Mark McGuiness has a good post about what poetry illuminates about creativity. (The title of the post refers to advertising, but don't let that put you off.)

I get tired of hearing creativity equated simply with idea generation, when that’s often the easiest and least interesting part of the creative process. Shakespeare wasn’t interested in creating ‘original’ plots, but his execution was pretty good - he was so intent on “getting every detail right, getting the structure and rhythm and balance right” that the originality took care of itself.

In my own humble way, I know that when I’ve made a conscious effort to write an original or new kind of poem, the strain shows in the writing - the most interesting things happen when I’m focused on something else, on trying to capture something accurately or tease out the little animating goblin in a word or phrase.

I'm not a poet myself, but I resonate with this. Creativity is constantly associated with the generation of novelty and I've long felt uncomfortable with that. It seems to go with a mindset that suggests it can run to a timetable, and all we need to do is get ourselves sufficiently stimulated and brainstorm.

Trying to be different usually leads to stuff that is rarely engaging. Brainstorming often leaves me feeling strained rather than inspired. So many efforts to support innovation end up feeling like a kind of Victorian potty training.

We use the word "original" to mean novel but what if we think of it as meaning "from the origin" - being true to ourselves rather than trying to be different. In improvisation, the truly inspiring moments are often the connecton of ordinary ideas, where the comedy is not apparent to the actors as they speak. It's not effortful or contrived, but feels natural and spontaneous.

Update: Synchronicity? I just spotted this in Paul Robinson's blog Vagueware:

The problem with innovation in this field, is it tends to not look very interesting at first glance - it might be the smallest of changes in a UI, or a weird library that re-implements something interesting discussed in an academic paper a couple of decades ago.