Until it's off your mind, like Kelly said.

Here's an example of a project I'm currently wrangling. I found out three days ago that I'll be teaching my own class next term. I'm visiting my folks for the holiday from the 23-29, I'm at a conference Jan 3-6, and then classes start on the 9th.

Basically, I need to have the syllabus done this week. So I sat down and made a list of things I knew I needed to do, which contained things like:

-Review syllabus from Fall 2011
-Email colleagues who've taught course for their syllabi
-Review colleagues' syllabi
-Review lesson plans and assignments from Fall 2011
-Skim articles to decide which ones to assign

(Some of these are mini-projects unto themselves--I'll need to download some of the articles I want to skim.)

Once I had a list that let me feel comfortable I had enough stuff down, I set out with the first thing I could think of to do, which was to email my colleagues for their syllabi. But the moment I did that and started "review colleagues' syllabi," I realized that about half my colleagues assign a reader edited by another colleague of ours. So suddenly I had an NA I hadn't planned on:

@Campus: check out reader from library

and

@home: skim reader

So, yes, you'll want to do some project planning that involves collecting some of the NAs and subprojects involved in a larger project. But chances are, you're not going to capture every single NA, nor should you really try. Plan enough to feel comfortable that the NAs you've identified are the next ones needed to move the project forward, and then move the project forward. Like Gardner said, more than a handful of NAs for any one project often either indicates that you need to track it as separate projects, or that your plan is going to get blown to hell, and you'd better be ready for that.