[In reply to message sent Wed, 30 Sep 87 00:42:22 EDT.]
Has your survey paper been published yet?
∂01-Oct-87 1404 JMC re: Committees
To: RICHARDSON@SCORE.STANFORD.EDU
[In reply to message sent Thu 1 Oct 87 11:39:57-PDT.]
I don't know what you are talking about. I have no committees
this quarter, since I'm in Texas.
∂01-Oct-87 1410 JMC recursivity
To: su-etc@SAIL.STANFORD.EDU
Rumor has it that someone in Jesse Jackson's campaign threatened
to reveal that someone in Dukakis's campaign was responsible for
revealing that Biden used material from Kinnock's campaign.
Kinnock is now accused of copying Thatcher's slogans. She has
been accused of being a reincarnation of Genghis Khan.
∂01-Oct-87 1418 JMC visiting professors
To: faculty@SCORE.STANFORD.EDU
CC: richardson@SCORE.STANFORD.EDU
Does anyone have proposals for a visiting professor for next year?
I seem to be in charge of this although away, and I suppose I can
collect proposals and suggestions by email.
∂01-Oct-87 1433 JMC Message of 1-Oct-87 14:23:53
To: richardson@SCORE.STANFORD.EDU
∂01-Oct-87 1425 Mailer@Score.Stanford.EDU Message of 1-Oct-87 14:23:53
Received: from SCORE.STANFORD.EDU by SAIL.STANFORD.EDU with TCP; 1 Oct 87 14:25:01 PDT
Date: Thu 1 Oct 87 14:24:25-PDT
From: The Mailer Daemon <Mailer@Score.Stanford.EDU>
To: JMC@Sail.Stanford.EDU
Subject: Message of 1-Oct-87 14:23:53
Message failed for the following:
GRAD-ADMIN@Score.Stanford.EDU.#Internet: File not found
------------
Received: from labrea.stanford.edu by SCORE.STANFORD.EDU with TCP; Thu 1 Oct 87 14:24:04-PDT
Received: from Score.Stanford.EDU by labrea.stanford.edu with TCP; Thu, 1 Oct 87 14:07:55 PDT
Received: from SAIL.STANFORD.EDU by SCORE.STANFORD.EDU with TCP; Thu 1 Oct 87 14:18:37-PDT
Date: 01 Oct 87 1418 PDT
From: John McCarthy <JMC@sail.stanford.edu>
Subject: visiting professors
To: faculty@score.stanford.edu
Cc: richardson@score.stanford.edu
Does anyone have proposals for a visiting professor for next year?
I seem to be in charge of this although away, and I suppose I can
collect proposals and suggestions by email.
-------
∂01-Oct-87 1438 JMC INS
To: su-etc@SAIL.STANFORD.EDU
It is indeed true that the INS is very bad to deal with. It is my impression
that it is the worst of any U.S. Government agency. Of course, the INS will
try to pass the buck by saying that it's Congress's faulty for not giving them
money for more staff. Is there evidence to confirm my impression that the INS
could be a lot more polite even without a larger staff.
∂01-Oct-87 1455 JMC re: Reminder
To: BSCOTT@SCORE.STANFORD.EDU
[In reply to message sent Thu 1 Oct 87 14:43:14-PDT.]
Thanks, I received your message and will take care of it.
∂01-Oct-87 1459 JMC re: Various Topics
To: RPG
[In reply to message rcvd 01-Oct-87 14:47-PT.]
I can't commit myself to new writing this quarter.
I'm agreeable to taking on your Clinger's travel assuming DARPA will
pay. Ask Les about appointing Clinger visiting scholar and to find
out whether that will do.
∂06-Oct-87 1444 JMC re: aids
To: PALLAS@SUSHI.STANFORD.EDU, su-etc@SAIL.Stanford.EDU
[In reply to message from PALLAS@Sushi.Stanford.EDU sent Tue 6 Oct 87 10:46:43-PDT.]
Like many other social groups, the homosexuals have leading
people. I don't know much about it, and I was relying on what the A.P.
story said about the book in question including its author being a
"homosexual activist" and its reference to "gay community leaders". It
said they initially regarded AIDS as a PR problem rather than as a medical
problem. I did not regard myself as arguing a conspiracy. However, it
was my opinion and still is, that this PR attitude has influenced the
medical community and the liberal community and has prevented AIDS from
being treated like other communicable diseases and especially sexually
transmitted communicable diseases. This has increased its spread.
As the article suggests, homosexual leaders have changed many of
their attitudes, presumably as a consequence of so many deaths. It's hard
to identify hindsight as the reason, because the AIDS epidemic has
developed almost exactly as was predicted when the disease was first
identified and its viral nature suspected. However, most still seem to
consider the danger of embarassment as more important than the danger of
death when they oppose, for example, tracing sexual contacts of AIDS
victims. I consider exposing and opposing this attitude worthwhile, if
minor, in reducing the toll from AIDS. Ibsen wrote a play, perhaps it was
"A Pillar of the Community", in which similar attitudes towards syphilis
had similar disastrous consequences. While Ibsen complained about the
politics of respectability, the politics of victimization has had
similar effects in suppressing common sense.
The view that Alzheimer's disease is unimportant, because it
affects the old is prejudiced. Consider that George Polya contributed to
society well into his nineties while his colleague Gabor Szego did not.
