Xref: utzoo comp.os.vms:38377 comp.object:3386
Path: utzoo!utgpu!news-server.csri.toronto.edu!bonnie.concordia.ca!uunet!odi!dlw
From: dlw@odi.com (Dan Weinreb)
Newsgroups: comp.os.vms,comp.object
Subject: Re: Software Costs vs Results
Message-ID: <1991Apr28.041052.1269@odi.com>
Date: 28 Apr 91 04:10:52 GMT
References: <1991Apr22.191804.3454@mits.com.au>
Reply-To: dlw@odi.com
Organization: Object Design, Inc.
Lines: 21
In-Reply-To: alistair@mits.com.au's message of 23 Apr 91 00:18:04 GMT

In article <1991Apr22.191804.3454@mits.com.au> alistair@mits.com.au writes:

   U.S. Government Accounting Office 1979 report (FGMSD-80-4)
   describes this breakdown of results from $6.8 million in nine federal
   software projects: 47% ($3.2M) were paid for but never delivered, 29%
   ($2.0M) were delivered but never used, 19% ($1.3M) were abandoned or
   re-worked, 3% (%0.2M) were used after change, and only <2% ($0.1M)
   were used as delivered."

Please note that "federal" software projects are not typical of
private-sector software projects.  The more you learn about how
procurement and contracting work in the Federal Government, the less
surprising figures like the above seem.  Some of this kind of thing is
actually intentional (two companies are given the same task, and only
the "winner" is chosen, the results of the other being discarded),
some due to process inefficiency (work is done as part of a greater
project, that is later cancelled when Congress changes its
priorities), some is inefficiency, probably a bit is out-and-out
fraud, perhaps some is pork-barrel, and so on.  If we looked
at Federal military procurement as a whole, I wonder how different
it would be?