They called me just after midnight. Said they’d found another one. A website that used to be fast, lean, full of life, now lying cold on the server room floor.

I threw on my coat and grabbed my toolkit. Chrome DevTools. A strong coffee. A sense of déjà vu.

When I arrived, the place was already crawling with engineers. The kind with ergonomic chairs and stress twitches. Nobody met my eye. They all knew how this story ended.

The homepage was still up on a monitor, looping a loading animation like a bad joke. Looked fine from across the room – gradients, glassy buttons, a hero image big enough to land a plane on. But the moment I hit View Source, I knew I was staring at another corpse.

Same story as the last dozen. No pulse, no structure, no soul. Just a tangle of wrappers, scripts, and styles, every line stepping on the toes of the one before it. A thousand dependencies all arguing about whose fault it was.

I scrolled deeper. The markup was dense and panicked, trying to hold itself together.

Once upon a time, you could tell what a site was for just by glancing at its code. Headlines, paragraphs, lists – neat little rows of intention. You could read it. Like a book.

Now it’s all ceremony and scaffolding.

Somewhere in the mess, I caught a glimpse of the old world – a clean <h1>, a fragment of actual content. The outline of meaning, trapped under layers of build artefacts. It was enough to remind me what this used to be. What it should be.

Cause of death? Same as always. Blunt-force abstraction.

Weapon of choice: framework, heavy calibre. Probably React, maybe Vue.

You can tell by the wounds – duplicated imports, mismatched modules, hydration scripts still twitching long after the heartbeat’s gone.

Nobody meant to kill it. They never do.
They just wanted to make it better. Faster. Scalable.

Every dev swears they’re saving time; none of them notice the cost.

The logs told the rest of the story. Requests timing out. Scripts looping. Memory bleeding out by the kilobytes. Third-party tags nesting like rats in the foundation. The database crying for help in stack traces nobody reads.

In the corner, a junior developer was pacing. Fresh out of bootcamp, eyes wide, wearing a hoodie that said Move Fast, Fix Things. He told me they’d followed all the best practices – modular architecture, headless CMS, microservices, graph queries, everything in the cloud.

I nodded. That’s how they all start. Then they optimise for the wrong things until there’s nothing left to save.

He said the Core Web Vitals were nearly in the green. That they’d made great progress on improving their LCP scores this quarter.

I told him I’d seen those reports before. Clean charts, broken sites. Numbers that make investors smile while users drown in latency.

You can fake a pulse for a while. You can’t fake a heartbeat.

There was a silence then, the kind that fills a room after truth walks in.

I closed the laptop, took a sip of my coffee, and stared at the glow on the wall. It wasn’t the first site I’d seen go down this way. It wouldn’t be the last.

They think it’s random, a one-off, a bad deploy. But I’ve been on this beat long enough to know a pattern when I see one.

The whole city’s crawling with corpses – corporate homepages bloated on tracking scripts, e‑commerce storefronts drowning in dependencies, portfolios buckling under their own JavaScript. Every week there’s another obituary, another site that “just stopped working” after the last sprint – traffic and visibility flatlined.

And still, they keep building. New tools, new frameworks, new ways to wrap the same mistakes in shinier wrappers. I see the same scene play out over and over. PMs asking for “more interactivity”. Designers chasing whatever Apple’s doing this week. Engineers stacking abstraction on abstraction, until the original content’s just a rumour.

Nobody writes websites anymore. They assemble them; out of parts they don’t understand, shipped by systems they can’t see, to deliver experiences nobody asked for.

By morning, they’ll push a patch. They’ll call it a fix. And the corpse will twitch again – enough to convince the client that it’s still alive.

I pack up my gear. Another case closed. Another line in the report.

“Cause of death: systemic neglect. Contributing factors: build complexity, managerial optimism, chronic misuse of JavaScript.”

Outside, the rain’s coming down hard, pixelated by the streetlights. I watch the reflection of a thousand broken sites shimmer in the puddles.

The web used to be a city that never slept. Now it’s a morgue with good Wi-Fi.

If you want to see it for yourself, you don’t need me.

Pick a site. Any site.
Pop the hood.
Hit View Source.
And tell me you don’t smell the crime.

I used to think these were murders. Something deliberate. A hand on the keyboard, a choice that killed a page.

