Software Writers' Workshops as Scientific Methodology

pieces of literature. A pattern is a configuration or solution to a problem in the design and/or implementation of software that is repeatedly found, though almost always in slightly different form depending on the circumstances. A pattern then is not a programming language construct or an idiom, though in some cases it might be possible to generalize a pattern that way. Many software systems are put together from a set of patterns in a structured way. In this case a pattern language might be in play. A pattern language is a set of patterns that can be used with each other in a particular way or set of ways to achieve a larger system. Patterns and pattern languages arise in the world because software designers and implementers tend to solve the same problems in the same ways—sometimes because developers communicate best practices with each other, sometimes because they see each others’ solutions and adapt them and improve on them, and sometimes because the best solutions are (somehow) cosmically attractive. The result is an artificial world of naturally occurring software that is filled with expressed pattern languages of all sorts, and it becomes an almost scientific endeavor to identify and describe them. Perhaps it’s not science in the most proper sense, but it is a science-like endeavor in that the world of software is being described by a set of pattern languages; these pattern languages form the model or theory that is the basis of this science. The endeavor then is to discover those pattern languages and to describe them, and that’s where the literature comes in. The software patterns community has decided that natural language is the best medium for describing discovered pattern languages. The structure of the natural language descriptions typically follow a format, so there is a hint of formality in this endeavor, but human language is the basis. Natural sciences tend to use either mathematics or other formal languages as the basis of describing their models and the world. That’s because the most important thing for these sciences is to unambiguously describe their knowledge. For the patterns community, the most important thing is to be able to pass on knowledge to practitioners—to software designers and implementers—who might not be Richard P. Gabriel