On Thu, May 8, 1997 at 11:02:36 (-0700) Louis Proyect writes: >This is the sort of systems development projects that I have been working >on for 28 years and they present a completely different set of problems >than creating shrink-wrapped software like a word-processor. The >difficulties cut to the heart of the contradictions of capitalism and have >little to do with psychology or whether such activity is more like >novel-writing than engineering. > >The problem is simply that there are powerful impulses in any capitalist >institution--including government agencies and universities--to set >completely arbitrary deadlines based on incomplete specifications. >Management structures in corporations encourage forced marches and I have >been on many. The question is, why did this (lack of predicted productivity gains) happen in software? I was trying to answer that specific question, and trying to refute what I saw as a vastly over-abstract idea from O'Conner. >Mostly, this type of procedure is not followed. Projects are started with >incomplete specifications. They are understaffed and target-dates are >completely arbitrary. This tendency existed throughout the 60s and 70s and >has only accelerated in the late 80s and 90s when corporations decide to >take shortcuts in the face of competition. And, just as likely, a fully-specified project will fall flat on its face because half-way through it, you realize that what you are creating is in fact not what you want. I think there are simple issues here with the very nature of the problem itself, rather than an idiotic Dilbert-esque system (common enough). Such problems will be seen, I contend, even in the most democratic (socialist) of systems---I guess that was one of my points. Also, the late 80s and 90s were a time of increased competition which caused companies to skimp on requirements? If anything, I've noticed a vast increase in the amount of requirements in the past 10 years, but my knowledge of these trends is certainly provincial. >One of the reasons socialism is such an appealing idea is that procedures >will be common across organizational lines. This will allow software to be >shared. If there was a single American university system, admissions >procedures would be simplified. Instead of having 10,000 computer >programmers working on separate systems at Columbia, Yale, U. of Cal, etc., >there would be 500 or so developing software for the entire institution. >The same thing would of course be true for financial and manufacturing >institutions. One bank, one automobile company, one steel company, etc., >all owned by the people. No advertising, no public relations, no separate >health and savings plans. I have a book at home called "The Waste of >Capitalism" or something like that which details this stuff. I should try >to read it and report on it. You should, it sounds quite interesting. The thing I find appealing about socialism/democracy is that one need not give up a fundamental human right in order to live. We need more practical models to point to and say---see, that's how it could be done! Very appealing to computer geeks like me!:-) I suggest those interested in this should take a look at what the Free Software Foundation has done. Coincidentally, our shop (Dejanews) is almost 100% run on "free" software (Linux operating system, FSF/GNU tools, etc.). The amount of sharing of ideas across firms is probably quite low, but free software does allow some of this to occur. That's also one thing notable about Usenet newsgroups---they allow a great deal of software development to be done by a widely-dispersed set of people who, more or less, get an equal say in how things are done. Bill
