John Roebling's bridge is still standing and still functioning despite problems with the quality of the supplied components and the unknown stability of some of its foundations.
How long does our software last?
![[many connected ideas]](map.gif)
I think that the rush to implementation has applied to attempts to develop CASE. It also applies to many proposed methods - programs executed by humans. The methods I have studied offer, at best, a good solution to a particular class of problems in a particular context. However they are sold as solving all problems in all situations(Silver Bullets).
I have a hunch that different socio-economic niches make different methods, processes, techniques and technologies rational: "Does One Size Fit All?"(1995) . I have been working out the consequences of adopting different methods and processes in different situations. I am supervising a survey of software development companies to see if their socio-economic situation is having an effect on their methods and processes (see theses).
I am therefore no longer designing and selling my favorite tools and methods. Instead I am collecting different software development tools, processes and methods. I am analyzing as many published methods and methodologies as possible. I am using the operations research paradigm developed by scientists and practitioners in the second world war. Here one develops mathematical models of how some process or method is carried out. The model can be used to make predictions and to predict optimal (or at least feasible) methods, processes,and/or techniques.
I have been collecting reports of what is done in practice from magazines. Similarly I monitor the state of the art as discussed on several Usenet Newsgroups. I therefore maintain several directories pointing to material on developing software development under the following headings:
This was an experiment to see if it was feasible to develop an ASCII based documentation language that would allow a software developer to mix mathematical and natural notations together at all stages in a project, for all life cycles, and all methods. The notation grows out of the well known EBNF notation for syntax but goes much further. In particular it allows parts of one document to be reused and extended in other documents.
One test was to develop a tool to translate the notation into HTML. This turned out to be useful for documenting languages and preparing WWW notes and handouts.
The above are part of a collection of samples of documentation and code based on my MATHS notation.