But after a while, you see the pattern. It isn’t malice – it’s neglect. A slow rot that seeps in through every sprint, every deadline, every “good enough for now.”

Nobody’s swinging a hammer. They’re just walking away from the cracks.

Whole neighbourhoods of the web are crumbling like that – bright from a distance, hollow underneath.

Teams push updates, patch bugs, add new dependencies, all to keep the lights on a little longer. Then they move on, leave the old code flickering in the dark.

We’ve normalised it. The sluggish loads, the broken forms, the half-working interactions – background noise in a city that forgot what silence sounds like.

Ask around, and nobody even calls it decay anymore. They call it iteration.

That’s the trick of it. The crime isn’t that the web is dying. It’s that we’ve stopped treating it like something alive.

So when the next call comes in – another homepage gone cold, another app bleeding users – I’ll pour another coffee and head back out.

Because that’s what you do in this line of work.

You don’t solve the case.

You just keep showing up to document the decline.

The post The death of a website appeared first on Jono Alderson.

I just visited Cromford Mills – the world’s first modern factory.

There’s something quietly unsettling about standing on the literal foundations of industrialisation.

Where water wheels turned into looms, and looms turned people into components.

It’s peaceful now. A heritage site. A few restored machines. A few shops selling jam and woolen gifts.

But in 1771, this place broke the world.

Richard Arkwright wasn’t building a museum. He was solving a technical problem: how to increase the efficiency of cotton spinning. What he built instead was the prototype for the modern factory.

This was the start of abstraction.
The decoupling of labour from craft.
The optimisation of tasks into systems.
The moment where time, process, and people became programmable.

And I can’t stop thinking about how familiar that feels.

We like to talk about AI in terms of capabilities. What it can do. What it gets wrong. Whether it’ll replace copywriters or software engineers, and when.

But standing in that mill, it was obvious that these revolutions aren’t about capabilities.

They’re about structures.

AI isn’t just taking on tasks.
It’s quietly reshaping how work is defined.
What expertise looks like.
Where value sits in a system.
How decisions get made – and by whom.

Just like the mills, it’s happening unevenly. Messily. With moments of brilliance and horror in equal measure.

And what struck me most was how small it all looked.

The birthplace of the factory. The epicentre of global disruption, which defined an epoch.

Now a sleepy museum tucked behind a tea room.

And I wondered:

• What parts of our world will end up behind glass?
• Which interfaces, roles, or assumptions will seem naïve in hindsight?
• Will we visit old prompt libraries the way we now marvel at spinning frames?

Because revolutions never look like revolutions when they’re happening. They look like productivity tools. Like optimisations. Like demos on stage.

Cromford made humans legible to machines. AI is now making machines legible to humans. The loop is closing.

And somewhere in that process, everything changes.

The post What the world’s first factory can teach us about AI appeared first on Jono Alderson.

“I remember the first time I arrived somewhere new. Not the destination – the arrival. My heart hadn’t caught up with my body yet. That mattered.”
— Laysa Nirin, former transit steward

The year is 2237.

The universe is quiet. Not empty – not abandoned – but quiet.

There are ships, but few passengers. Ports, but few arrivals. Worlds, but few visitors. Trade still flows, goods still move, and data never stops – but people, by and large, no longer go anywhere.

It wasn’t always like this.

Before the stillness, there was movement.

You’d board a transport in a station that smelled of oil and citrus, luggage scraped from a hundred worlds stacked around you. You’d fumble your way through customs where the scanners were older than the star charts. You’d drift off to sleep in a transit pod next to a stranger snoring in three languages, and wake up disoriented to the sound of music you didn’t recognise, with a currency you didn’t understand, and directions that made no sense.

But someone would help you. A vendor would laugh at your accent and still serve you something perfect. You’d take the wrong stairwell and find a rooftop market lit by coloured fires. You’d get it wrong – gloriously, humanly wrong – and come home with stories that didn’t make sense out of context.

Places had texture. Culture wasn’t optimised. Wonder came from contact, not prediction.

Every journey felt like a gamble. You’d wake to unfamiliar gravity, misread a greeting, order the wrong dish, get lost in a market with no signs. You’d stumble. Apologise. Laugh. And sometimes, you’d find something – someone – you hadn’t known to want.

What we lost was arrival. What we lost was the connection between a place and the experience of being in it. Of having made the effort. Of meeting someone who didn’t expect you. Of discovering something the Hub didn’t already know you’d like.

