Words Need an Address

Juan Castilla Brazales tilts his head back inside the Alhambra and lifts a pair of binoculars toward the roof.

Above him the muqarnas drop from the vault in thousands of plaster cells, a frozen honeycomb that catches the Granada light and throws it back in fragments. Inside those folds, fourteenth century craftsmen cut lines of Arabic verse. Castilla, the Arabist who leads the palace epigraphic survey, spent hours on those ceilings with the binoculars, reading poems that almost no visitor ever notices.

His team started the count in 2002. Seven years later they catalogued 3,116 inscriptions out of the ten thousand or so that cover the complex. They found writing in cellars. They found writing under staircases and around corners no eye reaches. One phrase recurs hundreds of times across the walls, wa la ghalib illa Allah, no victor but God. Three court poets supplied much of the rest. Ibn al-Jayyab, Ibn al-Khatib and Ibn Zamrak all held posts as secretaries in the Nasrid chancery, and masons carved their verse into the building that housed them. The survey filled eight bilingual volumes.

Sit with what those men did. They did not decorate a palace with words.

They built a book that a person enters on foot.

A courtier crossing that courtyard at dusk moved through an argument. Column by column, the poem advanced. His feet turned the pages. His shoulders learned where a line sat. He carried no device, opened no cover, and still spent a lifetime inside a text.

I spent thirty years on the other side of that arrangement. I recorded what already happened. I built newsrooms in two countries, chased the day's facts, and wrote a column in Stabroek News called Ways of Looking and Feeling. Every one of those years, I pressed language flat onto a rectangle and handed the rectangle to a reader who held it at arm's length.

The Nasrids solved a problem that we have re-created and cannot yet name. This branch names it.

Researchers at University College London and the University of Florida read 236,270 American time diaries and found a country walking away from its own sentences.

Their study, published in iScience in 2025, tracked the American Time Use Survey from 2003 to 2023. In 2003, twenty seven percent of Americans read for pleasure on a given day. By 2023, sixteen percent did. The participation rate dropped about three percent every year for two decades. Jill Sonke of the University of Florida, a co-author, refused to call the drop a dip. She described instead “a sustained, steady decline of about 3% per year.”

The American Academy of Arts and Sciences reads the same survey through a different lens and reaches a colder number. The share of Americans who spend more than twenty minutes a day reading for personal interest fell from 22.3 percent in 2003 to 14.6 percent in 2023. Over the same twenty years, the share who spend more than twenty minutes a day watching, gaming or using a computer for pleasure held near seventy nine percent. In 2023, eighty four percent of Americans gave reading five minutes or less on an average day.

Break the number by age and the floor disappears. Among Americans aged fifteen to twenty four, 9.5 percent read more than twenty minutes a day for pleasure. Close to eighty percent of that same group spent more than twenty minutes on a screen. The Bureau of Labor Statistics puts daily reading among fifteen to nineteen year olds at eight minutes in 2023 and nine in 2024. Gaming and computer leisure took about an hour and a half of the same day.

Now hold those figures against a second set.

Gloria Mark, Chancellor's Professor of Informatics at the University of California, Irvine, has clocked human attention on screens since 2004. That first year, she and her colleagues shadowed knowledge workers with stopwatches and measured an average of two and a half minutes on any one screen before a switch. By 2012, using software logs, they measured seventy five seconds. Recent measurements settle at forty seven seconds, with a median of forty.

Put the two data sets side by side and the popular story collapses. Nobody stopped looking at words. This generation reads more individual words than any generation in history, in feeds, captions, notifications, subject lines, chat threads and search results.

Word volume climbed. Reading duration fell through the floor.

People did not lose their appetite for language. They lost the ability to stay anywhere long enough for language to do its work.

Almost everyone blames the glowing rectangle, and the evidence for that verdict looks stronger at a glance than it does under a lamp.

Pablo Delgado, Cristina Vargas, Rakefet Ackerman and Ladislao Salmerón published the landmark meta-analysis in Educational Research Review in 2018. They pooled fifty four studies and more than 170,000 participants from 2000 to 2017 and titled the paper with a warning: do not throw away your printed books. Paper beat screen on comprehension, at an effect size of minus 0.21. Virginia Clinton, working separately in the Journal of Research in Reading in 2019, landed at minus 0.25. In January 2019, almost two hundred European reading scholars signed the Stavanger Declaration, the closing statement of the E-READ research network that Anne Mangen led from the University of Stavanger. They warned that paper still carries long informational text better than screens. Then they named the challenge: find a way to support deep reading of long texts inside a screen environment.

Read the moderators, though, and the verdict narrows to something far more interesting.

