Path: utzoo!utgpu!jarvis.csri.toronto.edu!mailrus!uwm.edu!uakari.primate.wisc.edu!brutus.cs.uiuc.edu!apple!snorkelwacker!spdcc!xylogics!world!madd
From: madd@world.std.com (jim frost)
Newsgroups: comp.software-eng
Subject: Re: CS education [engineering, mathematics, and computer science]
Message-ID: <1989Nov22.195230.11965@world.std.com>
Date: 22 Nov 89 19:52:30 GMT
References: <1989Nov20.160513.19313@world.std.com> <34804@regenmeister.uucp>
Organization: Software Tool & Die
Lines: 73

chrisp@regenmeister.uucp (Chris Prael) writes:
>From article <1989Nov20.160513.19313@world.std.com>, by madd@world.std.com (jim frost):
>> I've found mathematics to be priceless in both practical and
>> theoretical computing.  

>Some examples would be useful.

Most of the networks you're using use high-powered math, particularly
graph theory, to get reasonable performance.  Go pick up a networking
book worth its name and you'll find tons of those little squiggles all
through it.

Statistics are in use every day.  Ever used a paging kernel?  They use
statistics to predict future probabilities.  Which page should I get
rid of?  Which are likely to be used soon?  There's math behind all
that, since the pager is doing statistical analysis (often tempered
with a bit of generalizations like "programs run linearly" which may
or may not hold water).

Much of graphics is modelling real-life which works on physics which
works on mathematics.  Ever done ray-tracing?

These are off the top of my head.  Go get an algorithms book and go
through it and you'll find HUNDREDS of examples, most of which will be
better than mine because most of those guys are real theoreticians.

>> It certainly can avoid doing things the hard
>> way, and we can avoid the mistakes of our predecessors who had to do
>> so because of a poor understanding of the problems at hand.

>More mistakes and misunderstandings in computing and its development 
>can be traced to pseudo-mathematical pretences than can be traced to 
>mathematical ignorance.

Huh?  Most of the mistakes I've seen were the result of people not
understanding complexity, something which you learn early in
mathematics.  Bubble-sorts don't work on million-record databases.  2n
programmers don't create twice the code in the same amount of time.
Summing a series expicitly is not as effective as calculating a general
case and using it instead.  These are the kinds of mistakes you see
all over the place, most often with people who spend all of their time
worrying about having the most efficient code.

I'd like to see counterexamples.

>As did Dario, your argument raises the 
>question of how deep or shallow your comprehension of either mathematics 
>or computing are.

At least I acknowledge a relationship.  Do you?

>One thing that becomes particularly clear if one examines the development 
>of the field over the last forty years is that the pace of innovation has 
>decreased drastically in the last 20 to 25 years.  Kids come out of school 
>with 25 year old techniques and ideas, believing that this stuff is brand 
>new.  One can blame the slowing partly on the "maturing" of computing as a 
>field.

I'm confused here.  I mean, really confused.  I was shocked when
studying algorithms to find that most of the people involved are still
alive and working on things.  Development hasn't slowed, it's
accellerated to the point where it's impossible to keep up with it
all.  People building on previous theories or proving they degrade in
this-or-that circumstance or showing that some other technique is
optimal, or nearly so, or as optimal as possible given that you can't
know the future.  I read the technical reports and every month there's
something new.  If you don't see it, you're not looking for it, or
you're missing it because it's a much wider field than it used to be.
It certainly isn't going slow to me.

jim frost
software tool & die
madd@std.com