From newshub.ccs.yorku.ca!ists!helios.physics.utoronto.ca!news-server.ecf!utgpu!cs.utexas.edu!qt.cs.utexas.edu!yale.edu!cs.yale.edu!mcdermott-drew Thu Apr 30 15:22:03 EDT 1992
Article 5194 of comp.ai.philosophy:
Newsgroups: comp.ai.philosophy
Path: newshub.ccs.yorku.ca!ists!helios.physics.utoronto.ca!news-server.ecf!utgpu!cs.utexas.edu!qt.cs.utexas.edu!yale.edu!cs.yale.edu!mcdermott-drew
>From: mcdermott-drew@CS.YALE.EDU (Drew McDermott)
Subject: Educating Searle and Zeleny
Message-ID: <1992Apr22.145058.10183@cs.yale.edu>
Summary: Operating Systems 101
Keywords: virtual,processes,Searle
Sender: news@cs.yale.edu (Usenet News)
Nntp-Posting-Host: aden.ai.cs.yale.edu
Organization: Yale University Computer Science Dept., New Haven, CT 06520-2158
Date: Wed, 22 Apr 1992 14:50:58 GMT
Lines: 77

At some time in the remote past, Zeleny wrote:

  In article <1992Apr7.153156.10030@cs.yale.edu> 
  mcdermott-drew@CS.YALE.EDU (Drew McDermott) writes:

  >That's true.  Let me try to state the point noninsultingly: I have a
  >lot of respect for Searle, when he sticks to his home turf of the
  >"logic of introspection" (a phrase I just made up).  But it's obvious
  >that he has never taken a course on operating systems, and that his
  >style of rebutting arguments against computationalists would be quite
  >different if he had.

  Let me try to state the counterpoint noninsultingly: I have some respect
  for you, when you stick to your home turf of the non-monotonic logic (as
  little as I respect the subject itself).  But it's obvious that you have
  never taken a course on the philosophy of mind, and that your style of
  rebutting Searle's dismissal of the System's Reply arguments would be quite
  different if you had.

Searle has less of an excuse than I do for ignorance; there's less
computer science to know than there is philosophy of mind.  (At least,
as measured by cubic meters.)

  Incidentally, if you sincerely believe that any information about operating
  systems might change the outcome of this stalemate, I urge you to try it
  out on me.  Even though I've never bothered to take any courses in the
  computer "science", I doubt that you would be able to convincingly extend
  your condescending characterization of Searle to my own case.

Okay.  The key fact that Searle types overlook is the importance of
abstract entities in computers.  Almost all of the entities one takes
for granted in dealing with computers exist only "virtually."  A
virtual entity is one that exists because some ensemble of programs
behave as if it existed.  Here's an example.  Suppose a program is
running in a cluster of workstations, and the program's job is to find
idle workstations in the cluster and recruit them to work on some
task.  (Such programs do exist.)  At any time, there is a subset of
the cluster that corresponds to "the group working on the task."  By
virtue of what does this subset exist?  It's not physically
distinguishable in any way.  It exists because in the controlling
computer there is a table of currently recruited workstations and
corresponding data structures on the workstations pointing back to the
controller.  But of course these inert data structures do not cause
something to exist; for them to be effective, there must be programs
that read the data structures and interpret them in the intended way.
E.g., if the controller wants to communicate with every workstation
in the group, it consults the table to see where they are; and so
forth.

The point that computer novices overlook is that almost everything in
the machine exists in just this sense.  That object moving across your
screen, for instance.  (Assuming you're playing asteroids when you
could be reading comp.ai.philosophy.)  It's only "virtual."  Some data
structure somewhere describes its shape, position, and velocity, and
then programs "conspire" to make those features veridical.  And, of
course, that data structure, and almost all almost all other data
structures, are merely virtual.

A key type of virtual entity is a process, or thread of control.  It
is almost always misleading to talk about a computer doing something,
because most of what computers do is mediated by processes.  If you're
talking to an editor and a mailer, then you're dealing with separate
entities.  Furthermore, it won't do to say that the computer is
merely, as it were, shamming multiple-personality disorder, now
playing one role and now another.  The reason it won't do is that
processes can migrate from one processor to another; a process exists
independently of any particular physical embodiment, so long as it
stays embodied somewhere or other at any time.

This may sound mystical, but it's taught in elementary
operating-systems courses.  The reason it is relevant to Searle is
that if he knew about this stuff, he would realize that the Chinese
Room argument has to be rephrased in terms of processes, not
processors.  

                                             -- Drew McDermott



