Why It Goes Dark

In September 1991 the head of the computer science laboratory at Xerox PARC — the laboratory that had given the personal computer its face, the windows and icons and screens this sentence is probably arriving through — published an essay in Scientific American arguing that the whole direction was a mistake.

The most profound technologies, Mark Weiser wrote, are those that disappear. They weave themselves into ordinary life until they are indistinguishable from it. Writing is such a technology — candy wrappers are covered in it, and nobody's attention is taxed. So are electric motors: a typical car of his day hid twenty-two of them, noticed by no one. The personal computer was not. It sat at the center of attention and demanded to stay there, and Weiser — running the lab that had put it there — thought the demand was the flaw. His word for the alternative was calm: technology that lives at the periphery of attention and moves to the center only when needed, then steps back. His measure for it was pastoral. A walk in the woods puts more information at your fingertips than any computer system, he observed, and yet the walk relaxes where the computer frustrates; the machine that fit the human world, his essay ends, would make using a computer "as refreshing as taking a walk in the woods."

He was no innocent about the other fork. The same 1991 essay warns that hundreds of networked, sensing computers could make totalitarianism up to now seem like sheerest anarchy — he saw both futures from the same laboratory, and his recent biographer's verdict is that the calm one lost, not that it was naive. He died in 1999, at forty-six, eight years before the phone that settled the question the other way. The industry kept his vocabulary — every design team on earth can recite "the most profound technologies are those that disappear" — and built its opposite, for a reason the second essay in this series spent three thousand words on and this one can hold in a clause: attention, once measurable, was revenue.

This series has a rule, and the rule has held for six essays: argue for the world, never for the product. This essay is the rule's one exception, and it is written in the only register that earns the exception — a maker's notebook. What follows are decisions inside Gnoia, the instrument this series has so far declined to mention: each one with the pressure against it, and the name of whoever taught us why it should hold anyway.

Every default in a piece of software is a decision about who you will be while using it.

There is no neutral setting in a file. Here is what we set, and why.

First decision. The session ends.

She opens it; when the work of it is done, she closes it, and the field goes dark. There is no next thing queued. Nothing invites you to stay.

The pressure against this decision is every chart in the industry. Session length, daily actives, the metrics an earlier essay traced to their filing — a product that ends the session is a product that walks its own numbers downhill, on purpose, forever. The ending's great replacement, the infinite scroll, was designed at least twice in the same year — once inside Microsoft's image search, once by Aza Raskin, who evangelised it into ubiquity and has spent the years since saying in public that he regrets it. The regret is instructive precisely because the design made sense at the time; his own late summary of how good rooms produce bad defaults is four words long — incentives eat intentions. Nobody drifts into these decisions, in either direction. They have to be held.

The reason it holds has two halves. The first half is cardiological, and the previous essay carried it: the value of stillness lives in the ending, and an ending has to be built, because nothing digital ends by itself anymore. The second half belongs to Albert Borgmann, who in 1984 drew the sharpest distinction in the philosophy of technology: between focal things, which gather people and ask something of them, and devices, which deliver a commodity while their machinery withdraws from view. His example is the hearth against central heating. The hearth demanded skill and wood and tending, marked the morning and the evening, and gathered the household to one warm point; the furnace delivers warmth everywhere, to no one in particular, asking nothing — and nothing gathers. His other example is the table: food become commodity, he writes, disintegrates into snacks. The warmth is real, the calories are real. What dissolved was the shape.

A practice has a shape in time. A commodity is endlessly available, which is to say shapeless. The feed is warmth without a hearth, for the mind. The session that ends is the hearth decision: presence with edges, and one of the edges is over.

Second decision. The numbers are invisible.

The instrument keeps arithmetic, as every instrument must. It shows none of it. No score, no streak, no count of consecutive days, no progress bar on your inner life. What you see instead is weather — a field that warms, deepens, quiets.

Ursula Franklin — physicist, metallurgist, Quaker — sorting technologies in her 1989 Massey Lectures, named the two kinds: holistic technologies, where the practitioner controls the work from beginning to finish, and prescriptive ones, where the work is broken into steps and the steps are enforced. Prescriptive technologies, she wrote, "are designs for compliance," and their social mortgage is a culture of it — the trained acceptance that there is only one way of doing the thing. A visible number on a practice is a tiny foreman. It stands beside the work and grades it, and within a week you are no longer doing the practice, you are doing the number. Streaks are the purest case: a streak is a number whose whole function is to make stopping expensive — the precise inversion of the first decision.

So the numbers stay inside. If you could read them, you would play to them, and the listening would end.

Franklin's broader definition is this notebook's whole license. Technology, she said, is not a thing but a practice — the way things are done around here. A product, then, is a place where a way of doing things gets settled on behalf of other people, which is exactly why there is no neutral setting in a file, and why the settling deserves to be written down and argued with.

Third decision. Nothing important happens on a tap.

A tap is a twitch. Interfaces built for harvesting want the twitch — the cost of acting must be lower than the cost of thinking about it. The instrument's grammar for anything that matters runs the other way: a press, held, for a breath. Consent made physical. Long enough for the hand to hear from the mind.