Delgado and his colleagues found the paper advantage in expository text at minus 0.27, and no advantage at all in narrative text, where the effect landed at 0.01. The penalty appeared when readers worked under time limits, at minus 0.26, and thinned out when they did not. The gap also widened by about a hundredth of a point every year from 2001 to 2017. Young readers raised on screens performed no better than older ones. Familiarity fixed nothing.

Follow that logic where it leads. If the screen itself poisoned reading through its light or its glass, the damage would land on stories as hard as on textbooks. It does not. If practice cured the problem, digital natives would show a smaller gap. They show a larger one. If luminance did the harm, twenty years of better panels would have closed the distance. The distance grew.

Something other than the screen carries the loss, and the moderators point straight at it. The penalty concentrates where a reader must hold a structure in mind, must know where an argument began and where it now sits, under pressure of time.

The screen did not break reading. The screen broke place.

No human ever got born able to read.

Maryanne Wolf built her career on that sentence. Wolf, who directs the UCLA Center for Dyslexia, Diverse Learners and Social Justice, argues across Proust and the Squid and Reader, Come Home that the brain holds no reading organ. Reading recruits circuits that evolution grew for other jobs, and it stitches them into what she calls the deep reading circuit. Because culture built that circuit, culture can rebuild it. Wolf worries, with reason, that fast bursts of digital text starve the contemplative dimension that yields insight and empathy.

Take her premise at full weight and one door swings open. A reading architecture that human beings invented remains open to redesign by human beings. Nothing about a rectangle carries divine authority.

So ask a harder question. Which arrangement of text does the brain actually reward?

Martin Dresler and his colleagues answered a version of that question in Neuron in 2017. They scanned twenty three of the world's most successful memory athletes alongside matched controls. Then they trained ordinary volunteers for six weeks in the method of loci, the ancient practice of placing things you must remember inside rooms of a familiar building. The training rewired the volunteers' brain networks toward the athletes' pattern. The closer a volunteer's connectivity profile moved toward that pattern, the more that volunteer gained on memory tests. The gains held four months after the training stopped.

The method of loci does one thing. It gives words an address.

Now look back across every reading technology that ever worked. A clay tablet gave a line an edge and a corner. A scroll gave a column a position along a length of papyrus. A codex gave a sentence a recto, a verso, a quarter of a page, a thickness of paper under the left thumb. The Alhambra gave a poem a wall, a height, a direction of travel. Every one of those systems let a body file language by location, and every one of them let a reader return by walking back.

The scrolling rectangle deleted the address.

Text on a phone occupies no place. It arrives, slides upward, and vanishes into a coordinate system with no landmarks. A reader cannot point at where the argument turned. A reader cannot walk back to it. The finger drags, the words shuffle, and the brain, denied the spatial hooks it files by, treats the whole encounter as weather.

That reframes the entire problem, and it reverses the industry's answer. For twenty years the display business has moved words closer to the eye, smaller, sharper, brighter, until the newest headsets sit a centimetre from the cornea. Every step of that march strips more location out of text.

Push the other way. Move words further from the eye, larger, fixed, and spread across the surfaces of a room, and reading recovers the one property it cannot work without.

Mark Weiser opened his September 1991 article in Scientific American with a sentence that the computing industry has quoted for thirty five years and acted on for a fraction of them.

“The most profound technologies are those that disappear.”

Weiser, then chief technologist at Xerox PARC, chose his first example with care. He named writing itself as the original information technology, and he pointed at where it had gone. Not into books alone, he noted, but into street signs, shop signs, billboards, graffiti and candy wrappers. Literacy technology sits in the background of an industrial society, demanding no attention, ready at a glance. Weiser understood that a person absorbs a street sign without performing the act of reading.

Then he laid out the hardware. His group at PARC built three classes of device, sized to the writing surfaces a room already contains. Tabs took the inch scale, roughly an active sticky note. Pads took the foot scale, roughly a sheet of paper. Boards took the yard scale, the electronic equivalent of a blackboard or a bulletin board. He asked his readers to count the writing surfaces around them. A typical room holds over a hundred at the inch scale, ten or twenty at the foot scale, one or two at the yard scale. His stated deployment goal followed from the count. Hundreds of computers per room.

Richard Bruce and Scott Elrod built the boards. Each one measured about forty by sixty inches and displayed 1024 by 768 pixels in black and white. A user wrote on it with a piece of wireless electronic chalk that worked against the surface or from across the room. PARC placed these liveboards in ordinary conference rooms and open areas and let anyone walk up without booking.

One prototype deserves more attention than history gave it. Marvin Theimer and David Nichols built a system they called a scoreboard, a wall of public information that tuned what it showed to the people standing in front of it. A reader looked, wore a badge, and the wall adjusted. In 1991, inside a research campus in Palo Alto, two engineers demonstrated a wall that read its reader.

