It's a craft

When I had lunch at Stanford Research Institute one day in the late 80s, one of the researchers (whose name escapes me at the moment) argued that the development of expert systems was a craft, not an industrial process.

I believe he was and still is right, but that this applies not only to expert systems but also to the vast majority of today's software development projects.

Now, being a craft does not pose a problem by itself, but for (at least) two reasons we nevertheless suffer from it daily.

The first reason is that neither managers nor developers recognize that they're into a craft and constantly try to handle software development as if it were a predictable process where all activities can be estimated for both cost and time.

The second reason is that there are too few masters of the craft, and that most managers can't tell an apprentice from a master unless the latter one wears a robe and a pointy hat.