Programming and fatigue
I find programming much more tiring than it used to be. To judge from what I read, so do many other programmers. But I'm struck by how many of them think the fatigue is a bad thing and a sign they should change their workflows. I've been perceiving it as good, and even comforting.
My reasons are simple: if I'm mentally fatigued, it suggests that I've been (i) diligently at work (ii) at things that require human brainpower and judgment. This is comforting because it is compatible with an optimistic view of human-AI teamwork: the AI is taking over the repetitive and obvious parts of the job, leaving me to do less repetitive, less obvious, higher-judgment tasks. It's no surprise that these would be more tiring.
To be clear:
- I don't mean to glorify burnout. I do, however, suspect that diligent, professional programming will more often produce a kind of healthy fatigue, and that we might mistake this for impending burnout (because it would have more likely indicated burnout in pre-AI patterns of work). For most of us, the old world is gone.
- I can't prove that this fatigue arises from normal, healthy work instead of some inefficiency I ought to be eliminating. I think that the effort is to productive ends and not, say, a byproduct of frequent context-switching, but for all I know, observers of me might disagree.
- I am in some ways old-fashioned about my work patterns, which often resemble AI-accelerated versions of pre-AI patterns: I am automating what I reasonably can, but am too much of a Gall's Law guy to want to rush into full-blown Software Factory Mode.1
- I'm still finding AI-assisted programming analogous to online poker. I have spent enough time juggling 8 to 12 windows in poker that it doesn't automatically feel silly or disorganized to have many active threads in a programming session. This new kind of programming fatigue feels a lot like the fatigue from a poker session.
- This take is optimistic and conjectural: it entails that I'm still necessary, and more of my time can be spent on relatively skillful tasks. These hypotheses seem reasonable to me, but others might find them delusional rather than merely optimistic.
- Reading through logs and doing certain kinds of research can be taxing, and AI is great at those; conversely, some of the tasks that still fall to me are in fact easy and routine. So, although I do think that AI is disproportionately replacing less-fatiguing work, this isn't obviously true.
So: by all means, please take care of yourself, but don't simply assume that if programming is fatiguing you more than it used to, that means you're doing something wrong.
I'm quite eager to try new things and to use AI heavily, but I find it hard to predict where overriding or guiding the AI is most likely to help. The software-factory approach makes it hard to be flexible about this, so I'm not eager to implement software-factory-ish practices without being very confident they'll improve my work. But elaborating on this would be a different post.↩