Weiser also drew the line his own field would later cross. He rejected the goggle. Virtual reality, he wrote, gives a map and not a territory, and it throws out desks, offices, weather, grass, trees, walks and chance encounters in favour of a simulation. He wanted the opposite motion, computing pushed out into the room, machines fitted to the human world rather than humans fitted to the machine.

The industry shipped the tabs and the pads. It abandoned the boards.

Thirty five years later, a person carries one very important book in a pocket. The walls of the room hold nothing at all.

Two properties separate a wall that can carry a book from a television bolted above a fireplace, and both properties turn on light.

A television emits. A wall reflects. An emissive display fires photons at the eye and consumes power for every second it holds an image. A reflective display works the way paper and plaster and carved stone work. Ambient light lands on the surface, the surface returns part of it, and the reader sees words by the same daylight that shows the room. E Ink electrophoretic panels hold their image with no power at all once the particles settle, draw current only when the text changes, and gain contrast in bright sunlight.

Those two properties matter more than resolution. A reflective wall of text neither glares at a reader in a dark bedroom nor disappears on a sunlit porch. It costs almost nothing to hold a paragraph standing for a week.

The physics of these surfaces moved fast over the last two years.

In October 2025, Kunli Xiong of Uppsala University published in Nature, with colleagues at Chalmers University of Technology and the University of Gothenburg, a reflective display they named retina E-paper. They built pixels from tungsten trioxide nanodisks measuring about 560 nanometres, smaller than a wavelength of visible light, and reached densities above 25,000 pixels per inch. Their prototype reflects about eighty percent of the light that hits it, holds optical contrast near fifty percent, and runs full colour video above twenty five frames a second. It draws between 0.5 and 1.7 milliwatts per square centimetre. Xiong says the technology can “provide new ways to interact with information.”

Run the arithmetic on that last figure, because the figure decides whether a wall of words costs a household anything. A wall three metres wide and two and a half metres high covers 75,000 square centimetres. At the published draw, that entire surface consumes somewhere between forty and one hundred and thirty watts while its text changes, and far less when the text simply holds. A room of reading surfaces costs about what a room of light bulbs used to cost, and unlike the bulbs it repays the household in sentences.

Meanwhile the large formats arrived. E Ink and AUO Display Plus put joint investment into large-format module production, with mass production timed for late 2025. Manufacturers now ship colour electronic paper panels at 25 inches, 28 inches, 31.5 inches and larger, built for signage rather than for pockets. At Computex in June 2026, the manufacturer Seekink showed a modular tiled electronic paper wall, marketed for near-zero power decorative surfaces. Verify any single vendor's claims before you build on them. Read the direction of all of them together and the conclusion holds: the panel industry has started tiling walls.

The material problem closed while nobody watched. Only the intention stayed missing.

Turn every surface in a life into a page and one question decides everything that follows. Who writes on it?

The advertising industry answered first, and it answered years ago. Market analysts disagree on the size of digital out-of-home advertising. Estimates for 2025 run from about twenty billion United States dollars to thirty one billion depending on the research house, a spread wide enough to warrant caution about any single figure. They agree on direction. Every major forecaster projects the category doubling within a decade. The research firm eMarketer puts digital at 45.2 percent of all out-of-home advertising spend by 2028, against 22.0 percent in 2016.

Watch what that money does to public surfaces. Mordor Intelligence reports that municipal authorities in Singapore, Hong Kong and Vietnam now treat digital kiosks and bus shelters as dual-use assets. One panel delivers civic information and sells advertising. The city informs you and bills a brand for the privilege of standing beside the information. Advertisers already buy walls by the audience rather than by the location, adjusting the message to weather, traffic and hour.

So the surfaces will fill. That outcome no longer waits on anyone's permission. The open question concerns authorship, and Weiser saw it in 1991 before he saw anything else. He warned in the same article that hundreds of sensing computers in every room could make historical totalitarianism look like anarchy. He asked engineers to build privacy into the walls from the first day.

I write a ten volume novel called Nightmares of Knowledge about a man who merges humanity into one surveilled consciousness. I did not invent his method. I read it in a 1991 research paper written by an optimist.

Qualped therefore builds toward a hard refusal, and I want it on the record in this branch rather than buried in a policy page. A Qualped surface carries the reader's own text and the texts that reader chose. It sells no space. It runs no advertising. It reports no gaze. The wall works for the person standing in the room, or Qualped does not ship it.

Whoever writes the walls writes the population.

A generation raised inside rented sentences will think rented thoughts.

Picture the following morning. I have not reported it. I have designed it, and I want you to hold it as a designed future rather than a described present.

