Writing this article began organically. Which is a funny thing to even have to say in 2026. Consider what happens when you give a developer two days off, unlimited internet archive access, and too many ideas crammed into one article.
What does organic even mean now?
I did not write this on a mechanical typewriter.
I wrote it on a PC with my stubby index fingers running Windows software that, miraculously, does not blue screen every ten minutes anymore. It only took Microsoft thirty years to pull that off.
To the left sits an analog record player with some secondhand Yamaha bookshelf speakers I found at a charity shop; to the right of me sits a modern dark wood-paneled Zen PC case, a processor that would have occupied an entire room thirty years ago, and a GPU that can synthesize gargantuan piles of AI slop or brilliant code in roughly ten seconds flat.
And yet, for all that raw power, it still comes down to an algorithm. It always has.
The Sharper Image and the Death of Wonder
When I was a kid I used to walk into The Sharper Image store at Faneuil Hall Marketplace in Boston and just stand there. Looking at technology I could not afford while the staff watched me carefully to make sure I did not break anything.
I also grabbed some brightly colored rock salt candy; I loved that stuff, some core memories right there.
That feeling of picking up a piece of technology and not quite knowing what it did, like a ten-year-old ape holding something from another civilisation, you cannot replicate that in a sterile Apple store. The technology is better now. Genuinely better. Faster, smaller, more capable than anything those shelves held. But the sense of wonder at the unknowable object is completely gone.
Everything is explained before you touch it. Every product has a thirty-second video, a Reddit thread, a YouTube teardown, a comparison article, a spec sheet, and six AI-generated summaries of what other people thought about it. The mystery has been optimised out of the experience.
I did not know it at the time, but that shop was one of the last places where a kid could walk in and feel genuinely tactile wonderment about the future. Confused in a good way. The way that makes you want to figure things out via curiosity; they eventually went bankrupt and resurfaced as an online-only store.
That feeling is what I keep chasing when I go back into the archive, or when searching for used records, that rush you feel of finding something illusive and rare amongst a pile of James Last Trumpet a gogo records, man that German bandleader sold some records back in the 70's.
I once found a rare copy of Philip K. Dick's short story compilation, but it was in French. Absolutely gutted… how did that even end up halfway around the world in a charity shop in the suburbs? What a journey it went through.
Mondo 2000 and the Magazine That Dreamed Too Hard
Which leads me to Mondo 2000.
Mondo 2000 was a glossy cyberculture magazine published out of Berkeley, California, through the 1980s and 1990s. It covered cyberpunk topics: virtual reality, smart drugs “noots”, the coming digital revolution. It was a more anarchic and subversive prototype for the later-founded Wired.
Wired won commercially. Mondo had more cyber soul.
It started as High Frontiers in 1984, edited by R.U. Sirius, the pseudonym of Ken Goffman. It became Reality Hackers in 1988, then Mondo 2000 in 1989. It ran 17 issues and folded in 1998. In its tear through the early 1990s, Mondo brought an anarchic, drug-addled sensibility to the geeky world of computers, drawing a wiggly line from gonzo rock journalism to novelty-chasing tech speculation.
Wired reads, by comparison, like the operating manual for an IBM mainframe. That is not an insult to Wired. Wired won because it was legible to more people. But Mondo felt like it was made by people who were genuinely strange, genuinely excited, and genuinely unsure how things were going to turn out.
That experimental uncertainty was the best part. And they didn't waste 80 billion dollars on a Metaverse attempt. Cyberspace…
When I was a kid I mostly went to the back pages and read all the small ads. Strange devices. Gadgets I could not identify. Things that promised to change the way your brain worked. The joy was not in understanding any of it. The joy was in the sense that there was a whole dimension of reality operating outside the things people talked about in school, and somebody somewhere was living in it right now, more than likely in San Francisco, a hippie and tech soup.
*You can still read the full archive: *
https://archive.org/details/Mondo.2000.Issue.01.1989/page/106/mode/2up?q=cyberpunk
I have been thinking about doing a Vault series updating those old articles. Taking what Mondo predicted and comparing it to where we actually landed.
If only there were enough interest in nostaligia.
Scarcity Made It Matter
What made those magazines special was scarcity. There were not many magazines about any topic. A few for technology. Two for skateboards. One for surfing. One for electronics. People had to wait. They came out once a month. You had to order them by mail or physically go down to a newsagent and pick them up.
Or, if you were a teenager with no money, you stood in the shop and read as much as you could before the owner sneakily appeared from behind you and said:
This isn't a library, kid, are you buying it?
