From mboxrd@z Thu Jan  1 00:00:00 1970
X-Spam-Checker-Version: SpamAssassin 3.4.4 (2020-01-24) on polar.synack.me
X-Spam-Level: 
X-Spam-Status: No, score=0.7 required=5.0 tests=BAYES_00,INVALID_DATE,
	REPLYTO_WITHOUT_TO_CC autolearn=no autolearn_force=no version=3.4.4
Path: utzoo!attcan!uunet!tut.cis.ohio-state.edu!ucbvax!MBUNIX.MITRE.ORG!munck
From: munck@MBUNIX.MITRE.ORG (Bob Munck)
Newsgroups: comp.lang.ada
Subject: O-O and general S/W Metrics
Message-ID: <27053.618762611@mbunix>
Date: 10 Aug 89 14:30:11 GMT
Sender: daemon@ucbvax.BERKELEY.EDU
Reply-To: munck@MITRE.org
Organization: The Internet
List-Id: <comp.lang.ada>

Boy, I've got to agree with Ralph Johnson (johnson@p.cs.uiucf.edu):

   I am usually suspicious of software metrics.  In general, we do not
   know what to measure of software.  We really need to measure the
   IDEAS in software, not the lines of code, though nobody knows how to
   do that.  

It is, I believe, possible to ASSESS the quality of software by careful
and laborious reading and studying of it.  A very good programmer with
extensive experience in long-term, multiple-person projects can, I
believe, predict quite accurately the additional time needed to put a
piece of software into production use and the relative cost of
maintaining and enhancing it.  The entire field of software metrics
seems to me to be an attempt to discover mechanical short-cuts to do
this, and I've never seen one that convinced me it could.  Likewise, the
current fad of O-O* seems to me a similar attempt at short-cutting an
intrinsically difficult job.

A favorite book of mine, "Zen and the Art of Motorcycle Maintenance,"
makes the point that quality is not quantitative; we can recognize it,
but we cannot completely characterize it.  The little rules we make up,
and then measure with metric tools, are never absolutely true.  Given
such a tool, I can always write a bad program that it will rate as good,
and usually a good program that it will rate as bad.

Finally, I've never yet seen an attempt to validate a software metric
under "real world" conditions. That is, I've never seen a comparison of
the output of a metric tool to the actual life-cycle costs of the code
it was applied to.  Likewise, I've never seen a demonstration that O-O
(or Ada, for that matter) reduces life-cycle costs.  The closest thing
I've seen was Larry Weissman's graduate work at U of T two decades ago
(not the gerbil study).
                                     -- Bob Munck

* When I read "O-O" or "OO" (for "object oriented"), I always hear Fred
Gwynn (not sure of the name) in "Car Fifty-four, Where Are You" saying
his famous "Ooh! Ooh!"  It seems to fit.