The reason why not is that Szego had Alzheimer's disease.
∂06-Oct-87 2230 JMC re: Talk at Stanford
To: harnad@PRINCETON.EDU
[In reply to message sent Wed, 7 Oct 87 00:55:20 EDT.]
It seems that a talk or series of talks based on these
abstracts would be interesting, but I'm not in a position
to do anything until I return to Stanford in January. I'm
at the University of Texas in Austin until then. Also the
people with whom you would probably have most in common are those
at CSLI. Jon Barwise, John Perry and David Israel come to
mind. I could make introductions, but perhaps you know
them already. Suppes is also interested in the relations
between psychology and philosophy.
∂06-Oct-87 2338 JMC re: Talk at Stanford
To: harnad@PRINCETON.EDU
[In reply to message sent Wed, 7 Oct 87 01:46:55 edt.]
Barwise and Perry are barwise@csli.stanford.edu and
john@csli.stanford.edu. Suppes used to be pat@imsss.stanford.edu, but I'm
not sure he reads electronic mail. His telephone number is 415 723-3111.
∂06-Oct-87 2352 JMC AIDS
To: su-etc@SAIL.Stanford.EDU
I should explain that my present remarks are a revival of an argument of a
year or so ago in which I maintained that important public health measures
were being neglected. I revived the issue, because I considered that the
Shilts book supported my previous opinions - both that the measures were
necessary - and my guess as to why they were opposed. I wasn't attacking
any group, merely disagreeing with them. My remarks about quarantines
were not a specific proposal, but quite rigorous quarantines confining
people with certain communicable diseases to their homes were a common and
effective public health measure before antibiotics. They were not
considered discrimination. AIDS presents a special psychological problem,
because large parts of the population perceive themselves, probably correctly,
as not enangered.
∂07-Oct-87 1005 JMC book the Green Library should get
To: library@Score.Stanford.EDU
Today's NYT 1987 Oct 7, contains an article about
a Chinese author named Bo Yang and mentions a book
called The Ugly Chinaman and says it has been
translated into English. The man spent nine years
in prison in Taiwan, and his book has been banned
on the mainland. A Socrates search lists neither
the title nor the author. Could you pass a request
to acquire the book to the appropriate person?
Unfortunately, the article doesn't list a publisher,
but I suppose librarians are good at that.
∂07-Oct-87 1014 JMC re: gorbachev and circumscription
To: perlis@YOOHOO.CS.UMD.EDU
[In reply to message sent Wed, 7 Oct 87 13:09:00 EDT.]
I don't have a solution to the problem. While some kind of non-monotonic
reasoning is certainly appropriate, and circumscription will probably
work, my problem is with the formalization of the required general facts
about knowledge, i.e. precedes the non-monotonic part. I'm interested
in your ideas, and I'll think about it again now.
∂07-Oct-87 1318 JMC re: DAI Workshop Funding Request
To: gasser%pollux.usc.edu@OBERON.USC.EDU
[In reply to message sent Wed, 7 Oct 87 11:58:00 PDT.]
This week.
∂07-Oct-87 1430 JMC knowledge
To: halpern@IBM.COM
I'm teaching Epistemological Problems of AI, and I would
like the reference to your work or that of the others
at IBM San Jose most relevant to formalizing knowledge,
especially problems like the wise men and Mr. S and Mr. P
and "all he knows is".
∂07-Oct-87 1442 JMC same message, better address
To: halpern@ALMVMA.IBM.COM
I'm teaching Epistemological Problems of AI, and I would
like the reference to your work or that of the others
at IBM San Jose most relevant to formalizing knowledge,
especially problems like the wise men and Mr. S and Mr. P
and "all he knows is".
∂07-Oct-87 1550 JMC wrong address
To: halpern@IBM.COM
You might tell your secretary. Either she gave me a wrong address
or I misheard her. Thanks for the papers.
∂07-Oct-87 1540 mmdf@RELAY.CS.NET Failed mail (msg.aa12928)
Received: from RELAY.CS.NET by SAIL.STANFORD.EDU with TCP; 7 Oct 87 15:40:29 PDT
Date: Wed, 7 Oct 87 17:56:43 EDT
From: RELAY Mail System (MMDF) <mmdf@RELAY.CS.NET>
Sender: mmdf@RELAY.CS.NET
Subject: Failed mail (msg.aa12928)
To: JMC@SAIL.STANFORD.EDU
Your message could not be delivered to
'@ibm.com:halpern@almvma.ibm.com (host: ibm.com) (queue: ibm-sj)' for the following
reason: ' Unknown Host 'almvma''
Your message follows:
Received: from sail.stanford.edu by RELAY.CS.NET id aa12928; 7 Oct 87 17:43 EDT
Date: 07 Oct 87 1442 PDT
From: John McCarthy <JMC@SAIL.Stanford.EDU>
Subject: same message, better address
To: halpern@ALMVMA.IBM.COM
I'm teaching Epistemological Problems of AI, and I would
like the reference to your work or that of the others
at IBM San Jose most relevant to formalizing knowledge,
especially problems like the wise men and Mr. S and Mr. P
and "all he knows is".
∂07-Oct-87 1556 JMC re: more on gorbachev
To: perlis@YOOHOO.CS.UMD.EDU
[In reply to message sent Wed, 7 Oct 87 18:09:40 EDT.]