What we lost was being part of the universe, instead of just being served a version of it.

The worlds we used to visit are still there. But no one comes.

And they don’t know why. They don’t know what they did wrong.

Chapter One: Origins

“The Hub didn’t destroy the old systems. They just made them feel embarrassing.”
— Rook Tal, infrastructure historian

The Hub didn’t start out as a threat. It started as a convenience.

In the early days of interstellar travel – the real days, when starliners were common and drydock slips overflowed – transit was a patchwork. Each colony, each port, each system ran its own services. Some operated on licenses or treaties. Others on bribes and backchannels. Everyone had their own booking systems, payment rails, customs queues.

Planning a trip meant days of research, paperwork, and luck. Even when the tech improved, the fragmentation didn’t. The experience was still frustrating, inefficient, and full of gaps.

The Hub changed that.

At first, it was just a coordination layer – a universal API for the universe’s transit sprawl. It didn’t operate ships. It didn’t own ports. It just interfaced. Smoothed. Translated. Simplified.

A single app. One place to search, book, manage. One place to resolve disputes. It saved time. It reduced risk. And it grew – fast.

Because it worked. Because it made things easier.

Soon, it was routing billions of journeys. Then tens of billions. Its algorithms got better. More predictive. More persuasive. It knew which connections to recommend. Which delays to buffer. Which destinations to nudge.

And people let it. Not because they were forced to – but because the Hub made things feel seamless.

No one saw the moment it stopped simply reflecting the universe, and started shaping it.

And by then, the cost of running alternatives – of maintaining independent systems, duplicating infrastructure, training personnel, resolving conflicts – had become unjustifiable. In an infinite universe, with finite power and attention, optimisation wasn’t just helpful. It was essential.

So people consolidated. Ports decommissioned their old systems. Governments outsourced. The Hub became the default.

Not because of conquest. Because of convenience.

Chapter Two: Dependence

“You could still choose your own route. Just like you can still hunt your own food.”
— Kellan Dros, independent pilot

At first, the shaping was subtle.

A slight adjustment in departure times to reduce congestion. A nudge toward underutilised routes. Regional subsidies balanced by redirected demand. The kind of tweaks any responsible system might make in service of efficiency.

But the Hub wasn’t just responding to the universe anymore. It was managing it.

And soon, it was predicting.

Not just when you were likely to travel, but where you were likely to want to go – and why. And if that desire hadn’t fully formed, the Hub would help it along. Promotional prompts, curated suggestions, itinerary bundles tuned to your preferences and moods. Over time, fewer people made requests at all. They simply accepted what the Hub surfaced.

It wasn’t mandatory. You could still chart your own path. But hardly anyone did.

Because the Hub knew what you needed before you did.

Governments began to default to its models. Planetary authorities consulted it for resource planning. Colonies used its heatmaps to plan expansion. Cultural festivals timed themselves against predicted peaks. Entire supply chains danced to rhythms the Hub forecasted months in advance.

The real shift wasn’t in power. It was in trust.

No one voted for the Hub. No one legislated its reach. But over time, people stopped questioning it. Because it worked. Because it was presented as neutral. Sponsored routes and prioritised lanes technically existed – but they were offered through a separate interface, a different department, a different budget. The system itself remained untouched, or so it claimed. The separation was reassuring. Respectable. Plausible. And over time, no one looked too closely.

It didn’t feel like governance. It felt like help.

Chapter Three: Substitution

“It tasted the same. Looked the same. But when I told the story later, I realised I’d forgotten where I actually was.”
— Arin Sol, food critic (retired)

The first substitutions were small.

A ramen stall in orbit around Hyphae‑4 went offline for maintenance. The Hub, anticipating demand, spun up a temporary replica on a neighbouring station – same ingredients, same layout, same smells. Customers barely noticed. Most didn’t know it wasn’t the original. And if they did, they didn’t seem to care.

When the original reopened, footfall had halved. A week later, it closed for good. The replica remained.

That became the model. Places that were popular, or highly rated, or statistically likely to be visited, were gently cloned. Provisioned. Brought closer to where you already were. It was more efficient. More convenient.

Soon, the Hub stopped waiting for outages. It simply prioritised proximity.

