I think, in ten years or so, software development will be done in the way project management is done now: as a marshaling of resources rather than the craftwork that currently defines it. The resources will mostly be machine intelligence, bent to the task of mapping fuzzily-defined specifications into executable code. There’ll be a few old crutzers still building things by hand, but such bespoke work will be expensive, and valued more for the aesthetic appeal than the practical benefit.

Ian’s skill set will be much more relevant than mine at that point.