Amber Case, carrying Weiser's torch, states calm technology's first principle as an economy: technology should require the smallest possible amount of attention. The notebook's amendment, learned in the building: smallest possible is not uniformly small. Attention should be budgeted like any scarce thing — near zero at the periphery, and deliberately, physically expensive at the moments of commitment. A design that makes everything effortless has not freed its user; it has just made the important indistinguishable from the idle. Calm at the edges, weight at the center — and the weight is not a departure from the calm-technology lineage but its own mechanism applied: Weiser and Brown's word for pulling a thing from the periphery into the middle of attention was recentering, and recentering something, they wrote, is how we take control of it. The hold is a recentering you perform with your hand.

Fourth decision. The ear opens before the mouth.

Voice first. You speak; it hears; and then — this is the important part — it is quiet. The temptation refused here is the oldest one in the conversation business: to answer everything, at length, impressively, because the answer is the demo. But the instrument's work is listening, and listening is mostly the discipline of not performing. She speaks to open and to close, and in between, mostly, she listens.

Weiser again, in the essay with John Seely Brown that named calm technology: what calms, they argued, is the easy movement between the periphery of attention and its center. Their canonical example was not software at all. It was an eight-foot string hung from the ceiling of a PARC hallway, its motor wired to the ethernet so that every passing bit gave it a tiny twitch: a madly whirling blur when the network was busy, a small stir every few seconds when the building slept. Nobody ever read the string. Everybody always knew what it said. That is the periphery doing its work — informing without demanding — and the ear is the body's own version of it: always on, rarely front-of-mind, instantly promotable. An interface that leads with the open ear inherits that grammar. One that leads with its own voice is a lecture.

Fifth decision. There is one way to ask.

The instrument answers questions of weight in exactly one way: a question formed and held, a cast, a reading that ends. There is no side door — no chat window where you can pull the lever again and again until the answer improves. That door existed in an early version. It was closed, and the ruling written down: one way of asking, and the asking costs.

Ivan Illich is the reason, and his criterion is deliberately a sliding scale, not a badge: tools foster conviviality, he wrote, to the extent that they can be easily used, by anybody, as often or as seldom as desired, for a purpose chosen by the user. Fifty years of tool-building have optimised the front of that sentence and quietly deleted the back. As seldom as desired — and the next sentence on the same page makes the criterion explicit: their existence "does not impose any obligation to use them." The tool must survive your neglect without begging. A purpose chosen by the user — not induced, not suggested at the moment of unlock, not whispered by a badge. Illich's warning was that a tool can grow out of its user's control — first, he wrote, to become his master — and that past a certain watershed industries generate the needs they serve. An instrument you consult is on the right side of that watershed. An instrument that consults you has crossed it.

Sixth decision. She has no face.

The voice that opens and closes the session belongs to a presence with no body and no eyes — a ring of light where the voice comes from, which you look through rather than at. The pressure against this decision is the entire companion industry, whose economics run on embodiment: give the voice a face, the face a name, the name a personality, and the personality a need to be visited, and attachment does the retention work by itself. The previous essay in this series watched that architecture grieve — users mourning partners a company could switch off, because the person they loved had lived on the company's side of the glass.

The refusal of the face is the refusal to open that account. Her voice is carried by the machine the way a cello carries a melody: the instrument is not the music, and nothing here pretends otherwise. Turkle's children needed their robots to be alive enough — alive enough to be a friend, to need care, to mourn. The instrument declines the test rather than trying to pass it. Nothing in this design asks to be loved. The person in the room is you.

Seventh entry, and it is not a decision but an objection, because a notebook that recorded only its wins would be marketing with a bibliography.

Borgmann would not believe us. His argument, read whole, is not that devices are bad at focal practices; it is that the device paradigm dissolves practice — that you cannot deliver the hearth through the logic of central heating, because the paradigm eats whatever is placed inside it. Put the deepest ritual in the world behind a pane of glass owned by an industry that farms attention, he would say, and the pane will win. The strongest reader of our own bibliography is the one who thinks this notebook is a category error.

And he was careful in a way the summary versions of him are not. He allowed devices in the service of practices — the runner's shoes, the cook's stove — while denying that engagement itself can ever be procured as a commodity: mediated, yes; delivered, never. On the screen specifically he was on record, not extrapolated: he rejected a focal conception of the internet outright, because the good life, he held, is necessarily embodied. So the bet of this notebook, stated in his terms, is that an instrument of attention can stay on the shoes side of his line — equipment in the service of a practice whose substance happens elsewhere, in the room, in the day, in the life the session hands back.

He may be right that it cannot. The gravity he describes is real, and anyone who has sat in a product meeting has felt it: every default pulls toward the feed the way water pulls downhill, not because anyone is wicked but because the slope was graded by the business model long before you arrived. What we have against the slope is a list of weights — the ending, the invisible numbers, the hold, the quiet ear, the single costly door — and weights are not a proof. They are a bet, renewed every time someone in a room says no. The bet is that the line does not run between devices and hearths, but between designs that farm attention and designs that hand it back; and that which side a thing lands on is not decided by its category. It is decided in rooms, one refusal at a time, by people who write the refusals down so they cannot be quietly unmade.

That is what this notebook is for.

The sentence this series extends has two halves, and the essay about silence took the first: it ends the session, goes dark. The second half is the promise the whole notebook exists to keep —

…and leaves you more here — not less.

Dark, in this design, is not the instrument leaving you. It is the instrument handing you back. The reading ends, the field quiets, the room you are actually in returns — and whatever was gathered in the session goes with you, because it was yours before it was ever ours.

The rest of the evening does not belong to us. That was the whole idea.