Return
Everyone has a drawer.
Not necessarily a physical one, though physical is what I keep imagining. The bottom drawer in a desk that sticks when you pull it. Inside: a notebook with forty pages of handwriting that stops mid-paragraph. A folder of printed research. Notes in a margin that say “expand this” in handwriting you barely recognize as yours.
You know what’s in the drawer. You think about it the way you think about a friend you owe a call. Not constantly. But enough that it has weight. Enough that when someone asks what you’re working on, the drawer enters your mind before the honest answer does.
The notebook is a novel you started in 2019. Or a dissertation chapter. Or a business plan. Or an algorithms textbook you were writing, or reading, or working through — whatever the verb was when you were still the person doing it. The specific content doesn’t matter. What matters is the gap between the last entry and today, and what that gap has become.
The gap has its own physics.
The longer something sits untouched, the harder it is to pick up. Not linearly — exponentially. A project abandoned for a week costs you ten minutes of reorientation. A project abandoned for a month costs a full day. A project abandoned for three years costs something that can’t be measured in time at all, because the barrier isn’t confusion. It’s confrontation.
You have to look at your own work from a distance you didn’t choose. And the distance reveals everything. The sentences you thought were good. The architecture you thought was clever. The ambition that now looks like — what? Naivety? Arrogance? The gap between what you were trying to do and what you were capable of, preserved in amber, impossible to unsee.
A repo with commits from three years ago is a fossil record of a person who doesn’t exist anymore. The code they wrote reflects assumptions they held, skills they had, problems they understood. Opening that repo is an act of archaeology performed on yourself. Most people would rather start over. Starting over lets you be the person you are now. Resuming forces you to sit with the person you were.
Here’s what the re-entry actually looks like.
You open the file. The first thing you see is formatting that irritates you. You’d lay it out differently now. Then the prose, or the code, or the notes — and they’re not bad, exactly, but they’re alien. Written by someone who shared your name and your general intentions but processed the world differently enough that their sentences feel borrowed. You start reading with the intention of finding where you left off. Instead you find yourself editing. Rewriting. Redoing work that was already done, because the person who did it isn’t you anymore and your instinct is to make it yours before you can move forward.
Three hours later you’ve rewritten the introduction and made no progress on the actual gap. The activation energy didn’t decrease. You just spent it on the wrong reaction.
This is the pattern. Not inability. Misallocation. The cost of return isn’t ignorance of where you left off. It’s the compulsive need to re-establish ownership of work that your past self produced. You can’t continue someone else’s project, even when the someone else was you.
AI changes the cost structure. Not in the vague way. In a specific way.
The machine reads your old code so you don’t have to. It traces logic you wrote three years ago and reports back without judgment. It has no opinion about the gap between your ambition and your execution. It reconstructs context, tells you where things stand, and asks what you want to do next. The archaeological dig happens, but you’re not the one with the trowel.
The shame component of the activation energy drops to near zero. You’re not looking at evidence of who you used to be. You’re looking at a summary, prepared by something that doesn’t know or care who you used to be.
This isn’t new in kind — editors and coauthors have always served this function. A good collaborator can look at your abandoned manuscript and say “here’s where you were, here’s what’s missing, let’s pick it up.” What’s new is the cost. That collaborator used to be a person you had to find, trust, explain the project to, and pay. Now it’s free, or close enough that the decision to resume is no longer gated by a resource question. It’s gated only by whether you actually want to finish.
And here is where the drawer reveals what it actually is.
An unfinished novel is potential. It’s the book you could write if you had time, if circumstances were different, if you could just get back to it. The incompleteness protects you. As long as it’s unfinished, it can’t be judged as a finished work. It exists in a superposition of quality — potentially brilliant, potentially terrible, never collapsed into either.
Finishing is collapse. A finished novel is a specific novel. It has specific flaws. People can read it and say no. The drawer isn’t just where you put things you abandoned. It’s where you put things you’re protecting from completion.
When AI drops the cost of resumption to near zero, it removes the excuse. Not the fear. The excuse. “I’d finish it, but I’ve been away too long, I’d have to start over, I don’t remember where I was.” Gone. The machine remembers. The machine can reconstruct context instantly. The only question left is the one the drawer was hiding: do you want this to be finished?
The drawer was the architecture of avoidance, and it was load-bearing.
I have a drawer.
I keep writing around it. The algorithms textbook, the anonymous novelist at chapter twelve, the hypothetical projects in hypothetical desks. Analysis from a safe distance. The essay does the thing the essay describes.
What’s in my drawer is loops. Versions of the same problem attempted over and over, each one a little different, each one failing in a new way. Code I wrote that worked until it didn’t. Systems I built that were going to be the one that lasted. Notes that say “this time” in handwriting I recognize too well.
The gap between the last attempt and today isn’t long enough to feel archaeological. It’s recent enough to sting. And the thing I’m protecting from completion isn’t a book or a chapter or a business plan. It’s the answer to whether the pattern breaks. As long as I haven’t finished, I haven’t failed for the last time. The incompleteness is the only architecture that keeps the possibility alive.
So I know what the drawer is for. I know it’s load-bearing. I know that when someone offers to flatten the hill, the hill was never the problem.
This is what cheap return actually reveals. Not that people were lazy or lacked discipline. That the obstacle was structural, psychological, and often wanted.
Some people will discover their drawer held real work interrupted by real friction. They’ll pick it up and finish and it will be unremarkable in the best sense — a thing completed, finally, because the logistics stopped preventing it. The ten-year gap was a logistics problem, and now the logistics are solved.
Some people will discover their drawer was a room they built around something they couldn’t face completing. They’ll open it, see the excuse removed, and understand for the first time that the excuse was the point. They won’t finish. But they’ll know why, which is its own kind of progress — harder than completion, and worth more.
What I keep coming back to is the notebook with handwriting you barely recognize. The self that started the thing. The specific way they formed their letters, the specific references they made, the specific ambition they carried that led them to open a blank page and begin.
That person is irretrievable. Not because they’re dead — because you’re different. You read differently now. You know things they didn’t. You’ve failed in ways they hadn’t yet. The project they started was shaped by who they were, and you can’t resume it without acknowledging that you are not them and they are not you and the work belongs to both of you and neither.
AI can bridge the technical gap. It can reconstruct context and identify patterns and tell you what comes next. What it can’t do is resolve the strangeness of picking up a sentence someone else started and realizing the someone else was you, years ago.
The drawer is open now. The hill is flat. What remains is the weight you put there yourself, and the question of whether it was ballast or foundation. Nobody can sort that for you. Not even a machine that reads everything and judges nothing.