Limited editions were the holy grail.
We still have those same sources now, just a thousand times more of them. And somehow people are still not happy. They did not want the quantity. They wanted the magic of human-created personalisation. The sense that somebody made this specific thing for a specific kind of person, and you happened to be that person, a collective of like minds.
That feeling is almost impossible to manufacture at scale. Which is why so much of what gets published now, regardless of how it was made, does not produce it or comes out of China; sorry, designed in California…
Spell Check Is AI. It Always Has Been.
People rarely think of spell check as artificial intelligence, but that is exactly what it is. It just arrived early enough that nobody held street protests about it. There were no heated forum threads about the death of authentic writing when Microsoft Word started underlining words in red.
I wanted to track down the actual history, and this time I went the old-fashioned way. No AI assistant. Just Google, Wikipedia, and thirty minutes I will not get back. Normally I use AI for research because it is faster and usually goes deeper, but there was something fitting about doing it by index finger or mouse click for this particular article, most of it anyway.
The first spelling checker was created at MIT in 1961 by Les Earnest, using a list of the 10,000 most common English words. That was the whole database. The program could tell you that something was wrong, but it could not tell you what to do about it.
In February 1971, Ralph Gorin, a graduate student under Earnest, created the first true spelling checker as an applications program for general English text. He called it SPELL, and it ran on a DEC PDP-10 at Stanford University’s Artificial Intelligence Laboratory.
Gorin wrote it in assembly language for speed, and built the first real spelling corrector by searching the word list for plausible corrections that differed by a single letter or adjacent letter transposition.
He made it publicly accessible. It spread around the world via the ARPAnet, about ten years before personal computers came into general use. By the late 1970s, spell checkers had become standard on mainframes at universities and large corporations. By 1980, programs like WordCheck existed for Commodore systems. Word 95 brought the familiar red squiggle into millions of homes and nobody wrote a single angry forum post about it.
The first iterations were verifiers, not correctors. They highlighted mistakes but offered no suggestions for fixing them. The hard problem was not finding the error. It was suggesting the right fix. That required something more sophisticated.
The solution is called Levenshtein distance, named after Soviet mathematician Vladimir Levenshtein who described it in 1965. The idea is simple and straightforward: measure how many single-character edits it takes to turn one word into another. Deletion, insertion, substitution. Type “hte” and the spell checker computes how far that is from every word in its dictionary. “the” wins. “hat” does not.
Here is the dynamic programming version in JavaScript, because at some point every technical detour ends with the same question: do you want this in your stack?
function levenshtein(a, b) {
const n = a.length;
const m = b.length;
const dp = Array.from({ length: n + 1 }, () => new Array(m + 1).fill(0));
for (let i = 0; i <= n; i++) dp[i][0] = i;
for (let j = 0; j <= m; j++) dp[0][j] = j;
for (let i = 1; i <= n; i++) {
for (let j = 1; j <= m; j++) {
const cost = a[i - 1] === b[j - 1] ? 0 : 1;
dp[i][j] = Math.min(
dp[i - 1][j] + 1,
dp[i][j - 1] + 1,
dp[i - 1][j - 1] + cost
);
}
}
return dp[n][m];
}
A basic spell checker wraps around this by comparing your input against a dictionary and returning the candidate with the smallest distance. Real implementations layer on word frequency, keyboard proximity, and language model probability. But the Levenshtein core is still there, doing the heavy lifting, more than fifty years after Gorin first wrote it in assembly language.
This is an early form of artificial intelligence. The same as Photoshop’s first tools that allowed image manipulation. Nobody raged about those either.
I had to dig deep into the archives to find when humans started complaining specifically about AI, because there were many earlier iterations of the same anxiety: writing to printing press, camera to painting, photos to Photoshop. The shape of the argument never changes. Only the technology does.
Great visual explainer of how Levenshtein works if you want to see it animated: https://www.youtube.com/watch?v=d-Eq6x1yssU
What People Are Actually Loathing
When I try to distil what actually bothers people about AI in 2026, I keep arriving at the same four things.
Lack of Jobs.
Authentic Creativity
Loss of being human.
Connection.
The first three are obvious enough. If a tool can do part of your work faster and cheaper, people will worry about what that means for their role, their income, and their relevance. That anxiety is not irrational.
The fourth one is the most interesting to me.
