The Attention You Cannot Automate

There is a specific feeling that arrived quietly over the last few years, and most people I speak to recognise it before I finish describing it. You sit down to make something. Within about four seconds you become aware that a system could produce a competent version of it faster than you can outline it. The impulse does not disappear, exactly. It goes thin.

This essay is about what remains after that thinning — not as consolation, but as a working description of what is actually still yours. The short version: capability turned out to be portable and attention did not. Everything downstream of that sentence is the practical part.

The asymmetry, stated plainly

A model can write a sonnet, draft a contract, summarise a century, or generate an image of your childhood home from a two-line description. Each of these costs it effectively nothing. Not metaphorically nothing — nothing in the sense that no alternative was forgone. The system did not decline other work. It did not spend a Tuesday. It has no Tuesday.

You do. Every hour you give to one thing is an hour permanently unavailable to everything else, and there is a hard, unknown, non-negotiable total. This is not a disadvantage that technology has exposed. It is the entire basis on which human acts have ever carried weight.

Consider how obvious this is in ordinary life. A friend who drives four hours to sit in a hospital corridor has communicated something that no message, however well written, replicates — and the content of what they say when they arrive is close to irrelevant. We have always known that presence outranks output. We simply never had to make it explicit, because until recently output was hard to obtain and therefore functioned as a proxy for presence.

The proxy has broken. The underlying thing has not.

Three kinds of attention, and only one is scarce

It helps to separate what we lazily call attention into three distinct operations, because automation has affected them very differently.

1. Processing — reading, scanning, sorting, extracting. Fully automatable, already automated, and no loss to mourn. Nobody was ever enriched by manually reconciling a spreadsheet.
2. Judgement — deciding what matters, what to cut, what is beneath the standard. Partly automatable, and the interesting battleground. A model can rank; it cannot care about the ranking.
3. Presence — being the specific mortal in the room while it happens, at cost, without the option of doing something else with those minutes. Not automatable in principle, because the scarcity is the substance.

Most anxiety about creative obsolescence is really anxiety about the first category, expressed in the vocabulary of the third. If your work was mostly processing dressed as craft, that is a genuine and painful economic problem — but it is not a metaphysical one. If your work involved presence, the ground has not moved at all. It has only become visible.

What deliberate attention actually looks like

Advice about attention usually arrives as prohibition: delete the apps, block the sites, discipline yourself into a monk. That framing fails because it treats attention as a resource to be defended rather than a currency to be spent. Defence has no upper bound and no reward. Spending does.

Five practices that survive contact with a normal life:

• Name the expenditure before you make it. “I am giving ninety minutes to this and those minutes are gone.” Said out loud, this filters an astonishing amount of activity that survives silent scrutiny.
• Keep one thing deliberately inefficient. One practice where you refuse assistance, not from principle but because the cost is the point. Handwriting the letter. Walking the route. Cooking the thing that takes four hours. The inefficiency is the signal, to yourself first.
• Distinguish the draft from the decision. Let machines produce drafts freely; they are good at it and drafts were never sacred. Guard the decisions. What stays, what goes, what standard applies — that is where a person is located in a piece of work.
• Schedule noticing. Attention goes to change, not to importance, which means the durable things in your life are structurally invisible. Ten minutes a week aimed at something that has not changed — a face, a street, a sky — corrects a bias you cannot correct by intention alone.
• Finish things at cost. The last fifteen percent of any work is where the expenditure concentrates and where all the temptation to outsource lives. It is also the only part anyone can feel.

The objection worth taking seriously

There is a real counter-argument here and it deserves better than a dismissal. It goes: you are romanticising inefficiency. A hand-carved chair is not better than a machined one; it is worse, and more expensive, and the only thing it transmits is your own vanity about having made it.

Partly right. Effort is not a virtue in itself, and there is a genuine failure mode where cost becomes a substitute for quality and the maker demands credit for the suffering rather than the object. The corrective is precision about which costs carry signal. Nobody wants your struggle. What lands is the choosing: that out of a finite life, you allocated this portion, here, to this, for someone.

The distinction is between cost that is incidental and cost that is directed. Grinding through a task a tool could do better is incidental — waste dressed as devotion. Spending nine years on forty pages because they were yours is directed. Same hours. Entirely different objects.

Why this is not a coping mechanism

It would be easy to read all of this as an elaborate way of feeling better about being outperformed, and I want to be careful here, because that reading is available and it is wrong.

The claim is not that human work is secretly superior. In most measurable respects it is now demonstrably not, and pretending otherwise has already cost people their livelihoods and their bearings. The claim is narrower and harder: that a category of value exists which is constituted by finitude, and that this category cannot be entered by anything that does not run out of time. That is not a competition being graded. It is a different kind of object.

The practical consequence is unglamorous. Stop competing on volume and speed, in your work and in your own private accounting of your worth. Start being deliberate about a resource you have been spending unconsciously since you were old enough to have it. There is no technique that makes this easy and no tool that does it for you, which is, at this point, most of the recommendation.

Related reading: what abundance does to meaning traces the economics underneath this argument, and how to keep a human record is the applied version — what to do with a life that leaves traces automatically. If you would rather have the whole thing as narrative than as argument, begin with the story.