When I was a teenager, I was a big fan of the show (and books),
Red Dwarf. One part of the story describes one of the characters, Rimmer, preparing for an exam. He spends so much time preparing his revision timetable and improving it iteratively, over and over again, that he completely forgets to revise and the night before the exam realises he has a
beautiful revision timetable, but no time left to revise. "Rimmer's Revision Timetable" was a shorthand in several places I worked for "we're doing too much preparing, not enough doing, let's get on and just do".
I once spent 9 months trying to figure out how a piece of software should be built - I wanted it to be perfect - while my bosses agonised over the delay. One night, in a bout of anxiety and just trying to get something to show my bosses, I sat down at around 7pm and just started riffing away in the editor. I ended up working through the night. By 5am it was done. By 9am it was in production. By 11am the first customers were giving us money. I didn't need to edit it again for 3 months. That was the moment I decided to ignore an old quote I'd been handed and had until the point believed had real truth in it:
"Give me six hours to chop down a tree, I'll spend the first three sharpening the axe" - attributed widely to Abraham Lincoln, but as the other saying goes "90% of the quotes you read on the Internet are misattributed to the wrong person" - Mark Twain.
There are some situations where the preparation is necessary. If your axe is blunt, it'll never cut down the tree, even in 6 hours. Getting it sharp in the first 3 hours means you might finish the task early, for less physical effort.
But more often than not, you're just procrastinating. You could go find another axe. You could find that the axe you have is just fine and gets your tree down in an hour rather than three hours of sharpening followed by half an hour of chopping. But by spending your time sharpening, and not chopping, you have the illusion of telling yourself you're working towards your goal, but you're not. And guess what, sharpening is a lot easier and a lot less sweaty than chopping, so it's more comfortable too - what's not to like?
YAGNI, Yak Shaving, axe sharpening, Rimmer's Revision Timetable, they all bring into focus different perspectives of the same problem:
Doing a thing that isn't the thing, isn't doing the thing.
Yes, if your editor config is distracting and you want to fix something, go find a quick thing and move on. Installing a tiling window manager and a global keyboard shortcut app of some sort might make things flow a little better, but spending days optimising is likely poor ROI.
Just try and do the thing. Build something. Stop planning to build something. If you aren't sure what to build, spend 5 minutes trying to build something and figure out what's actually in your way (hint: it's not your editor, it's not your lack of gdb knowledge, it's not that you're not using the hottest new programming language or that new agent/skill/model people are talking about), and find the possible quickest way to get back to making progress.
Yes, you need to think about prod. The more senior you are, the easier this comes naturally, but writing tests is doing the thing, not procrastination. Putting in operational and business metrics is doing the thing, not procrastination. Why? Because those are part of the thing. Your prompts are not part of the thing, and neither is your tmux config.
As an aside, LLMs can maybe help here (but don't over-engineer the harness): I have a side project where I feel I need to pick up a lot of theory on reinforcement learning, I could do with a refresher on partial differential equations, and I also want to learn some more advanced socket programming in C (trust me, all these things combine into a single thing).
I could have spent the next three months reading books and doing courses and then started on my project, but what I've done instead is get Fable to build out a curriculum where I build the relevant parts as I'm learning.
The first spec it came back with told me I had about 7-10 months of work ahead of me, but I could see (at least in the early stages which I understand better than the later stages), I'd be in a virtuous cycle of learning → applying → learning → applying → ... and within the first couple of weeks I'd have some basic harnesses that showed progress and have some of the theory down too.
This, I think, is a valuable way to structure the learning without procrastination. A personalised tutor who can help you fill in the gaps in your knowledge as you build, so you build the right thing and learn, but you're not just procrastinating up-front, feels good. And, if I get 3 months in and have enough knowledge to realise the next three months are not quite right, I will have the knowledge to re-prompt, re-draft and move forward.