Leaving the formalism aside, it seems to me that our beliefs about Gorbachev's
knowledge derives from two assumptions.
(1) Gorbachev's knowledge of specific events derives from his experience.
(2) The experience he can have is consistent with Reagan either sitting
or standing.
This avoids having to assume anything about Gorbachev's specific mental
processes.
∂07-Oct-87 1602 JMC receipt of paper
To: ft100%utep@FORSYTHE.Stanford.EDU
To: Teodor Przymusinski
Thanks for your Non-monotonic reasoning vs. ... .
I thought I had a disagreement, which is what induced me to ask
how to send you a message, but it went away. Please acknowledge
receiving this. Regards,
∂07-Oct-87 2232 JMC supreme court
To: su-etc@SAIL.Stanford.EDU
WASHINGTON (AP) - A Jewish group has hailed passage of the
''yarmulke amendment'' to a defense authorization bill, allowing
military personel the right to wear ''neat and conservative''
religious apparel.
The legislation came after a Supreme Court decision upheld an Air
Force policy forbidding a Jewish officer to wear a yarmulke.
Rabbi David Saperstein, co-director of Reform Judaism's Religious
Action Center, says the legislation ''will eliminate the conflict
faced by observant Jews in the armed services between their religious
obligations and military duties.''
In my opinion the Supreme Court acted correctly in
declining to intervene in how the armed services regulate
uniforms on the grounds that there wasn't a constitutional
question. Also Congress acted within its powers to regulate
the armed forces in agreeing to allow military personnel the
right to wear "neat and conservative" religious apparel. They
also acted intelligently in not further defining "neat and
conservative" leaving that to the Defense Department.
Presumably there could be a future dispute as to
whether the Defense Department is reasonably interpreting
"neat and conservative". Only after administrative remedies
were exhausted, could the losers reasonably appeal to the
Federal Courts, where the judge would have Congresses law
to interpret with the record of the debate as a guide.
I suppose it won't be very controversial that this
was the proper course of events. Greater passions lead to
greater desires to get one's way at the cost of damaging
the constititutional system.
∂08-Oct-87 1625 JMC re: supreme court
To: HADDAD@SUSHI.Stanford.EDU
[In reply to message sent Thu 8 Oct 87 14:32:30-PDT.]
I didn't pay attention to the part about the appropriation bill.
(In fact it was the authorization bill, not the appropriation bill.)
In general I agree that attaching unrelated issues to bills is a
bad practice. I suspect that in this case it was getting around
Congress's own procedures, since the Armed Services committees still
haven't (I believe) finished this year's appropriation bill and wouldn't
have had time to take the yarmulke matter up separately, and it
probably went through without opposition.
∂08-Oct-87 1632 JMC re: NSF Centers
To: nilsson@TENAYA.STANFORD.EDU, DEK@SAIL.Stanford.EDU,
guibas@Score.Stanford.EDU, ullman@Score.Stanford.EDU,
RWF@SAIL.Stanford.EDU, goldberg@Score.Stanford.EDU,
ZM@SAIL.Stanford.EDU, mayr@Score.Stanford.EDU, pratt@Score.Stanford.EDU
[In reply to message from nilsson@Tenaya.Stanford.EDU sent Thu, 8 Oct 87 16:12:49 PDT.]
I would be interested in participating but couldn't be point man.
∂08-Oct-87 2044 JMC
To: ARK
.bb Mr. S and Mr. P
%2Two numbers ⊗m and ⊗n are chosen such that 2_≤_m_≤_n_≤_99.
Mr. S is told their sum and Mr. P is told their product. The following
dialogue ensues:
%2Mr. P: I don't know the numbers.
%2Mr. S: I knew you didn't know. I don't know either.
%2Mr. P: Now I know the numbers.
%2Mr. S: Now I know them too.
%2In view of the above dialogue, what are the numbers?"%1
∂08-Oct-87 2117 JMC report on ebos
To: AIR
I received a phone message from Mark Wegman saying that IBM
needs a final report on ebos before they will pay the final
bill. It doesn't have to be long, and it needs to say that
it is the final report to IBM on the project. Please do it
in the next couple of weeks. Unfortunately, I haven't been
calling my answering machine at Stanford, so I don't know
how old the message is. May be you should telephone Wegman
or try a message to wegman@ibm.com.
∂09-Oct-87 1106 JMC reference
To: VAL
Where was your Formal Theories of Action published. Is that the IJCAI-87 paper?
∂09-Oct-87 1110 JMC re: Sept Auditron Readings
To: GILBERTSON@SCORE.STANFORD.EDU
[In reply to message sent Fri 9 Oct 87 09:31:36-PDT.]
I am in Texas for the Fall and can't read them. I suppose the 358 auditron
is in that secretary's office. As for 356, I suppose my daughter Sarah,
who was acting as a secrretary during the summer packed it up with the
other things from my desk for storage till I return in January. Her
home number is 916 758-6544 if you need to ask where it is.
∂09-Oct-87 1203 JMC please send paper
To: PHY
In the file cabinet to the right of my office door is my reprint
collection. Could you send a copy of a paper entitled Common
Business Communication Language to
Dr. Robert Weigle
U.S. Army Research Office
RTP
North Carolina, 27511
∂09-Oct-87 1225 JMC re:aids
To: bothner@PESCADERO.STANFORD.EDU
[In reply to message sent 8 Oct 1987 1357-PDT.]
