I should have noticed the warning signs the moment Gordon introduced himself as “Gordon, founder of the internet’s friendliest community.” He said it with the same tone a cult leader might use when welcoming you to a free weekend retreat in the woods. But he was a friend of a friend, and I needed work, and he needed “a few small backend features.” A little coding for rent money. Easy.
He sent me a spec sheet the size of a novella. The Friendly Internet Forum: A welcoming space for constructive communication. The phrase appeared at least five times before the table of contents. Buried deeper were the actual feature requests, each one just innocuous enough to slip past suspicion—at least at first.
The first request was simple: implement a like button. Except he also wanted it to morph into something else. After three minutes, it turned into an angry red face. The system would notify the original poster that “<username> is angry about your post.”
Why? I asked.
“People like being liked,” Gordon said. “Then they like being un-liked even more. It keeps them checking.” He sipped coffee like a man describing a new kind of solar panel. Not like someone designing digital interpersonal cruelty.
The second task: put a large, bright “Report This User” button next to every text field. If you clicked “Reply,” the first thing you saw was a button encouraging you to report the person you were replying to, before you even replied to the post.
“This encourages mindful communication,” he said.
“This encourages paranoia,” I said.
“Same thing,” he answered.
The system didn’t send the report to moderators. No, the report went straight to the user being reported. Instant notification: “<username> reported your post as offensive, abusive, or disrespectful.” No explanation. No evidence. Just pure accusation.
At this point, I assumed he was misguided, not malicious. Forums are weird. People want strange features. I once built an online knitting group a function that let them exchange “cozy vibes.” Everybody has their thing.
But Gordon’s “features” started spilling over into what can only be called engineered interpersonal warfare.
Take the Opinion Contrast Chart. If you posted something longer than a paragraph, the system generated a little graphic underneath: “You are 87% incompatible with <username>’s beliefs.” It was meaningless. Random numbers. But the sight of it reliably set people off. The first time I saw two users fighting under the influence of one of these charts, I realized I was watching digital bloodsport.
Then there was the Shadow Quote function. Whenever someone quoted your post in a reply, the system subtly tweaked the quoted text. A phrase like “I don’t care for that show” became “I obviously think the show is garbage and anyone who disagrees is delusional.” Small edits, but just large enough to start an argument.
Users would respond with “Why did you write that?” And the original poster, seeing the quote, would shout back, “I never wrote anything like that!” They were right. They hadn’t.
Whenever users blocked each other, the system sent out a notification to both parties: “<username> blocked you because of your behavior.” Not content, not preferences—behavior. Then it sent each of them a transcript of their last ten interactions entitled “Clarity Report.”
“You’ve gamified hatred,” I said.
“I’ve simply amplified what’s already there,” he replied.
Another feature: the Invisible Draft Bubble. When a user started typing, everyone else in the thread saw a floating message: “<username> is preparing a response…” No content yet. Just the hint of an argument forming. Users began attacking each other before the reply even appeared.
One of the ugliest features was the Who Really Posted This? algorithm. A nonsense machine-learning script stamped every post with a fabricated claim: “Verified: 62% chance this user is lying.” Users reacted exactly as you’d imagine: suspicion, accusation, dogpiling. Gordon called it “healthy skepticism.” I called it a partition of distrust.
Worst of all was the Misattributed Post glitch. Posts would appear under the wrong user’s name for thirty seconds before correcting. Long enough to trigger outrage. Then it would flip to the correct author, prompting cries of “Stop impersonating me!” and “Don’t lie about what you posted!” Gordon said it created “dynamic conflict arcs.”
I spent weeks implementing individual pieces without seeing the whole puzzle. It wasn’t until he asked for the Crumbling Confidence System that I finally understood.
He wanted anonymous messages sent to every user: “Some members think your posts are condescending.” “Several users dislike your tone.” “Others wish you would stop dominating discussions.”
They were all generated by the software. No real complaints. Just simulated whispers behind people’s digital backs.
“Why would you do this to people?” I asked.
“To show them the truth,” he said.
“What truth?”
“That everyone hates everyone else, whether they admit it or not.”
“That’s not true,” I said.
“Isn’t it?” he asked.
We had a long silence then. The kind of silence where the shape of a man becomes visible.
I tried to convince myself I was simply a contractor following instructions. Piecemeal tasks. Discrete code chunks. But deep down I knew I was constructing a machine for human misery. A social architecture designed for implosion.
The final straw came when he asked for a feature called the Honesty Spotlight.
He wanted the system to randomly select one user per day and post a banner across the entire forum: “Today’s Honesty Spotlight: <username>. Please vote on whether you believe they’ve been truthful lately.”
Votes were public. Naturally.
“What does the winner get?” I asked.
“Winner?” he repeated.
“Yeah. What’s the reward?”
Gordon smiled a slow, chilling smile. “The knowledge of how the community really feels about them.”
Then: “Can you have it done by Thursday?”
I stared at the request for a long time. Then I stared at my monitor. Then I stared at the small reflection of myself in the black corner of the screen.
I scheduled a call.
“I’m done,” I said the moment he joined.
“No, you’re not,” he said calmly.
“I’m not working on this anymore.”
He tilted his head. “Do you believe you’re a good person, Lance?”
“I try,” I said.
“And isn’t trying the same as failing, but slower?” he asked.
“You’re sick,” I said.
“No,” he answered. “I’m simply paying attention.”
That was it. I closed the laptop without waiting for a goodbye.
Out of a blend of guilt and horrified curiosity, I visited the forum later that night. Five users arguing in a thread titled “Introductions.” One had been given the tag Most Hated User of the Hour. Another was flagged as Under Community Review.
Posts were being misquoted in real time. Draft bubbles floated like storm clouds. Angry faces pulsed on every comment. A fake community sentiment message popped up on the screen: “Some users think you post too often.” It wasn’t even meant for me. It was just part of the ambiance.
The worst part wasn’t the chaos. It was knowing that Gordon was somewhere watching it all unfold like a man tending a bonfire that never went out.
I closed the page. Shut my machine down completely. Went outside into the cold air where the sky didn’t ask for my feedback. And for a moment, I felt something that resembled relief.
Then, softly, to no one at all, I said, “God, I miss typewriters.”
