Whenever I build something (in an app I use that includes note-taking) that makes certain notes appear automatically (without me having to look for them), I will read them and edit them.
My (tiny) collection of writings on LessWrong, though, is not like this at all. I have to actively look for posts. There's nothing that shows me my writings, no auto-linking from where I'm thinking or writing to the post, the path to doing an edit is very annoying (scroll up, click the 3 dots, click edit, do editing, click save). Even though I was very motivated to make an updating collection of ideas, it just wasn't the right substrate.
(Very concretely, the trigger for "performing a maintenance" is that you read something you wrote and decide that it's dumb, then, you'll traverse the path to doing the edit, even if it's behind multiple actions and not a single hotkey/mouse click, so, *automatically displaying writings/code/ideas is what yields maintenance*)
(There's some holes in this, as you'll find out if you ask an AI to play devil's advocate -- the biggest one is "over-optimization of visible parts")
----
This is what I originally wanted to reply with, the above was a side point but became the main point..
If the problem with maintenance is uneventfulness, you can just make systems where you have "maintenance events", or generally make maintenance more appealing. We have a lot of control over our world and the way it's presented to us.
I don't like how this post is essentially an elaborate description of a problem and that it implicitly says, "maintenance is important, here's why we ought to put more resources into doing it".
How about just making it less difficult lol.
(imo, "use more effort" should be an absolute last resort to "solving" a problem and barely counts as a solution, it's admitting defeat)
Your conclusion is, "I wish my elaboration of the problem was even longer". I think 1 problem description with 1 solution would be more valuable than 10 problem descriptions with 0 solutions.
(I did enjoy the post, I think laying out problems is the first step to solving them, so maybe in that sense this post is super useful, and just incomplete)
----
There's another angle too that seems big and that I don't wanna dive into right now.
Since maintenance is so expensive, or hard, or since decay inevitable, maybe we ought to let things decay more often, plan for it, maybe even design systems that naturally decay in ways that are useful. Maybe we even free up "maintenance energy" by explicitly letting things decay, which we can then spend on things that matter.
(Do you really wanna maintain all friendships? Sometimes it's good enough that it was fun while it lasted and that you still love each other from a distance)
Anecdotally, uneventfulness is not the bottleneck to maintenance, editability and (especially) easy findability is.
(here's why:)
Based on ideas in this Shortform https://www.lesswrong.com/posts/MCBQ5B5TnDn9edAEa/atillayasar-s-shortform?commentId=aETYdQKhaXyAWPEoY and based on my own experiences with a note-taking + note-retrieval app: the main barrier to maintenance might be simply that it's annoying to look up and "edit the entry" (rules, post, idea, code) underlying whatever you're doing.
Whenever I build something (in an app I use that includes note-taking) that makes certain notes appear automatically (without me having to look for them), I will read them and edit them.
My (tiny) collection of writings on LessWrong, though, is not like this at all. I have to actively look for posts. There's nothing that shows me my writings, no auto-linking from where I'm thinking or writing to the post, the path to doing an edit is very annoying (scroll up, click the 3 dots, click edit, do editing, click save). Even though I was very motivated to make an updating collection of ideas, it just wasn't the right substrate.
(Very concretely, the trigger for "performing a maintenance" is that you read something you wrote and decide that it's dumb, then, you'll traverse the path to doing the edit, even if it's behind multiple actions and not a single hotkey/mouse click, so, *automatically displaying writings/code/ideas is what yields maintenance*)
(There's some holes in this, as you'll find out if you ask an AI to play devil's advocate -- the biggest one is "over-optimization of visible parts")
----
This is what I originally wanted to reply with, the above was a side point but became the main point..
If the problem with maintenance is uneventfulness, you can just make systems where you have "maintenance events", or generally make maintenance more appealing. We have a lot of control over our world and the way it's presented to us.
I don't like how this post is essentially an elaborate description of a problem and that it implicitly says, "maintenance is important, here's why we ought to put more resources into doing it".
How about just making it less difficult lol.
(imo, "use more effort" should be an absolute last resort to "solving" a problem and barely counts as a solution, it's admitting defeat)
Your conclusion is, "I wish my elaboration of the problem was even longer". I think 1 problem description with 1 solution would be more valuable than 10 problem descriptions with 0 solutions.
(I did enjoy the post, I think laying out problems is the first step to solving them, so maybe in that sense this post is super useful, and just incomplete)
----
There's another angle too that seems big and that I don't wanna dive into right now.
Since maintenance is so expensive, or hard, or since decay inevitable, maybe we ought to let things decay more often, plan for it, maybe even design systems that naturally decay in ways that are useful. Maybe we even free up "maintenance energy" by explicitly letting things decay, which we can then spend on things that matter.
(Do you really wanna maintain all friendships? Sometimes it's good enough that it was fun while it lasted and that you still love each other from a distance)