A woman wakes in Durham Region at ten past six. Frost holds the outside of the bedroom window and the room stays dark, so the wall beside her holds nothing. Reflective surfaces need light, and she needs sleep, and the two facts agree with each other for the first time in the history of the screen.

She opens the blind. Grey Ontario light crosses the plaster and the east wall comes up in text at the pace of a photograph surfacing in a tray. Three paragraphs. She wrote them the previous night, and QwaiAI placed them here because she works better on an argument she can see before she speaks to anyone.

She reads standing, brushing her teeth, at a distance of two metres. Nothing glows. Her eyes do the work that eyes did for four thousand years, catching light bounced off a surface. Her body does what a body does, holding its own weight, turning its head, walking closer to a line she wants to check.

In the hallway, stem four of her Simsbook holds the north wall at the height of her shoulder. She passes it eleven times before noon. On the seventh pass she stops, because the third sentence has gone wrong, and she says so aloud. The wall marks the sentence. She will fix it tonight.

Her son reads a chapter of his own book off the wall above the kitchen table while he eats. He can point at where the villain changed. It sits above the toaster, four lines down, and it will sit there until Thursday.

Nothing in that room competes for her attention. Nothing in that room notifies her. The words hold their positions like furniture, and she navigates them the way she navigates furniture, by memory of place. By Friday she can walk the argument from the bedroom to the front door and recite it. She stored it in a building rather than in a scroll.

She has not read a book. She has lived inside one for five days.

That scene requires no invention beyond intention. Every component of it exists today in a laboratory, a factory, or a shipping catalogue.

The Qualped screenless reader turns the surfaces of a life into the pages of a Simsbook. Four parts do the work.

The Simsbook supplies addressed text. A Simsbook tree divides into forty branches and each branch into ten stems, which gives four hundred addressable units in a single work. That structure exists for a reason most publishing never needed. A chapter divides by argument. A stem divides by surface. Four hundred stems map onto the walls, doors, panels and passages of a life the way four hundred lines of Nasrid verse map onto a courtyard.

QwaiAI works as the cognitive executive. Hold that term hard, because the industry will abuse it within a year. A cognitive executive writes nothing for the author. It runs the schedule of attention. It decides which stem belongs on which surface at which hour. It keeps a long argument in sequence across days, marks the sentence the author flagged in the hallway, and refuses to move text the reader has not finished. An executive function in a human brain holds a goal against distraction. QwaiAI performs the same office for a room.

The surfaces carry the words without emitting light into the eye. Reflective panels for permanence, projection for scale, ordinary plaster and paper wherever the budget says so.

Qualped Club supplies the readers. A person surrounded by their own sentences and by nobody else's judgement will drift into a private echo. The club exists to argue back.

A chapter divides by argument. A stem divides by surface.

You need none of that on Monday.

Print one paragraph you would defend. Tape it to a wall you pass eleven times a day. Leave it there for a week. Watch what your memory does with the fourth sentence, the one at shoulder height beside the light switch. Then compare that against the four hundred paragraphs you swiped past on a phone in the same week.

Write two hundred words of your own and put them where your eye lands without a decision. Give the words a wall, a height and a direction of travel. Walk past them until you cannot stand a weak verb in the second line.

Then teach a child to do it. Put a chapter above the breakfast table where a nine year old eats, and let the villain change above the toaster.

None of those actions costs money. All of them restore the one property the rectangle stole.

Return to Granada, and to the man with the binoculars, and to the poems cut into a ceiling where almost nobody would ever read them.

That detail troubled me for weeks. A craftsman climbs scaffolding and cuts a line of verse into a plaster cell high above a courtyard. He climbs down knowing the courtiers below will catch, at best, a suggestion of language in the shadow. Modern instinct calls that waste. Modern instinct measures a word by the eyeballs it collects in the first hour.

The Nasrids used another measure. They built for a person who would cross that courtyard ten thousand times across forty years. They understood something about human beings that the attention industry has spent two decades unlearning.

A word placed in the world does not need to seize you. It needs to wait for you.

It needs to hold the same spot, at the same height, in the same light, through every mood you carry past it. One morning your life arrives at its meaning.

A feed cannot wait. A feed exists to move. The rectangle in your pocket has never held one sentence in one place for one day. The country that lives inside that rectangle now gives reading five minutes.

Walls wait. Walls have always waited. Egypt wrote on them. Rome left so much carved Latin that the modern corpus cataloguing it runs to some 180,000 inscriptions. Granada filled one palace with ten thousand. Every one of those civilisations understood the page as a place you go rather than a surface you hold.

Qualped now returns the page to the wall, and hands the pen to the person who lives in the room.

Look up from this text at whatever surface stands in front of you. A wall, a door, a window, a ceiling above a bed. Blank, all of it, in a life you claim to want to fill with meaning.

Write on it.