Algorithms changed my sentences.
I’m writing this from Tirana, with the radiator ticking and my wheelchair tyres humming softly on the tile as I pivot back to the desk. The air smells like burnt coffee—my fault—and the screen throws that cold, aquarium-blue light that makes every draft feel guilty before it’s even born.
If you write online, you know the new boss. The algorithm isn’t just a tool in the room; it acts like the landlord, the judge, and the bouncer, all at once. And the part that messes with your head is how quiet it is—no shouted edits, no slammed doors, just a ranking that rises or sinks like a tide you can’t see.
My Personal Problem With “Good” Writing
The keyboard clacks like tiny castanets, and I catch myself doing it again—smoothing edges, trimming jokes, ironing out the odd rhythm. Not for you, not even for an editor, but for a machine-shaped expectation that sits in my peripheral vision like a silent metronome.
Years ago, the fear after you hit “send” was blunt: “This doesn’t work, rewrite everything”. Now the fear is sneakier, and it smells like plastic wrapping fresh off a gadget: “It looks like it was written by AI” .
That accusation is nastier than plain failure. If you blew it, at least it was your mess, your fingerprints on the glass.
And yes, I hear the irony as my chair creaks and I type this: we’re terrified of being mistaken for the very machines that were trained on the way humans write. The room is quiet, but the pressure has teeth.
Before we go on, I’m going to do one thing I always do at Free AstroScience when a concept gets technical: I’ll simplify it on purpose, so you can hold it in your hands. Algorithms, ranking systems, and text models aren’t magic; they’re pattern machines, and they love whatever looks like an average of everything they’ve already swallowed.
Three Ideas That Should Make You Uncomfortable
Let me throw three sharp stones into the pond and listen to the splash.
The first: the mainstream worship of “clarity” is turning prose into sterile packaging, and it’s not making readers smarter—it’s making them drowsy . You can almost hear the vacuum seal tightening around each sentence, that soft hiss of air being sucked out.
The second: SEO and “content structure” don’t serve human curiosity; they serve predictability, and predictability is the enemy of surprise . You can feel it in your wrists after a long day—writing starts to resemble filling out a form, not telling the truth .
The third: our obsession with spotting AI text is training people to distrust any writing that sounds competent. The suspicion that a piece is “too similar” to a machine becomes a stain that doesn’t wash out, even if the author is, as the source puts it, “flesh, blood, and caffeine”.
If those three ideas make you bristle, good. That bristle is the sound of your human ear catching something off-key in the culture.
The Trap: “Simplify Everything”
Here’s the cultural damage I recognise in my own drafts: simplification at all costs. Every thought must be neat, linear, instantly processable, like it’s being run through a scanner that hates smudges.
But reality—the kind you and I actually live, with odd pauses, mixed motives, and embarrassing memories that pop up in the shower—is not neat. Reality is disordered, sometimes absurd, often poetic in the wrong place, and full of “useless details” that still make you feel the scene.
When we scrub those details out, we don’t get “better” writing. We get polished prose that looks flawless and feels like chewing on cardboard, what the source calls “Maximum Informative Efficiency” and, more honestly, “Maximum Boredom Guaranteed”.
AI loves averages and hates rough edges. It hates the little, unnecessary gestures, the sentences that vibrate for no documented reason. And if you start writing as if your only job is to avoid confusing a machine, you shrink into the creative freedom of a change-of-address form .
I know, because I’ve done it. I’ve watched myself do it, with the faint buzz of the laptop fan sounding like a warning.
The Other Trap: “Engineer Everything”
Then, just when you try to rebel by getting weird, the ecosystem pushes you toward complication . Not the fun kind—no playful style, no risky leap—but “narrative engineering,” where every paragraph must sit in the right place like pipes in an industrial plant .
SEO demands structure, Google demands relevance, and generative models demand sentences so predictable they can be summarised like a datasheet . Meanwhile the reader—poor soul—just wants not to be bothered by the machinery .
The source nails this as an impossible triangle: humans want brilliance, Google wants consistency, and models want compressible predictability . You can almost hear the grind of gears when you try to satisfy all three at once, like a shopping trolley with one bad wheel.
And it gets darker when the machine refuses to explain itself. The moment that stuck with me is painfully familiar: someone asks a model why one text gets indexed and another disappears, and the answer is the universal shrug—“It depends on the context” . That line has the texture of wet soap: it slips right out of your hands.
The Opposite Argument, Proven by One Tiny Story
Now I’m going to challenge my own anger with one story, because a single scene can do more work than a thousand hot takes.
A while back, I wrote two pieces close together. Same care, same late-night silence, same smell of coffee gone cold, same little ache in my shoulders from leaning forward too long.
