Relay-Version: version B 2.10 5/3/83; site utzoo.UUCP
Posting-Version: version B 2.10.2 9/18/84; site leopard.UUCP
Path: utzoo!watmath!clyde!burl!ulysses!bellcore!leopard!perry
From: perry@leopard.UUCP (Perry S. Kivolowitz)
Newsgroups: net.cse
Subject: pitfalls in computer science education
Message-ID: <149@leopard.UUCP>
Date: Fri, 14-Feb-86 16:33:27 EST
Article-I.D.: leopard.149
Posted: Fri Feb 14 16:33:27 1986
Date-Received: Sat, 15-Feb-86 05:10:58 EST
Distribution: net
Organization: Rest Home For Aged Bit Twiddlers
Lines: 59

First, let me state the premis from which I base my opinions:

	A ``good'' Computer *SCIENCE* Education is INVALUABLE

I doubt anyone would claim that formal training will contribute to
*inferior* computer scientists. So what's the rub?

The problem here is one of terminology. Questions:

What is computer science education.
What is *good* computer science education.
What is computer science and what is its relationship to the average
position needing to be filled in industry.

Computer science is  not  hacking.  It isn't coding. It's understanding.
The science of computing is the understanding of the nature of problems.
The understanding of the nature of algorithms. It deals with the art  of
programming (note difference  between  science  and art) only in as much
that programming is the means by  which  computer science can be applied
to become a tangible benefit.

A computer science  education  teaches thinking. Data processing educat-
ion teaches coding.

A *good* computer science education stresses the fundementals of comput-
ing theory and  practice.  Very  importantly, is should teach underlying
principles in a way which facilitates the  translation of the principles
to new and different areas  and  problems  (a new field of research or a
new work assignment).

The main problem with computer  science education, as  I see it, lies in
inadequate measuring of  the understanding held  by students. CS courses
often test (and therefore  in  the  student's  minds *STRESS*) the wrong
things completely!

Testing (and grading) should be  based  upon critical understanding. The
ideal (pie in the sky in most real world classes) way to measure student
learning (and thus influence  what  will be learned) is to subject every
student to a verbal comprehensive examination or a laboratory or both.

Don't get me wrong. I don't mean  ``make  them write a program.'' I mean
structure an examination which measures true understanding of principles
rather than simple coding or memory tricks.

There are very few computer  *scientists*  out there. Certainly very few
filling most computer  industry jobs.  Incidently,  I have seen very few
doctoral disserations recently  which are worth  the paper their written
on so the failing in  computer science  education  does  not  end at the
undergraduate level.

In summary - given the worlds greatest hacker and a well rounded computer
scientist - and a  large  number  of  arbitrarily  chosen problems, there
should be no doubt  that  the  super  hacker  will  prove inferior to the
scientist. 

*scientist - one who is  adept at  applying  established knowledge in the
solution of new problems or the derivation of new knowledge*

Perry S. Kivolowitz