I have been puzzling over how I might make my postings less irritating to
people who disagree with my positions. I would actually welcome
suggestions about how I might do it. I was aware from the review that
Shilts also criticized the Reagan Administration, but it seeems to me that
I acted fairly in the matter by posting the entire article and not just
the part that supported my opinion. My problem with your suggestion is
this. I believe that the whole approach to AIDS, especially the
evaluation of uncertain factors has been affected by liberal ideology and
by positions taken by the leaders of the "gay community". These include
the following.
1. The relative importance assigned to preserving privacy vs.
reducing the spread of AIDS. This has caused AIDS to be treated quite
differently from the much less serious venereal diseases with regard to
pre-marital tests and contact tracing. The arguments about the
uncertainty of the tests and the effect of tracing on people coming in for
treatment are applicable to the other diseases and strike me as specious.
2. There was the old argument about a hospital transferring a
doctor with AIDS from direct contact with patients. I speculated on
bboard that this might be justified by the chance that his weakened immune
system would cause him to be a source of ordinary infections. I was
assured by someone in contact with experts that this wouldn't happen and a
year later read that people with AIDS were involved in the somewhat
increased incidence of tuberculosis for essentially that reason.
It occurs to me that I could not estimate the relative importance of
general liberal ideology and the specific influence of the "gay
community" in these matters, and I may have overestimated the former.
Now I may be wrong, and what I regard as irrationalities caused
by ideology may indeed be entirely objective scientific judgments.
Indeed this is why I pay special attention to changed evaluations,
that indicate that uncertain matters were not treated prudently.
Finally, I should say that I am not in sympathy with the occasional
fundamentalist who says that AIDS is God's way of punishing
immorality.
∂09-Oct-87 1236 JMC re: One More About Your Lecture
To: kam%unsun.riec.tohoku.junet%utokyo-relay.csnet@RELAY.CS.NET
[In reply to message sent Fri, 9 Oct 87 18:00:45+0900.]
Lifschitz's Formal Theories of Action was in IJCAI-87.
∂09-Oct-87 1445 JMC
To: aaai@SUMEX-AIM.Stanford.EDU
Pending Proposals
Leslie Burkholder, cmu
"lb0q#"@andrew.cmu.edu
workshop material june 9
Les Gasser
msg.msg[1,jmc]/431p
John,
I sent you a previous note about funding, but perhaps it got lost.
I'm organizing the 8th workshop on Distributed AI, and
am writing this note to request funding for it from AAAI.
The workshop will be held at the University of California's Lake
Arrowhead Conference Center, May 22-25 1988, at a cost of
$75.00/person/night (inclusive). I expect to have no more than 40
people, which would bring the cost to $9,000, and would like to have
some extra to pay for airport transportation, preparation and mailing,
xeroxing of papers, etc. I'd like to get $10,000 in support from
AAAI.
I'd like to know as soon as you can tell me when I might expect an
answer on AAAI support, as I must place a deposit of $1800.00 at the
Arrowhead Conf. Center ASAP.
A preliminary description of the workshop appears below. The planning
committee has included:
Miro Benda, (Boeing AI Center)
Phil Cohen, (SRI)
Lee Erman, (Teknowledge)
Mike Genesereth, (Stanford)
Mike Georgeff, (SRI)
Carl Hewitt, (MIT)
Mike Huhns, (MCC)
Victor Lesser, (UMASS)
Nils Nilsson (Stanford)
N.S. Sridharan, (FMC Corp)
Michael Fehling, (Rockwell)
A tentative description follows.
If you need more information, please let me know.
-- Les
Dr. Les Gasser
Computer Science Department
University of Southern California
Los Angeles, CA. 90089-0782
gasser@usc-cse.usc.edu
----------------------------------------------------------------
8th Workshop on Distributed AI
Lake Arrowhead Conference Center (Tentative)
Planned for May 1988
The 8th DAI workshop will focus on on issues of coordinating fairly
large-grain "agents," and not on issues of language-level concurrency,
fine-grained parallelism, concurrent machines, or "connectionist"
approaches. The driving focus will be be synthetic and pragmatic,
addressing questions of how we go about integrating theoretical and
experimental ideas about knowledge, planning, negotiation, action,
etc. so as to build working interacting agents?
Suggested topics:
How to describe problems, decompose them, and allocate them among a
collection of intelligent agents, including resource allocation,
setting up communication channels, dynamic allocation, etc.
How to assure coherent, coordinated interaction among intelligent
agents, including how is control allocated, how is coherence
determined, what is the role of communication in coherence,
plan synchronization, etc.
How to reason about other agents, the world, and the state of the
coordinated process, including plan recognition, prospective
reasoning (expectations), process, cognitive, knowledge, and belief
models, representation techniques, what needs modeling in what
situations, etc.
How to recognize and resolve disparities in viewpoints,
representations, knowledge, and goals (including dealing with
incomplete, inconsistent, and representationally incompatible
viewpoints) using techniques such as communication, negotiation,
conflict resolution, compromise, deal enforcement, specialization
and credibility weighting, etc.
Problems of language and communication, including interaction
languages and protocols, reasoning about communication acts
(e.g. when, what, how to communicate), dialogue coherence, etc.
