Organize saves by intent, not by topic

Everybody builds the same folders. Food. Fitness. Travel. Tech. Home. Maybe Funny, for the ones that do not fit. It is the obvious structure, it takes about four minutes to set up, and within a couple of months it has stopped doing anything. The folders are still there, still correctly labelled, still full — and you still cannot find anything, and you still are not going back to any of it.

The usual diagnosis is that the categories were not fine-grained enough, so people split Food into Recipes and Restaurants and Meal Prep and the same thing happens again a month later. The problem is not granularity. It is that topic is not the question you have at retrieval time. You do not open a saved folder wondering what your videos are about. You open it because you are standing in a supermarket, or you have twenty free minutes, or a friend is moving flat and you know you saw something. Those are questions about purpose, and a topic folder has no answer to them.

Two axes, only one of them gets folders

Every save sits on two independent axes. Topic is a property of the content: this reel is about pasta, whether you saved it or someone else did, and it will still be about pasta in five years. Intent is a property of you at the moment you tapped save: what you meant to do with it. The same pasta reel could be a thing you intend to cook this weekend, a technique you want to learn properly, or something you saved purely to send to your brother, who is smug about pasta.

That is the whole distinction, and it decides which axis deserves your folder names. Topic is inherent in the video, which means it can be worked out later by anything that reads the video. Intent exists nowhere except in your head, which means if it is not captured at save time it is gone. There is a short list of intents, and almost everything falls into one of them:

• Do this soon — a thing with a real occasion attached. Cook it Sunday, try the stretch tomorrow, book the place before it fills up.
• Buy this — a product, a tool, a specific model number, waiting on money or a decision.
• Learn this properly — worth real attention later, not a two-minute skim. The long explainer, the technique breakdown.
• Share with a specific person — saved not for you at all, but for someone you had in mind while watching.
• Keep as reference — no action attached now, but you want it findable when the situation comes up. The tax rule, the wiring diagram, the packing list.
• Pure entertainment — you enjoyed it and you are never going to do anything with it, and that is a legitimate answer.

Read that list against your own saved folder and something becomes clear quickly: the buckets you actually need are not the buckets you built. There is no "Food" in there.

Intent decays, topic does not

The two axes also differ in how well they survive time, and that is the practical argument for the whole approach.

Topic is durable because it is recoverable from the artefact. Open a reel you saved a year ago and within four seconds you can tell it is about pasta. You did not have to remember that; you read it off the video. Nothing was lost, because nothing needed to be stored in you.

Intent is not recoverable, and it decays fast. A week after saving that reel you can no longer reconstruct that you meant to cook it on Sunday, that the Sunday in question has been and gone, or that you had picked it specifically because your sister was visiting and does not eat dairy. All of that was real, load-bearing information, and none of it is anywhere in the video. What you are left with is a pasta reel and a vague sense that it mattered.

A workable split

Hand topic to whatever automatic categorization you have. It reads captions, transcripts and on-screen text, and it will tell pasta from deadlifts perfectly well without your input. It will also be wrong occasionally, and an automatic system is the right place for that to happen: the error costs you one extra scan rather than your faith in the filing.

Then keep a very small number of intent buckets. Under six, and deliberately coarse. Coarseness is a feature, not a compromise: the point of a coarse bucket is that filing into it requires no thought and no judgement call, and only a step that asks for neither survives being made while you are barely paying attention. "Do soon" takes a quarter-second to decide. "Do soon — kitchen" versus "Do soon — errands" takes a moment of deliberation, and the moment is where the habit dies.

Fine buckets also drift, in the way any hand-maintained taxonomy drifts. You make a narrow one in a burst of enthusiasm, use it four times, forget it exists, then make a near-duplicate three months later. Neither of the two is authoritative after that: the reel landed in whichever one happened to exist that week, and nothing about the reel tells you which week it was. Coarse buckets have no near-duplicates to drift into. Five wide categories you actually use beat twenty precise ones you half-use, every time.

Why "watch later" is not an intent

Almost every system grows a "watch later" or "read later" bucket, and it is always the biggest one and always the least useful. It is worth being precise about why, because the failure is structural rather than a matter of discipline.

An intent has a completion condition — something that can happen, after which the item leaves the bucket. You cook the recipe. You buy the drill. You send the link. "Later" has no such condition. It is not a statement about what you will do, it is a statement that you have declined to decide, and there is no future event that resolves it. So nothing ever exits, arrivals continue at the rate you scroll, and the bucket grows without bound until it is functionally a second saved folder with the same problems as the first.

The fix is to force the decision that was deferred. When something is genuinely worth your attention later, it belongs in "learn this properly" — a bucket you can finish an item in by actually watching it and deciding it is done. When it is not, it is entertainment, which is a real and respectable answer that costs you nothing to admit. Almost everything currently sitting in your "later" pile is the second one.

None of this requires a new system or a weekend of reorganising. It is one substitution: stop naming folders after what things are, and start naming them after what you meant to do. Topic can be recovered from the video any time you want it. Intent is gone by the end of the week unless you spend the quarter-second to write it down — and it is the half you will actually be searching for.