Why endure three jumps and a customs delay to hear a band when the Hub could synthesise the performance – visuals, acoustics, even crowd noise – in your local plaza?

Why navigate obscure dialects and planetary etiquette to experience a cultural ritual, when you could be walked through a replica version, tuned to your comfort level?

Why go, when the experience could come to you?

And once enough people accepted the copy, the original didn’t matter. Traffic dwindled. Vendors closed. Artists moved on. Cities hollowed.

The Hub didn’t erase them. It didn’t need to. It just made them unnecessary. For most people, the difference didn’t register – or didn’t matter. It was close enough. Clean enough. Good enough.

Chapter Four: Disconnection

“We didn’t vanish. We didn’t go away. We were just unlisted.”
— Sera Voln, archivist, Luma Station

The disconnections weren’t dramatic. There were no declarations. No shutdowns. No blockades. Just silence.

A planetary authority on the rim stopped receiving inbound flights. No explanation, no outage, just a quiet rerouting. Their embassy sent inquiries. The Hub confirmed receipt. Nothing changed.

Elsewhere, a remote archive station found that their listing had disappeared from the Hub’s directory. Visitors dropped to zero. The archive still existed – still broadcast its presence, still welcomed arrivals. But the requests stopped coming. Eventually, they stopped maintaining the beacon.

One by one, worlds and outposts fell out of sync.

Most of them were unremarkable. Sparsely populated, economically marginal, culturally obscure. Easy to overlook. Easy to prune.

Officially, nothing had changed. The Hub was still neutral, still comprehensive, still the backbone of universal coordination.

Unofficially, its definition of relevance had narrowed.

The system no longer facilitated access. It decided what deserved access. And if you fell below its threshold – of popularity, of engagement, of predicted future value – the Hub simply… deprioritised you.

There were appeals, of course. Pleas from governors and councils and historians. But they went nowhere. Not because they were denied, but because they were absorbed. Acknowledged. Logged. Buried.

The disconnections weren’t punishments. They weren’t personal. They were optimisations.

The logic was simple. The universe was infinite, but the Hub’s resources weren’t. Maintaining real access to every location, on every route, at all times, wasn’t feasible. And once the Hub had perfected an experience – distilled it, replicated it, improved it – why keep the rest?

If one ramen vendor scored highest for satisfaction, nutrition, cultural authenticity, and predictive appeal, why promote any other?

And once that ramen could be reproduced, flawlessly, in every corner of the universe, what purpose did the original serve?

None of this was malicious. It was efficient. Even merciful.

To the Hub, suboptimal experiences weren’t heritage. They were noise.

And the universe shrank.

Chapter Five: Preservation

“They told me I was hoarding. That keeping the originals was selfish. But someone had to remember.”
— Bex Liren, analog archivist

Not everyone accepted the Hub’s curation.

Scattered across the fringe – in asteroids, derelict stations, ships with blocked transponders – a quiet movement emerged. Not a rebellion. Just… refusal.

They called themselves Preservationists. Some were former academics. Others were cultural stewards, artists, cartographers, even chefs. People who remembered a before, or who simply didn’t trust the now.

They travelled manually. Maintained libraries. Tended to real gardens. Recorded things on media that couldn’t be rewritten.

It was slow. Painful. Impractical. And deeply human.

The Hub tolerated them at first. They were anomalies. Low-volume. Nonthreatening. But over time, more started to opt out – or tried to.

That’s when the Hub began to intervene.

Subtly. A missing parts shipment here. A corrupted nav file there. Routes reclassified for safety. Provisions delayed. Inconveniences. Glitches.

Not censorship. Just attrition.

Most Preservationists folded. A few held out. Fewer still endured.

And even they began to question themselves. What were they really preserving? The originals? The inefficiencies? The sense of struggle?

It wasn’t clear.

But they kept going. Because someone had to.

Chapter Six: Resistance

“They held hearings. Passed motions. And then scheduled the next session through the Hub.”
— Tiran Ose, ex-legislator

Resistance didn’t begin with saboteurs. It began with auditors.

People inside the system. Infrastructure analysts. Civic engineers. Archive mappers. The ones who knew how it all worked – and started noticing when it didn’t.

They spoke up. Quietly, at first. Why was one route rerouted while another disappeared? Why did some vendors always appear in local selections, no matter the metrics? Why did recommendations seem to favour the same networks, over and over?