Epistemological problems such as concept formation, mutual
knowledge, the mutual construction of language and coherence,
situation assessment with different frames of
reference, the problem of "shared meanings," etc.
Practical architectures for and real experiences with building
interacting intelligent agents or distributed AI systems, including
the limitations faced, resource bounded reasoning, etc.
Appropriate methodologies, evaluation criteria, and techniques for
DAI research, including comparability of results, basic assumptions,
useful concepts, canonical problems, etc.
Format:
Prospective participants should submit an extended abstract (8-10
pages) describing original work in DAI. Preference will be given to
work addressing basic research issues in DAI such as those outlined
above. A small number of "interested observers" will be selected for
participation and need only submit a request to attend with
some justification.
A number of submitted papers will be selected for full presentation,
critique, and discussion. Other participants will be able to present
their work in a "poster session." There will be ample time allowed for
informal discussion.
Participation will be limited to 35-40 people.
∂10-Oct-87 1737 JMC re: `America's Secret Wars Unveiled,' by Hodding Carter III
To: SINGH@SIERRA.Stanford.EDU, su-etc@SAIL.Stanford.EDU
[In reply to message from SINGH@Sierra.Stanford.EDU sent Sat 10 Oct 87 16:50:42-PDT.]
As I recall, Hodding Carter III had a fairly high position in the Carter
Administration, perhaps in charge of information for the State Department.
Most likely his views about the U.S. in world affairs today are not
enormously different from what they were then. From a conservative point
of view, that men of his views held such high positions in the U.S. Government,
(he was one of many that held office under Carter) is frightening.
It is also surprising from that point of view that they didn't do more
damage than they did. From a liberal point of view, I suppose it is
surprising that they didn't do more to diminish the influence of the
U.S. in world affairs.
It is not the purpose of this message to argue with Carter's point of view;
maybe I'll do that later.
Congratulations to Harinder on his typing; he probably didn't have our
opportunity to take typing in junior high school.
∂10-Oct-87 2041 JMC
To: suppes@CSLI.STANFORD.EDU
U.S. - Soviet collaboration on computer education of mathematically talented
high school students.
This is a potential project to which both sides make equal and large
contributions. Americans have more experience with computers, but it is my
impression that more first class Soviet mathematicians and scientists have
paid attention to the education of high school students. It is also more
suited for collaboration than computer education of high school students
in general, because the latter is extremely dependent on difficult to alter
institutional traditions and on the availability of money for hardware and
the kinds of hardware that are available. Programs limited to mathematically
talented students will be smaller, and it will be easier to make them
comparable. It is will also be easier to involve scientific organizations
that are more used to international collaboration than are the much larger
educational organizations.
In so far as there are immediately interested people in other countries,
they might also be involved with less trouble.
∂10-Oct-87 2118 JMC
To: pat@IMSSS
I forgot to say that if there is a successful summit, joint projects will be wanted.
∂13-Oct-87 2016 JMC re: DARPA Umbrella Contract
To: LES
[In reply to message rcvd 12-Oct-87 15:49-PT.]
I am asking Vladimir to be my proxy, so please include him in subsequent
messages. If you think qlisp needs representation at this time, please
ask Igor.
∂13-Oct-87 2018 JMC
To: VAL
Please be my proxy at Les's meeting Oct. 15. I have told him.
∂12-Oct-87 1549 LES DARPA Umbrella Contract
To: binford@WHITNEY.STANFORD.EDU, Buchanan@SUMEX-AIM.Stanford.EDU,
cheriton@PESCADERO.Stanford.EDU, Engelmore@SUMEX-AIM.Stanford.EDU,
Feigenbaum@SUMEX-AIM.Stanford.EDU, Genesereth@SCORE.STANFORD.EDU,
latombe@WHITNEY.STANFORD.EDU, DCL@SAIL.Stanford.EDU,
JMC@SAIL.Stanford.EDU, Nilsson@SCORE.STANFORD.EDU, BScott@SCORE.STANFORD.EDU,
Rindfleisch@SUMEX-AIM.Stanford.EDU, Wiederhold@SUMEX-AIM.Stanford.EDU
The planned extension to the DARPA umbrella contract that is currently
supporting a number of projects in the department, as well as Luckham's
group in EE, is scheduled to lapse 12/1/90. KSL has another umbrella that
I understand finishes on 8/21/88; I trust that either it will be extended
or that the projects on that one will be moved to the extension of the
other umbrella before then.
We need to have a successor contract in place before the existing one(s)
finish or we will be in a lot of trouble. As I understand it, DARPA would
like to have just one such contract in order to minimize administrative
complexity.
We have been less than impressed with SPAWASYSCOM as a contracting agency.
One possible contracting agent that Bill Scherlis suggests is NASA Ames.
Peter Friedland there says that he is willing to consider it if the
proposed work lies within the interet areas of his group. I heard from
Bob Engelmore that DARPA is now doing some contrcts in-house, so that may
be another possibility.
In any case, we need to put together a proposal outlining the scope of
work and preliminary budgets for the various known component projects.
I propose that we aim for completion of this proposal by November 13
(which happens to be a Friday!). In order to get organized, we should
have a meeting soon -- say on Thursday, 10/15 at 3:00pm in Nils'
conference room. Please try to come or appoint a proxy.
