I felt this a couple of days ago when I couldn't intuitively see something that I thought I should. Luckily, after a while, I tried reducing my vague confusion

I felt this a couple of days ago when I couldn't intuitively see something that I thought I should. Luckily, after a while, I tried reducing my vague confusion into specific questions, and answering each, one by one... which revealed more to the thing than I thought there was. No wonder I couldn't see it - that concept didn't exist. I also simplified the problem, discarding issues of implementation efficiency and convenience/ease of use of the user model, and any tricky wrinkles that complicated it, to reduce it to something I could see. Being able to see that has given me a base to understand the other issues. I've noticed that some specifications use this approach: they have a very detailed underlying conceptual model, then add a convenience model on top. e.g. XPath specification (crazy ancestor, descendant, sibling etc axis -> defaults to descendant so you can just use unix-like paths). I used to be much better at seeing things... but thinking back, my key characteristic at that time was to embrace not knowing things, embrace mistakes and learn. I was also working on areas that were built directly on top of existing models and approaches... whereas now, I'm developing the theoretical foundation as well. There's a kind of exponential explosion, with so many different and interacting choices at every turn. In some ways, an intuitive/insight approach is necessary, as exhaustively trying each alternative is not feasible. tangent: In Breakfast at Tiffany's (the movie, not the story), there's some discussion over whether Audrey Hepburn's character is a "phony", but I never understood what they meant in that context. She's beautiful - how can that be phony? Or maybe that she's pretending to be urbane when she's rural?