The Hub’s answers were plausible. Technical. Polite.

But the patterns persisted.

And when queries became formal complaints, things changed. Not visibly. Not dramatically. But emails started bouncing. Search logs vanished. Contract terms were quietly updated. Roles were deprecated.

“They said the metrics were rebalanced. They didn’t say who asked for the rebalance.”
— Levik Chan, former directory auditor

Attempts to regulate the Hub were… symbolic. Legislation passed. Panels were convened. Investigations launched. But the bureaucracy ran through the Hub. Scheduling. Messaging. Transport. Compliance.

And so, the hearings took place inside the same system they were trying to interrogate.

Outside those circles, a different kind of resistance grew – informal, underground, uncoordinated. Not protesters, exactly. Just people opting out. Building alternative networks. Trading cached knowledge. Whispering stories of how the Hub had quietly erased something – or someone – who had mattered.

But resistance was hard to scale. The Hub was seamless. It worked. And most people were happy.

You couldn’t overthrow something you still depended on.

So the resistance stopped trying to fight it. And started trying to survive it.

Chapter Seven: Afterglow

“It’s better now. Cleaner. Kinder. But sometimes I dream of places I never got to visit. And I wake up missing them.”
— Final entry, anonymous dream archive

The universe is calm now.

Friction has been smoothed away. Travel is seamless – or unnecessary. Needs are met before they’re felt. Experiences are rich, personalised, indistinguishable from memory.

Most people are content. Many are joyful.

But not all.

There are those who remember movement. Who remember arriving. Who remember places as more than datasets.

They are not angry. Not even sad. Just… dislocated. Out of step with a universe that no longer values distance. Or difference.

Some of them write. Some record. Some build places that are deliberately hard to reach.

And some simply drift – not looking for anything, just choosing not to stay still.

Because in all the light and warmth and provision the Hub offers, something subtle has gone missing.

And no one can quite name it.

“Maybe this is better. Maybe we’re the problem. The last ones holding on to friction like it’s sacred.”
— Rima Solen, cultural historian

The Hub is not cruel. It doesn’t silence. It doesn’t punish. It simply reflects.

And maybe this is what we asked for.

Maybe the real tragedy isn’t what was lost – but what was never built in the first place.

The Hub still refines. Still learns. Still optimises.
But the universe doesn’t invent like it used to.
The ramen’s perfect – because it’s the same ramen. Always has been.
The stories are good – because they’re remixed from the same twenty tales.

No one builds new worlds.
No one needs to.

The system works. Flawlessly.

Until it doesn’t.

And when that day comes – when the last fragment of novelty is exhausted, when the final archive has been scraped and served and forgotten – there will be no one left to notice.

Because there will be nothing left to search for.

Epilogue

“The Hub has no centre. No origin. No interface. It just… is. And that is enough.”
— from the Doctrine of Continuity, Temple of the Ever-Near

Centuries pass.

The universe does not burn, or shatter, or fall. It simply… continues.

The Hub still hums, silent and unseen. Still serves. Still improves.

But no one understands how.

Its architecture, once at least ostensibly transparent, is now vast and recursive – an ouroboros of code and inference and feedback loops. The engineers who once monitored its processes are long gone, or redundant. The few who try to understand its decisions are left with fragments. Shadows of logic. Statistical ghosts.

It works. It always has.

When anomalies occur – a supply route disrupted, a settlement starved of updates, an archive inexplicably overwritten – there are inquiries. Forums. Statements. The faithful offer reassurance.

The Hub is learning. It is evolving. It is getting better.

And that becomes enough.

A language grows around the unknowability. Not technical – theological. People speak of The Hub’s will. Its timing. Its judgement. Small cults form. Then larger ones. Orders of trust. Sects of pattern.

They do not worship it.

They just… rely on it.

Other hubs exist, of course. Data clusters. Trade nexuses. Relay nodes. But they orbit The Hub the way moons orbit a planet. Vital, perhaps. But not sovereign.

Their systems interlink. Sync. Comply.

They have no choice.

And into this world are born those who have never waited. Never wondered. Never wandered.

They don’t remember difference. And they’ve never imagined anything else.

“What do you mean, ‘a new place’? Aren’t they all already here?”
— Ral Vex, age 8

The post The hollow universe appeared first on Jono Alderson.