Yup. As far as I can tell Sussman is coming a little late to the party.
I agree with the spirit of what he says, inasmuch as he gets specific,
but others have been saying similar things.
There is a general idea that computer systems could be built in such a
way as to rely more on weak constraints (not to be confused with the
kind of constraint programming that appears in Logic based AI). The
connectionists were saying this loudly in the early 80s. I have been
saying it privately (assuming, naively that everyone already understood
it to be obvious), in the specific context of building software tools,
for almost as long.
Also, doesn't some of what Sussman says have echoes with the work on
Generative Programming (e.g. Czarnecki and Eisenecker, 2000)?
Correct me if I am wrong, anyone (I don't have time to read Sussman in
great detail), but does he suggest a particular new theme that I missed?
Richard Loosemore.
Mark Waser wrote:
I think that it's also very important/interesting to note that his
subject headings exactly specify the "development environment" that
Richard Loosemoore and others are pushing for (i.e. An "Infrastructure
to Support Generalizability" with "Generality of parts" and "Extensible
generic operations" that allows you to "Generate and test" -- or later
-- "Infrastructure to Support Robustness and Evolvability" with
"Combinators", "Continuations", "Backtracking and concurrency",
"Arbitrary association" and "Dynamically configured interfaces").
----- Original Message ----- From: "J. Storrs Hall, PhD."
<[EMAIL PROTECTED]>
To: <[email protected]>
Sent: Monday, February 26, 2007 2:08 PM
Subject: [agi] Sussman robust systems paper
I recently ran across this paper by Sussman on robust software systems:
http://swiss.csail.mit.edu/classes/symbolic/spring07/readings/robust-systems.pdf
And I was flabbergasted to find that there was about a 50% overlap
with the
ideas behind the system I'm working on. (It's also interesting to note
that
many of his examples come from classical dynamics and electronics, i.e.
dynamical systems modelled by diff. eq's.)
Key quote:
We are taught that the "correctness" of software is paramount, and
that
correctness is to b e achieved by establishing formal specification of
comp o-
nents and systems of comp onents and by providing pro ofs that the
specifi-
cations of a combination of comp onents are met by the specifications
of the
components and the pattern by which they are combined. I assert that this
discipline enhances the brittleness of systems. In fact, to make truly
robust
systems we must discard such a tight discipline.
Josh
-----
This list is sponsored by AGIRI: http://www.agiri.org/email
To unsubscribe or change your options, please go to:
http://v2.listbox.com/member/?list_id=303
-----
This list is sponsored by AGIRI: http://www.agiri.org/email
To unsubscribe or change your options, please go to:
http://v2.listbox.com/member/?list_id=303
-----
This list is sponsored by AGIRI: http://www.agiri.org/email
To unsubscribe or change your options, please go to:
http://v2.listbox.com/member/?list_id=303
