home *** CD-ROM | disk | FTP | other *** search
- Newsgroups: comp.software-eng
- Path: sparky!uunet!zaphod.mps.ohio-state.edu!wupost!csus.edu!netcom.com!mcgregor
- From: mcgregor@netcom.com (Scott Mcgregor)
- Subject: Re: Software Designs
- Message-ID: <1992Nov19.040908.6796@netcom.com>
- Organization: Netcom - Online Communication Services (408 241-9760 guest)
- References: <1992Nov18.051309.27011@latcs1.lat.oz.au>
- Date: Thu, 19 Nov 1992 04:09:08 GMT
- Lines: 57
-
- In article <1992Nov18.051309.27011@latcs1.lat.oz.au> baragry@latcs1.lat.OZ.AU (Jason M Baragry) writes:
-
- > I would like to be able to analyse some real-life designs and
- >determine some way of representing design knowledge or simply generalize the
- >design so it can be understood independantly of the application domain it is
- >used in.
-
- I sympathize with the desire to have understand things in this way,
- but I would caution Jason and others to be careful for what they wish
- for. For the most part software exists to solve a problem in an
- application domain. Common software horror stories revolve around
- inadequate understanding of the application domain by the developers.
- In such situations, methodologies and techniques that help you do the
- application domain independent stuff, i.e. "doing things right" (as
- opposed to "doing the right things" the application domain specific
- part) just ensure that you skewer your jet plane into the ground more
- surely and swiftly. The needs of the application domain MUST impinge
- and affect the development if the project is to achieve the desired
- results. Analysis of information that strips out this crucial
- dependency may undermine the very assumption of independence which
- would be the basis for any extrapolations or conclusions when you try
- tor apply the results.
-
- I found this same situation in the CoCoMo prediction models. The
- predictive power of the CoCoMo model is based upon the assumption that
- project durations are independent of each other, and that the major
- determinants of duration are organizationally independent: complexity,
- amount of functionality, etc. But in the project databases I have
- looked at this assumption is easy to disprove. In fact, the duration
- of the previous projects done by the individuals, or authorized by the
- management usually has significantly more explanatory power and thus
- yields better predictive estimates. A lot of people don't like the
- fact that this is so. They WANT models that are more dependent on
- things like complexity than on organizational or individual
- history--so they still prefer CoCoMo despite real evidence questioning its
- base assumptions.
-
- This is the same danger that an application domain independent
- analysis of methods may have: it may ignore the most significant
- contributor to success or failure. I don't know that it will, but
- given the CoCoMo case, I would suggest that this be carefully
- scruitinized, and not just assumed.
-
- I'd be very interested in the results whatever turns out.
-
-
-
- --
-
- Scott L. McGregor mcgregor@netcom.com
- President tel: 408-985-1824
- Prescient Software, Inc. fax: 408-985-1936
- 3494 Yuba Avenue
- San Jose, CA 95117-2967
-
-
-
-