"Random Walk Programming" is a software development approach that appears to draw inspiration from the successes in research and practice of GeneticAlgorithms, EvolutionaryAlgorithms and GoogleAndPasteProgramming, to enable unqualified (or just lazy) programmers to produce billable results without having to go through all the pain and difficulty of doing real work.
The basic method seems to be for the programmer to follow a procedural algorithm much like the following:
- Given a programming task, search the code, largely at random, until you find some place that seems to be related, more or less, to the functionality you're supposed to be working on.
- Change the code, largely at random, without any real need to think very much about how it might be related to your task, or its impact on the rest of the system.
- Observe the obvious results of your efforts, on user-visible screens and reports. (These are, of course, well known and widely recognized as the only things that really matter. They're all the managers and users can see or care about, anyway. ;-)
- Keep changes that seem to produce better results. "Undo" when the results seem to get worse.
- Repeat.
- Do random simple refactorings of some of the code you stumble across that is clearly and obviously bad. This will show that you're following a widely recognized methodology.
Anyone else
HaveThisPattern? ...or have observed it "in the wild?"
Random Walk Programming