According to Tim Menzies, David Owen, and Julian Richardson
The physicist John Archibald Wheeler advised that “in any field, find the strangest thing and explore it.” Accordingly, we explore the strangest thing about software—that it ever works at all.
Modern software is so complex that it should never work. […] Any software assessment or verification and validation process must negotiate this complexity by effectively covering only a fraction of the software’s possible internal configurations. So why does software work? Why aren’t our incomplete test methods missing all too many critical errors?
We propose that software works because internally it is surprisingly simple. This proposal is based on recent results from artificial intelligence research. AI has discovered certain previously unrecognized regularities that developers can use to quickly find solutions. A careful reading of the software engineering literature shows that these regularities have also been seen in conventional software. For problems with those regularities, much of what we can find via complex and costly methods, we can also find by random search.
A cliched interview question for an academic or research position is
What’s the most important problem in your field?
The game is to hope that the applicant’s response naively leaves an opening for a cliched follow-up question
Why aren’t you working on it?
It’s hard to imagine anyone still falling for that, yet I’ve never seen a top-ten list of clever responses. How would Oscar Wilde have answered?
Those questions actually are powerful ones to seriously ask yourself, as long as you get straight first on your definitions. Important in what way to whom?
More blog entries on Computing.