One piece travelled. The other sank.
So I did what the source describes—I asked the machine why it “indexed one text and ignored another,” and I got that soft, cloying reply: “It depends on the context” . And in that moment I realised the opposite of what I’d been telling myself.
The problem isn’t that we need to outsmart the algorithm. The problem is that we’ve started treating the algorithm as a god that deserves ritual purity, when it’s really just a pattern engine with no skin in the game.
That answer—“It depends”—taught me a blunt lesson: you can’t write your way into guaranteed permission. The promise of total optimisation is fake, and the sound it makes is a gentle lie, like elevator music over bad news.
So what do you do with that?
You stop sacrificing your voice to a system that won’t even explain its rules.
Ken Robinson Was Right, and We’re Doing the Reverse
There’s a quote in the source from Ken Robinson’s 2006 TED Talk that lands like a dropped mug on a kitchen floor: “If you are not prepared to be wrong, you'll never come up with anything original” . I can almost hear the crack of ceramic, that sharp punctuation.
And what are we training writers to do right now? The opposite—we’re training writing to never be “wrong,” meaning never original, never alive, never dangerous .
That word “dangerous” matters. Not dangerous like harm, but dangerous like truth that hasn’t been approved yet, like an idea that walks into the room wearing muddy shoes.
As a science communicator, I’ve spent years translating space and physics into plain speech for people who didn’t ask for equations. The funny part is that the universe itself doesn’t behave like an outline—stars wobble, data has noise, instruments drift, and discoveries arrive with messy edges.
So why are we pretending our writing should sound like labelling on a bottle?
A Gentle Rebellion That Actually Works
The source doesn’t suggest a macho war against AI. It suggests coexistence with a soft kind of resistance, and I can feel the relief in that idea like warm cloth on cold hands .
Machines will beat us on volume, speed, and efficiency . That’s not a moral failure; it’s arithmetic.
But machines won’t beat the human move of writing a slightly lopsided sentence that still has heart . They won’t know what to do with a digression that goes nowhere yet makes the piece breathe .
They won’t taste irony that rises out of a crack in thought . Irony has a smell—metallic, like rain on a railing—and it usually appears when you admit you’re not fully in control.
So, what exactly is this “gentle rebellion” in practice?
It’s deciding that your job is not to produce “optimised” text, but text that breathes . It’s writing something an algorithm struggles to classify, not because you’re trying to be obscure, but because you’re trying to be honest.
What I Do Now When I’m About to Self-Censor
When I feel that invisible supervisor behind my shoulder, I do a few simple things, and I’m telling you because you can do them without buying anything or installing anything. I pause and listen to the room—the traffic hiss outside, the fridge’s low groan, the faint squeak of my chair—and I ask, “Would a real person say this out loud?”
Then I let one detail stay that I’d usually delete. A weird metaphor, an off-beat rhythm, a moment of confession that doesn’t “serve” the outline but serves the reader’s trust.
I also stop treating structure like a cage. Yes, you should guide people, especially if you’re explaining science or culture, but guidance isn’t the same as sterilising everything until it shines.
And when I’m tempted to write like a brochure, I remember that line about “synthetic oil” versus water spilling over the edges . Water is messy. Water is alive. Water makes a sound.
Why This Matters Beyond Writing
This isn’t only about articles. It’s about what kind of humans we become when we’re trained to speak in machine-friendly shapes.
When your sentences are always clean, your thoughts start to feel clean too, and that’s dangerous in a different way. Real thinking is often loud in the head, full of half-steps, and it leaves fingerprints.
If you’ve ever cared for someone, fallen in love, lost a friend, moved countries, or watched your body change, you know life doesn’t arrive as a neat paragraph. It arrives as a jumble of smells, textures, and noise—street smoke, hospital antiseptic, the dusty sweetness of old books, the click of a wheelchair brake.
So I don’t want a writing culture that punishes the “useless details” that make us recognise each other . I don’t want young writers learning that the safest sentence is the best sentence.
The Future I’m Betting On
I keep coming back to the closing image: the future of writing isn’t a tug-of-war with machines; it’s a ballet where they create order and we create chaos, and somewhere between those moves, creativity shows up . I can hear the soft scuff of shoes on a stage, the breath before a leap.
In 2026 and beyond, the pressure to sound “acceptable” will rise, not fall. Detection tools will keep guessing, ranking systems will keep humming, and people will keep asking for “cleaner” copy the way they ask for brighter lights.
My bet is still on the messy human page.
So write like a person again. Let one sentence lean. Let one paragraph carry a smell from your real life. Let a thought stay “wrong” long enough to become original .
And if the machine doesn’t like it?
Let it choke a little. Your voice was never meant to be chewable data.
Post a Comment