I read a piece today on usability of social interfaces on joelonsoftware.com — Whereas the goal of user interface design is to help the user succeed, the goal of social interface design is to help the society succeed, even if it means one user has to fail.
Joel talks about the design of social networking sites, including a note that Defensive Design is a naive approach to site design. Defensive Design is the design of software that acts like a good citizen - pointing out mistakes that a user makes & suggesting how they can correct their error in the future. He talks about how this type of software design is too easily taken advantage of my freeloaders, scammers, and other miscreants and should be avoided. Overly polite social software leads to abuse and general lowering of the usefulness across all users.
This reminded me of something that happened today at work - I was speaking with a fellow engineer about moving our QA process from a wide but not deep exhaustive (and exhausting) testing approach - every field of every new object is tested (either manually or in an automated fashion) for default values to a wide & deep probabilistic testing approach - we should check not only across the breadth of each component of the system, but also (randomly, perhaps) some number of permutations of values in different fields. The question arises - should the tests that we develop while developing software be deterministic systems - executing the same code-path each time they’re run?
At the moment, I’d argue that it would be more useful to develop test suites that test not only the usual expected range, min & max values for functions and attributes in a system; but also permutatious of values — outputting copious amounts of data that can be easily filtered to help determine the state of the system when breakdowns occur. This is a fairly extreme departure (say, a 180 degree switch) from my previous instance that tests be reproducible, deterministic, self-documenting, and obvious.
I also spent an hour or so after work talking to one of my good friends who’s a statistician. He always gets me thinking so probabilistically.