Les Earnest
∂13-Oct-87 2022 JMC reply to message
To: PHY
[In reply to message rcvd 13-Oct-87 17:03-PT.]
Please reply to Bjorner that I can't stay till Monday.
∂14-Oct-87 1158 JMC re: Dangers of D & D
To: helen@PSYCH.STANFORD.EDU, su-etc@SAIL.Stanford.EDU
[In reply to message from helen@psych.stanford.edu sent Wed, 14 Oct 87 04:06:03 PDT.]
Helen's grumble about D&D glorifying war is accompanied by "I don't think
it's gonadal, frankly. I think it's learned, and as long as society teaches
people to get their kicks that way, war will be with us." It seems to me
that Helen is expressing a personal opinion about a matter in which
psychology and sociology ought (but probably don't) have something
scientific to say. I guess anthropology and history are also relevant.
My own opinion is that fighting is natural in the sense that primitives
fight a lot and so would children left to grow up unguided. Modern
middle class society does very well in reducing the level of violence.
The number of well brought up middle class boys who kill anybody is
low enough to that this is a minor cause of death. Certain subcultures,
e.g. the so-called "ghetto culture" are much worse, and murders in fights
and as a method of competition in illegal businesses is a major cause
of death. In contrast to this, the American military maintains sufficient
discipline that murder is a minor hazard in the service, even though it
is full of young men, who are attracted by military values and taught
to kill. I would also bet that the level of murder among men who have
been through the military is less than among those the military rejects
as unsuitable.
These theoretical remarks have the practical application is that the
way to reduce killing is to break the ghetto culture. Probably it
would take a lot of arrests to eliminate the attractiveness of the
gangs to 12 to 15 year old boys. There should be an experiment with
a massive anti-gang campaign is some "ghetto" small enough so that
the number of people who would have to be emprisoned wouldn't strain
the capacity of the penal system of the given state. Of course, the
area chosen would have to be at some distance from othe gang-ridden
areas.
∂14-Oct-87 1203 JMC reply to message
To: PHY
[In reply to message rcvd 14-Oct-87 07:55-PT.]
You're right. Please reply to Steensgaard-Madsen.
∂14-Oct-87 1238 JMC re: References
To: HALPERN@IBM.COM
[In reply to message sent 13 October 1987, 17:50:21 PDT.]
Thanks for the papers. The only published discussion I can cite
is in the old book "The Concept of Mind" by Gilbert Ryle.
However, that may have triggered further discussion among philosophers.
Ryle doesn't use logic. I suppose Thomason would know if anyone
has formalized it.
∂14-Oct-87 1258 JMC re: Kyoto Prize
To: goguen@CSL.SRI.COM
[In reply to message sent Tue, 13 Oct 87 15:53:23 -0700.]
Many thanks for your proposal to nominate me. Prof. Takayasu Ito
of Tohoku University, Sendai, Japan nominated me the last time
the prize was offered in something like computer science and
communication. I believe it was awarded to John Pierce for his
work on communication satellites. Anyway you should send a message
to Ito, who is
Professor Takayasu Ito
Department of Information Engineering
Faculty of Engineering
Tohuku University
Sendai, JAPAN 980
* (office: 0222 22-1800, x4269)
* (home: 0222 79-2945)
* (date: 1987 Jan)
* (electronic: ito@aoba.tohoku.junet (date: 1986 Dec))
I suppose the 5 papers should be
1. Programs with Common Sense which started the logic approach to AI.
2. Computing with Symbolic Expressions, the first paper on LISP.
3. Time-sharing Computer Systems in Management and the Computer, the
first paper on time-sharing in the sense of many people simultaneously
using a computer for general purpose computing.
4. A Basis for a Mathematical Theory of Computation, proving properties
of programs regarded as equations relating functions
5. Circumscription - a method of non-monotonic reasoning.
In this I have chosen my first paper in each area, not the most
comprehensive paper. If you leave out the time-sharing paper, because
it doesn't fully document my contribution, you might add Ascribing
Mental Qualities to Machines.
I'll send you a copy of my biography which has the other material you
asked for.
Thanks again.
John
∂14-Oct-87 1313 JMC Please print
To: PHY
biojmc.dvi[f87,jmc] and send it to Joseph Goguen at SRI.
∂14-Oct-87 1325 JMC re: McCarthy's Anti-Gang Campaign
To: SIEGMAN@SIERRA.STANFORD.EDU, su-etc@SAIL.Stanford.EDU
[In reply to message from SIEGMAN@Sierra.Stanford.EDU sent Wed 14 Oct 87 13:09:53-PDT.]
Maybe the social workers' approach will succeed if implemented vigorously enough,
but it hasn't yet. There are enough gang-ridden communities in the U.S. to
simultaneously try approaches with quite opposite rationales on different ones.
I regard the the "ghetto culture" that makes heroes of gangsters, pimps with
fancy cars, and rich drug dealers as a predatory culture analogous to predatory
cultures of the past, e.g German robber barons, Norman knights and predatory
gangs on the fringes of various civilizations. Sometimes these cultures have
conquered their victims and, over many generations, developed monarchies that
enforced order and developed civilization. Perhaps Christianity was even a
civilizing force as is claimed. Sometimes they were suppressed. Since
the "ghetto" culture isn't going to conquer the U.S., the first alternative
is foreclosed.