Most humans, most of the time, want to interact with other humans. Not because human content is always better, it often is not. But because when you add another layer of binary transformer magic in between, it dilutes something. It creates skepticism. It degrades the connection while it speeds up the productivity. The message arrives faster but with less of a person inside it. You can feel it even when you cannot prove it.
To the Archive!
I could have done this old-school manual way by scrolling endlessly through search results.
Instead I ran a small experiment with modern cutting-edge tools, two different methods of reading the same archive. One was a standard web fetch tool built into Claude. The other was Cloak Fetch, a stealth browser tool from Vektor Memory that approaches the web the way a human would, bypassing the layers that block automated requests. Same question, same archives, different answers.
What made it interesting is that both tools found two different answers.
Cloak stealth browser crawling comp.ai Usenet threads surfaced a December 1989 post titled “STRONG AND WEAK AI”, a researcher named Mike Coffin complaining that AI-simulated worlds were “very unlike any type of perceptual reality.” Technical frustration, not moral panic. The same search run through standard Claude web fetch tool surfaced something older and stranger: a 1991 post by Mauro Cicognini arguing that artificial intelligence would never surpass human intelligence, and that the correct counter-argument was reproduction; training a human being would “always be less expensive.” He was not entirely wrong about the economics.
I found it strange and interesting how to modern search tools using the same database could produce such different results with the same prompt…
Also, a sobering lesson to not say things on the internet you may regret, as they stay up there forever.
Go back further and the anxiety predates the internet entirely. Norbert Wiener’s 1950 book warned that machines capable of learning and making decisions “will in no way be obliged to make such decisions as we should have made, or will be acceptable to us.” That was before most people had a television. The same sentence, adjusted for vocabulary, appears in op-eds published last Tuesday.
What the archive actually contains, when you go looking, is not a single moment of panic but a slow accumulation of unease across decades. The July 1987 comp.ai thread on “the symbol grounding problem”, 91 replies deep, researchers from Apollo Computer, Rutgers, MIT, was arguing about something that sounds remarkably current: that AI programs manipulate symbols without understanding what those symbols mean. A chess engine doesn’t know what a queen is. It moves a token. The complaint was philosophical, not social, and it came entirely from insiders. The public wasn’t watching yet.
By June 1989 the tone had shifted. Barry W. Kort’s 345-reply thread on the Chinese Room Argument captured a different kind of frustration, not that AI was dangerous, but that it was producing hype without results. The complaint was that AI was too weak, not too powerful. Researchers were arguing amongst themselves that there were “no real solutions here.” The field had over-promised and the people closest to it were the angriest.
Then December 1989: Mike Coffin on “STRONG AND WEAK AI,” complaining that the simulated worlds AI was building had no grounding in perceptual reality. By 1991, Cicognini had given up on the technical argument entirely and suggested the correct response to artificial intelligence was to simply have more children.
And by February 1995, someone on comp.ai had posted a thread titled “Are there non-humans lurking on Usenet?” genuinely asking whether bots were already posting alongside humans without anyone knowing, the first thought or suspicion on the dead internet theory, possibly…
That last one landed in 1995. It is 2026. That question has not been resolved so much as quietly abandoned.
The arc across those eight years goes: philosophical doubt, technical frustration, economic dismissal, and then the first flicker of something stranger, the idea percolating that the boundary between human and machine communication was already blurring, and that most people hadn’t noticed.
Every stage of that arc has repeated itself with each new wave of the technology. We are somewhere in the middle of the current repetition right now, arguing about whether the outputs reflect reality, whether the field has over-promised, whether the economics make sense, and occasionally, in comment sections, whether the correct response is to simply have more children.
Philip K. Dick Understood This Before Anyone Else
I have read Do Androids Dream of Electric Sheep? once. I have watched both film adaptations somewhere around twenty times each, and I love them both for many different reasons. They are genuinely inspiring.
Trying to synthesise the book and the movies into a single complete theory is nearly impossible in one short article, so I will give you two of my tiny ideas.
The obvious one is memory.
The entire premise of the story is a question about what counts as a real experience. If your memories were implanted rather than lived, does that make them less yours? If you cannot distinguish the synthetic from the genuine, does the distinction still matter?
The less comfortable one is loss of control.
Roy Batty is not tragic because he is artificial. He is frustrated because he cannot negotiate the terms of his own existence. He did not set the parameters of his life. He cannot change them. He can only rage against them in the time he has left and snap his creator's head.
Poor Tyrell, he shouldn't have answered that lift/elevator call from Sebastian and gone back to bed.
