Several of the bløgs I read have pointed to this excellent piece
by the typically excellent Paul Graham about the difference between the "manger's schedule" (which almost the whole world runs on) vs. the "maker's schedule". The article is definitely worth a read because the idea is very well expressed, but I'll boil it down to the essentials here so I can riff on it: "Managers" (and most of the world) run on a schedule broken down into hour long blocks, with frequent changes of activity (meeting A, then some emailing, then meeting B, etc). "Makers" (which Graham uses to mean writers and programmers, and to which I'd add practicing scientists — more below) instead want to work in large, half-day chunks with no interruptions. Meetings are the bread and butter of managers, but they are a disaster for makers because they not only use time, but also break the "flow" of creative thinking they require; a single meeting can destroy a whole aftermoon.
How does this fit into my life? Well, I'm basically transitioning from a maker to a manager now that I'm the PI of my own group. Instead of spending most of my days in 3-4 hour chunks of either coding or being deep in analysis, I am finding most of my days interrupted with at least a couple of meetings, which make it almost impossible to actually accomplish any "making". This realization is important, because I don't want to end up being only a manager of other makers. As scientists get promoted they tend to flow from making to managing, but different individuals succeed to different extents at maintaining a balance. Put another way, I want to keep spending a large fraction of my time with an xterm and an emacs buffer open, rather than my email client and a Word document (ugh).
I obviously have to spend a reasonable amount of time meeting with people, so how do I strike the balance I want? What Graham describes from his younger days, and what I've seen a lot of scientists do is basically work two full days every day: 9-5ish as a manager and then 9PM-2AM (or whatever) coding, analyzing data, or whatever. I'm not quite ready to work that much, so what I've resolved to do (we'll see how it goes) is to actually schedule half-day chunks into my calendar for "making" when I won't be free to accept meetings with anyone else. I'll close my door, not answer the phone, and focus on a project in that time.
Wish me luck!