∂14-Oct-87 1329 JMC re: LISP in mathematics
To: ILAN@Score.Stanford.EDU, su-etc@SAIL.Stanford.EDU
[In reply to message from ILAN@Score.Stanford.EDU sent Wed 14 Oct 87 13:20:52-PDT.]
J. Wellington Tartaglia was a student of Allen Newell's, so I assume he
did it in IPL-V.
∂14-Oct-87 1359 JMC re: Les's meeting
To: VAL
[In reply to message rcvd 14-Oct-87 13:49-PT.]
Yes, it does. DARPA finds the contract award process very complicated,
because of all the relevant rules. It is easier for them to add "tasks"
to an existing contract, i.e. it takes fewer months. Therefore, there
is a broadly written umbrella contract written for considerably more
money than they have any immediate intention to award. Our contract
is a task under a present umbrella; even then there are bureaucratic
problems. Incidentally, we have NSF grants and DARPA contracts.
The difference is that Stanford undertakes to deliver something,
only reports, on a contract. Les understands all this, including
any new developments, and I suggest you talk to him and get him
to explain it.
∂14-Oct-87 1516 JMC re: Kyoto Prize
To: goguen@CSL.SRI.COM
[In reply to message sent Wed, 14 Oct 87 14:01:13 -0700.]
No, need to call back. I phoned after seeing that I had a message
from you, but SAIL was down so I couldn't read it.
I'll send the papers. I don't have a regular secretary at present,
but Don Knuth's secretary, Phyllis Winkler, if filling in temporarily.
I didn't know about the confidentiality, which I suppose is mainly
to keep the fact of nominations out of the press. My idea was just
that Ito knows the ropes, but it certainly isn't necessary, and
I suppose independent nominations add some force.
∂14-Oct-87 1521 JMC more for Goguen
To: PHY
Please send the six papers mentioned to Goguen.
I believe they're all in the file cabinet to the
right of the door in my office.
∂14-Oct-87 1258 JMC re: Kyoto Prize
To: goguen@CSL.SRI.COM
[In reply to message sent Tue, 13 Oct 87 15:53:23 -0700.]
Many thanks for your proposal to nominate me.
I suppose the 5 papers should be
1. Programs with Common Sense which started the logic approach to AI.
2. Computing with Symbolic Expressions, the first paper on LISP.
3. Time-sharing Computer Systems in Management and the Computer, the
first paper on time-sharing in the sense of many people simultaneously
using a computer for general purpose computing.
4. A Basis for a Mathematical Theory of Computation, proving properties
of programs regarded as equations relating functions
5. Circumscription - a method of non-monotonic reasoning.
In this I have chosen my first paper in each area, not the most
comprehensive paper. If you leave out the time-sharing paper, because
it doesn't fully document my contribution, you might add Ascribing
Mental Qualities to Machines.
I'll send you a copy of my biography which has the other material you
asked for.
Thanks again.
John
∂14-Oct-87 2035 JMC Brink's proposed modifications
To: hemenway@SCORE.Stanford.EDU
CC: brink@SUSHI.Stanford.EDU
in his Master's programs in his two messages for you and me seem
reasonable to me, and I hereby approve them.
John McCarthy
∂14-Oct-87 2058 JMC reply to message
To: PHY
[In reply to message rcvd 14-Oct-87 16:05-PT.]
Now I remember. The title of that paper is Recursive Functions of Symbolic
Expressions and their Computation by Machine. part I
discard the Bridge and forward the package. I'll reward Kelly.
∂15-Oct-87 0820 JMC
To: PHY
Notices of the American Math Society
fwd
News release: Information Interational and NeEASI-WEBER announce a better
solution to ad management
disc
American Academy of Arts and Sciences - meeting schedule
disc
many advertisements about books from SEAI Technical Publications
disc
Call for papers nineteenth annual Pittsburgh Conference on Modeling and Simulation
May 5-6, 1988
disc
AAAI-88 Seventh National Conference on A.I. call for papers Aug 22-26, 1988
St. Paul, Minnesota
fwd
CRC Press 1987 New Titles Catalog
disc
Stanford News - The Centennial Campaign
disc
Dean's Innovation Fund - invitation to apply
fwd
Values, Technology, Science, and Society
Please check for any error - Virginia Mann 723-2565
F Technological Opportunities for Humanity
McCarthy 60-62P T Th 1:15 3 units
fwd
Science magazine
fwd
hardcopy of Report of the Workshop on Environments for Computational
Mathematics, held July 30, 1987 during the ACM SIGgraph Conference in Anaheim
from Xerox
file
letter from Artificial Intelligence journal - about a review by Nigel Ward
on Computation and Cognition: Toward a foundation for cognitive science
by Z. Pylyshyn. Mark Stefik, A.I., wants you to give another review of
this book. If you are interested, and do not already have the book,
Stefik will rush you a copy of it.
fwd
NTT Procurement Seminar announcement - 2 seminars this fall.
disc
A group of reports from (University of Goteborg and Chalmers University of
Technology) Programming Methodology Group:
fil
`On a nonconstructive type theory and program derivation' by Jan Smith
by Jan M Smith
fil
`The Independence of Peano's fourth axiom from Martin-Lof's type theory
without universes' by Jan M. Smith
fil
`Inverse Image analysis' by Peter Dybjer
fil
`An efficiency comparison of some representations of purely functional
arrays' by Annika Aasa, Soren Holmstrom, Christina Nilsson
fil
`Views: a way for pattern matching to cohabit with data abstraction'
by Philip Wadler
fil
`Projections for strictness analysis' by P. Wadler, R.J.M. Hughes
fil
Invitation to nominate candidates for the Kyoto Prizes 1988 from
The Inamori Foundation
fwd
∂15-Oct-87 0854 JMC reply to message
To: PHY
[In reply to message rcvd 15-Oct-87 08:37-PT.]
