NetNews Usenet Archive 1992 #26/NN_1992_26.iso/spool/sci/logic/1964

Internet Message Format | 1992-11-09 | 5.6 KB

Xref: sparky sci.logic:1964 sci.philosophy.meta:2414 sci.math:14611
Newsgroups: sci.logic,sci.philosophy.meta,sci.math
Path: sparky!uunet!stanford.edu!CSD-NewsHost.Stanford.EDU!Sunburn.Stanford.EDU!pratt
From: [e-mail] (Vaughan R. Pratt)
Subject: What is Computer Science (was: Natural Kinds)
Message-ID: <[e-mail]>
Sender: [e-mail]
Organization: Computer Science Department,  Stanford University.
References: <[e-mail]> <[e-mail]> <[e-mail]>
Date: Mon, 9 Nov 1992 00:52:41 GMT
Lines: 101

In article <[e-mail]> [e-mail] (Angus H Rodgers) writes:
>To make this post a bit more interesting, I'll venture a definition of
>"computer science". I've tested (!) this against Vaughan Pratt's list
>of questions, and it seems to bear up, or at least not to look too
>silly to be worth mentioning:
>
>   Computer science is the science of discrete processes. The
>   automatic computing machine -- which is one kind of discrete
>   process, possessing a universal property which enables it
>   in principle to simulate any other such process -- is
>   conventionally taken to stand (both metonymically and
>   synecdochally) for the whole field. So the problem with
>   the phrase "computer science" lies not in the word "science",
>   but in the word "computer": the tool has eclipsed its object.
>
>Or, again:
>
>   Computer science is discrete applied mathematics. But this
>   is not to say that the discrete mathematics needed for the
>   applications always exists before the applications do. So
>   computer science is not applied discrete mathematics; the
>   qualifiers do not commute.
>
>(Comments? Flames? Alternatives? - Myself, I'm not sure whether
>the adjective "complex" should somehow also be worked into the
>definition(s).)

Gee, I just got through posting the following concerning discreteness
in CS on sci.physics, I see now I should have cross-posted it to
sci.logic.  But perhaps this topic is leaning more towards sci.math,
crosslisted here (also the subject line was overdue for a change).

A propos of discreteness in the list Angus cites (sci.logic,
[e-mail]), my questions 14-17, in
the area of computer graphics, are largely analog questions.  Also
question 2 on polynomial multiplication, although indeed about the
discrete Fourier transform, can be used to speed up polynomial
multiplication with coefficients from any ring supporting the Fourier
transform, including the reals and the complexes.

To me computer science is about four interactions: information and
time, computation and communication, models and reality, and
perspective and users.  Discreteness is neither emphasized nor
deemphasized in this view, the integers and the reals are both
important in CS.
--
Vaughan Pratt


In article <1992Nov8.154955.15938@prim> [e-mail] (Dave Griffiths) writes:
>After a while, the implications of working with zeros and ones just start
>to sink in. The digitization of life has been proceeding so gradually that
>maybe you don't realize what a revolution we are going through. Music is now
>nearly all digital and most of us can't tell the difference. Occams razor
>might suggest that there are _no_ continuums out there. We can get by OK
>without 'em.
>
>Now you might argue that computers are just a tool, no big deal. But
>I would say that technology leading to digitization is no coincidence,
>rather the digitization of life _reflects_ the underlying nature of reality.
>
>Nature is trying to give us a hint.
>
>Poor old physics is still stuck in the analog nineteenth century. The whole
>mighty edifice has been built on sand and the tide is coming in...
>
>Dave Griffiths

You are forgetting that scalability is a big concern today in CS.
Digital models don't scale gracefully, the bits look bigger.  Smooth
scalability demands a smooth view of what one is scaling.  A digital
view inhibits scalability.

When I was just starting as a Ph.D. student in Berkeley in 1969 I heard
a lecture by someone whose surname I think was Greenspan.  His topic
was the numerical solution of differential equations, and throughout
the lecture he kept making points of the above ilk, how everything was
really discrete when you got right down to it and numerical analysis
needed to take much more advantage of all this discreteness in the
world than it did.  Well, the two numerical analysts present, Vel Kahan
and Beresford Parlett, were incensed.  At question time Kahan ripped
right into Greenspan. challenging the efficacy of his "digital"
approach.  Never having witnessed such a ferocious attack in a lecture
before, I asked Parlett his opinion afterwards for calibration and he
assured me in very strong terms that he too felt that Greenspan's
digital emphasis was not at all helpful (he used a word that I expect
he would prefer not to have attributed to him in this forum).

The situation today with fonts is like this.  In the early 1980's many
people *identified* computer fonts with bitmaps.  Only a very few back
then in the desktop computer business realized that bitmaps would
become increasingly expensive as pixels shrank.  Today you can buy a
600 dot per inch printer for $2K, and at that granularity bitmap fonts
are a joke, smooth outline fonts are the only way to go.

That computers are digital has no more significance than that they are
silicon based.  With your argument you could equally convincingly show
that poor old biology is still stuck in the carbon-based nineteenth
century.
--
Vaughan Pratt