There is a sentence I have been putting at the top of source files for about five years:
Around here, however, we don't look backwards for very long. We keep moving forward, opening up new doors and doing new things, because we're curious, and curiosity keeps leading us down new paths.
That is Walt Disney. The wording above is the longer version that closes Meet the Robinsons, where it appears on screen with his name under it. I have quoted it for five years and it belongs to him, which is worth stating plainly at the top rather than leaving a reader to assume the sentence is mine.
Not in a README, where people look. In the source, where they mostly do not. It was a note for whoever got curious enough to open the file, which is a small enough audience that leaving it there felt less like a statement and more like a message in a bottle.
I want to write about what it did to me, because it did a great deal, and not all of it was good.
What it gave me
I found it not long after I started, which was March 2020, ten years old, in a flat I could not leave.
Nobody tells you the actual difficulty when you begin. You expect it to be that the material is hard, and it is, but that is not what stops people. What stops people is permission. There is an infinite amount to learn and no obvious order to learn it in, and everybody older than you has a curriculum they would like to hand over, and every one of those curricula begins by telling you that the thing you are actually interested in is not the right place to start. Learn the fundamentals first. Do it properly. Come back to that later.
That sentence is permission. It says the reason to open the next door is that you are curious about what is behind it. Not that it fits a plan, not that it is the correct next step, not that you have earned it. Curiosity is treated as sufficient justification on its own, which when you are ten and have no other credentials to offer is the only justification available.
Almost everything I know arrived through that door and not the other one. I taught myself by building things that people then used, rather than by finishing a course, and the reason was never discipline. It was that the building was the interesting bit and the course was not. This year I got curious about why large models seem to gain abilities in jumps, and that question walked me backwards out of machine learning entirely and into a physics paper from 1972, which is not a route any syllabus would have sent me down. I am reading a clinical neurology textbook at the moment, next to an artificial intelligence one, for no better reason than that I want to know how much of the metaphor is real.
None of that was planned. All of it came from treating the next interesting thing as reason enough.
What it cost
Here is the part that took me longer to see.
Curiosity is superb at starting and completely indifferent to finishing. A new door is always more interesting than the room you are standing in, always, because the room you are standing in is known and the door is not. That is the entire mechanism, and it does not come with a stopping rule. Nothing in the sentence tells you when to turn round.
So I started things and kept them running well past the point where I should have shut them, because shutting something requires going back and looking hard at it, and going back is exactly what the instruction discourages. I left work half-finished in favour of work that was newer. I moved on from mistakes quickly, which felt like resilience and was sometimes just avoidance wearing better clothes.
And then the version of it that actually mattered. Over a few years the way I described my own work drifted from what was true. Not one lie I could point at and date, which would have been easier to catch. It went the ordinary way: a phrase that was slightly generous, repeated, then repeated somewhere firmer, then written down, then written down somewhere other things linked to. Each individual step was small enough to be invisible while I was moving.
No amount of forward motion would ever have found that. Momentum is precisely the wrong instrument. The drift was sitting in things I had already said, in places I had stopped reading, and the only thing that surfaced it was going back and reading them in a row, which took days and which felt, the whole time, like the sort of backward-looking the sentence warns you off.
That is the asymmetry I would want to hand to anyone who takes this line as seriously as I did. Failure gets smaller as you walk away from it. A claim that was never quite right gets larger, because you keep walking and it comes with you, and every place it is repeated becomes another place it has to be corrected from. Moving on does not resolve a false statement. It distributes it.
The half I had been skipping
When I finally read the sentence properly, the thing I had been getting wrong was not in it. It was in what I had added.
It does not say never look backwards. It says not for very long. That bounds the duration; it does not forbid the act. Those are entirely different instructions, and the difference is the whole useful content of the line. I had been reading a prohibition into a limit for five years.
The man who wrote it makes the point better than I can. Having said it, he then built a park, and put at the entrance a street modelled on the small town he had lived in as a boy, so that every single visitor walks the length of a rebuilt past before reaching anything new. He looked backwards constantly. He was rigorous about it. He simply did not stay, and he did not let it become the destination.
It took me an embarrassingly long time to notice that the person I had been quoting was not doing what I thought he was telling me to do.
Why it stays in the file
I am keeping it, and I am keeping it exactly where it was.
A quotation at the top of a profile is a claim about yourself that you have not done anything to earn. This is the kind of person I am. It costs nothing to write and cannot be checked, which puts it in the same family as everything else I spent this year removing, only more socially acceptable. I would rather the work said it.
At the top of a source file it is a different object. Nobody arrives there by accident. The only person who reads it is someone who went looking, which is to say someone curious, which is to say the exact person the sentence is about. It is not addressed to an audience. It is addressed to whoever turns out to be like that.
So it stays, and I will keep putting it there, with one thing added that I had to learn the slow way: go back regularly, go back thoroughly, and then leave.