Ink Blot Software Engineering

The belief that software engineering is mostly a psychological endeavor to understand the specific culture or individual needs of customers and/or developers. It is in contrast to the One Right Way belief that there is a single canonized "best practices" to be "discovered" like a math theorem.

Rejected are ideas like ExtremeProgramming or ParadigmX are the only or best ways to achieve results for everybody in every company. Note that trying something is not excluded, just the extrapolation of your own success to everybody else.

Smells like something out of California :-)

As with many ideas expressed in their most extreme form, this is easy to parody and ridicule but contains an element of truth. We have modified the original vision of PairProgramming because it simply doesn't work with me. I am a convert to ExtremeProgramming now, but in a modified form necessitated by psychological differences. See TheWorldIsaVeryBigPlace for more comments.

I don't believe that there is One Right Way, a single canonized "best practices" to be "discovered" like a math theorem. Nor, however, do I believe that everyone should just "do their own thing". The truth is in the balance, as we are taught from the GameOfGo.

But the GameOfGo has one clear-cut metric that software engineering lacks: One either wins, loses, or draws (ties). Thus, it has an objective final metric. In software design, beyond mere producing the right results, there is nothing so stark. One man's elegance is another's pasta mess where each party ends up saying, "if you just were able to think properly like me, then you would like it". Software metrics are stuck in psychological tar.

I don't know about that. The clear-cut metrics are whether the code works and whether it is delivered on time. And the methodology of Go is infamous for being too difficult to formalize (otherwise we'd have good Go-plaing software). So actually they seem pretty similar.

"Works and on-time" ignores longer-term maintenace issues. Few claim that their favorite paradigm delivers faster (except maybe FP). Most brag about having code that is easier to change.

CategoryHumanFactors