That feeling of being served something you did not ask for and cannot override, runs through almost every online argument about AI right now. People are not necessarily arguing that the technology is bad. They are arguing that they did not consent to the shift. They woke up and the world had quietly moved in a direction nobody voted for, and they are pissed about it the way Roy was pissed. Not irrationally. Just without a clear target.
The longing for nostalgic days of organic creation is real. It is not sentimental weakness. It is a reasonable response to having authorship removed from things that used to carry it.
An extra thought: Deckard was a terrible detective. Fish scales? Come on, who keeps a pet fish in a bathtub? Shakes head
AI helps me with research and makes technical writing possible at a scale that would otherwise take weeks. The technical sections in this article would have taken me that long to write without help, and that is just not feasible for one person with no floor of fact-checkers and editors getting paid a hundred thousand dollars a year to produce one article a week.
Everything in life is a compromise between quality and quantity. In the past, people spent their time hunting for content. Now there is too much, and the quality is low.
But there is a clear difference between using a tool and disappearing behind it.
The tells are obvious once you see them. The em dashes in every other sentence. The generic opener about today’s fast-paced landscape. The numbered list format that appears whether the content actually has a natural list structure or not. The corporate confidence about topics the writer clearly has not thought through.
The staccato 3-part rebuttal is like a lawyer proving their point on a big case in front of a courtroom judge. Why AI?
The complete absence of any moment where the author’s personality shows up and says something that could not have been generated by a probability distribution.
The absence of the person who was supposed to be reviewing the work. Instead of writing authentic ideas and stuff with your own stubby fingers.
Posting raw output without reading it is the tell. The model hands you a draft. That is the beginning of the work, not the end. Your value in that loop is the editing, the judgment, the moment where you say “this is wrong” or “this misses the point” or “this needs the part about standing in The Sharper Image as a kid, because that is the only part of this that is actually mine.”
The Hand Behind the Artifact
Spell check did not kill writing. It changed it. Photoshop did not kill photography. It morphed into a fast toolset hidden in 20 dropdown boxes with added AI and more complexity in getting a layer to stay active.
AI will not kill human expression. It will change what effort looks like and what the evidence of that effort is.
What people are mourning, when they complain about AI slop, is not the technology. It is the loss of legible signs that a person was present. The rough edges. The unexpected opinion. The joke that only makes sense if you know the writer. The argument that goes slightly sideways because the writer was working something out in real time and did not quite get there.
Those things are not inefficiencies. They are the signal.
Think about the charity shop table. I walked in recently, and there was a Jenga-sized pile of books for fifteen dollars a bag. Not junk either. A full hardcover Harry Potter set. Romance novels. Popular science. Sitting there, completely unwanted, because most people have moved on to Kindle and iPad. The books did not get worse. The scarcity that gave them meaning just evaporated, and the medium changed.
That is the same thing happening to writing now, and to every other form of content that used to feel like it came from somewhere specific. The abundance did not improve things. It just made the signal harder to find.
The Sharper Image is gone. Mondo 2000 is gone. The newsagent who caught you reading without buying has pivoted into lotto, consumer trinkets, and vapes. But the impulse that made those things matter, that sense that somebody somewhere was genuinely trying to show you something weird and real and theirs, that impulse is still here. It did not go anywhere.
It just needs a real person behind it.
The machine did not take the trip that produced this article. Or wandered through a charity shop, remembered a Boston gadget store staffed by helpful but nervous salespeople, landed in a Berkeley mansion where people were making a cyberpunk magazine nobody quite remembers correctly, and ended up at Stanford University in 1971, watching Ralph Gorin write a spell checker in assembly language for a computer the size of a refrigerator.
That is a real trip. With real memories behind it.
They are my memories and not generated by AI, as that would be weird.
You can feel the tactile difference. That is the whole point; the most defiant anarchist revolting action would be creating a painting, writing a book, or building an analogue thing.
Because William Gibson wrote Neuromancer on a manual typewriter, and so should you: no pain, no gain…
Now go read an old paperback book or listen to a dusty record, buy some old nostalgic treasure, and store it in your apartment or house because it is yours to own and made by a human’s touch.
Sources: Stanford University History of AI Spotlight Exhibit; Les Earnest, “The First Three Spelling Checkers,” 2011; Wikipedia: Spell checker, Vladimir Levenshtein, Mondo 2000, R.U. Sirius; Document Journal: “Inside Mondo 2000,” 2021; Mondo 2000 Archive: archive.org
Ai Slop
Nostalgia
Cyberpunk
Memories
