Claude has enabled a sort of exuberant play recently, as I’ve figured out what it is good for and how it can help me do stuff that I’ve thought would be interesting or fun to play with, but not practical to implement on my own. That’s been fun, but also it has felt a little weird because it has led me fully into the territory of doing things because I can, not because they’re really useful.
Several years ago I read Cal Newport’s Digital Minimalism, going into the book thinking it would be one thing and learning it was something else I was much happier to think about, which was “why am I doing any of the things I do?” I spent a lot of time inventorying, thinking, and writing about that question, figuring out what my purpose was.
Some of that was painful to do. A few weeks ago Al asked me for some advice on a problem she was dealing with at work, and my opinions on the matter were so reflexive and so anchored on what is turning into a big portion of my life thinking about management problems that I felt a little dizzy and had to just stop and reflect on where that all even came from.
“I wish,” I said, “that I knew less about solving management problems than I seem to know. I could have been doing something else with all that bandwidth.”
So “what is my purpose? Why do I do these things?” were hard to ask because I had to spend a lot of time thinking about what the answers might have once been, and consider what I thought of being pulled away from that.
But from the exercise came some clear answers about what I do: I write, I take pictures, I read. Almost everything I screw around with, fiddle with, experiment with, or futz with has something to do with one of those things and some idea I have about how it will make one of those things better somehow. My reasons for doing those things are going to remain my own. I could share the ones that don’t feel super personal to me, but they’re the kinds of reasons you probably have yourself if you do any of those things much.
I’ve made a sort of mantra out of that list.
My rabbit holes are adjacent to my interests most of the time, but their utility is never completely guaranteed: Sometimes parts of my personality come to the fore that make my rabbit holes and digressions feel wasteful. They don’t pan out, or they do and they aren’t as useful as I hoped, or they triggered a bunch of fussing around with something that was fine as is. When I’m done, I feel a little dulled and uneasy, and I think “I could’ve been writing, or reading, or taking pictures.”
When I’m feeling very reflective I feel pretty self-conscious about those time-sucks. They feel like failures. I try not to be too hard on myself, because I’ve also come to realize there is probably something going on it’d be good to try to tease out. Like, I can tell you exactly why I haven’t been reading or writing much lately, and it has been helpful to know that and remember it.
So back to Claude: It has powered a sort of “oh, that’d be cool” digression factory that is always in sight of what I care about, but is always not doing the things I care about. It got really bad with one small project that spun out until I’d burned a whole evening on it. The time was just gone. And I wish I’d noticed for some reason besides “this didn’t pan out, why did I spend this time on it?”
So today I made the “focus guardian” prompt and stuck it in Claude’s user instructions. I think people tend to use that for facts about themselves, or the tone they prefer in responses, or whatever. I decided to use it to get a little help with my digressions. It’s a little lengthy, and I’m saying enough about the inside of my own head to even discuss this in public, so I’ll just outline what it does in the 350 words it takes:
- Looks for red flag expressions and requests
- Intervenes with a mix of questions and principles
- Suggests some alternatives
It caught me this afternoon, asked an intervening question, and provided some suggestions. I folded my laptop shut for a few minutes and wondered what I had been about to do, and decided I might as well enforce this new protocol I’m trying to implement in my own skull by writing about it.