Hmm. I guess I ran out of reprints. It was published in CACM, April 1960.
Could you make 10 copies from the library, send one to Goguen, and
put the rest in the reprint file cabinet suitably labelled?
∂15-Oct-87 1201 JMC class next week
To: VAL
I told them you'd talk about computing circumscription and your theory of action. They
were asked to read your Formal Theories of Action, Computing Circumscription and Semantics
of STRIPS in that priority. Suezette Branton, AI.Branton@r20.utexas.edu, will help with
reservations, etc. The class meets on Tuesday and Thursday, 12:30-2pm, and it has some
very good students in it. Many thanks for agreeing to do this. - John
∂15-Oct-87 1337 JMC re: class next week
To: VAL
[In reply to message rcvd 15-Oct-87 13:12-PT.]
Taylor 2.106. Also is there anyone you would especially like me to alert
to your presence besides Woody and Bob?
∂15-Oct-87 1946 JMC re: [*,RA]
To: ME
[In reply to message rcvd 15-Oct-87 16:16-PT.]
I will not be able to look at them until about the end of the month.
However, there's no reason not to change the password immediately.
Probably you should ask her if she would like any of the files printed.
∂16-Oct-87 0836 JMC re: CIS vending machine.
To: stantz@HELENS.STANFORD.EDU, su-etc@SAIL.Stanford.EDU
[In reply to message from stantz@helens.stanford.edu sent Fri, 16 Oct 87 00:55:31 PDT.]
Now if you were to use the CSD vending machine in MJH and volunteer to be one
of the people who loads it, you could put the burritos in yourself. Alas,
I suppose the CIS vending machine isn't eater-operated, because both the
level of technology and the level of co-operation are lower than they
were in the Stanford Artificial Intelligence Laboratory.
∂16-Oct-87 0841 JMC re: CIS vending machine.
To: stantz@HELENS.STANFORD.EDU, su-etc@SAIL.Stanford.EDU
[In reply to message from stantz@helens.stanford.edu sent Fri, 16 Oct 87 00:55:31 PDT.]
Addendum to previous message. I have to point out that, at least
chronologically, it was Helen's generation that was responsible for
both the technology and the co-operation involved in initiating
the eater-operated vending machine.
∂16-Oct-87 0856 JMC Computer Forum
To: RDZ@SAIL.Stanford.EDU, JJW@SAIL.Stanford.EDU
Ramin, how would you like to speak to the Computer Forum, which is being held
Feb. 10 and 11? One of your present talks would be suitable.
Joe, have you spoken yet? Do you have a talk different from what the Forum
has heard?
time table:
November 1 nominations must be received
November 15 student speakers will be notified
December 1 working titles due
January 15 abstract of talk and photo ready copies of viewgraphs
February 8 Forsythe Lecture
February 9 Forsythe Lecture - Forum Twentieth Annual Meeting
February 10 Meeting continues
February 11 conclusion: reception at Faculty Club from 4:30-6
∂16-Oct-87 0857 JMC
To: VAL
How is Gorbis surviving?
∂16-Oct-87 1519 JMC re: CIS vending machine.
To: stantz@HELENS.STANFORD.EDU
[In reply to message sent Fri, 16 Oct 87 14:45:44 PDT.]
Is the CIS vending machine an ordinary commercial machine or is it
operated by one of the CIS computers?
∂16-Oct-87 1533 JMC re: Helen's generation
To: WALT@SUSHI.STANFORD.EDU, su-etc@SAIL.Stanford.EDU
[In reply to message from WALT@Sushi.Stanford.EDU sent Fri 16 Oct 87 15:01:52-PDT.]
The grumble about "the generation that endowed us with the threat of extinction
via nuclear war" seems like wishful thinking. Who do you think might have done
what differently, and how would this have avoided "the threat of extinction
via nuclear war"? Stalin might have, but he wasn't a generation by himself,
and probably wasn't whom you had in mind.
∂16-Oct-87 2022 JMC re: Re: Helen's generation
To: WALT@SUSHI.STANFORD.EDU, su-etc@SAIL.Stanford.EDU
[In reply to message from WALT@Sushi.Stanford.EDU sent Fri 16 Oct 87 16:15:07-PDT.]
Sorry to have misinterpreted. It was the phrase "the generation that
endowed us with" that misled me.
∂16-Oct-87 2140 JMC (→20773 26-Oct-87)
To: "#___JMC.PLN[2,2]"
I am at the University of Texas in Austin for Fall 1987 and
will return to Stanford for Winter Quarter. However, I
will continue to receive email as JMC@SAIL.STANFORD.EDU.
My office number in Texas is 512 471-9558 and my home
number is 512 328-1625. U.S. mail sent to Stanford is
forwarded, by my address in Texas is
Computer Science Department
University of Texas at Austin
Austin, TX 78712
I will be in Europe from October 17 thru October 25.