The Manifesto

The reflex

In the 1970s and 1980s, people hated answering machines. They were impersonal. Sterile. The death of real conversation. If everyone started using them, society would fragment into atomized silence.

The hate was real. People avoided them deliberately. The complaints were backed by genuine unease.

The prediction was wrong. Voicemail became invisible furniture. Now people don’t even think about it. The infrastructure that seemed dystopian became so mundane that complaining about it sounds ridiculous.

The pattern

The reflex tracks technology’s friction with existing habits. The bigger the change, the more real the hate. The hate is not evidence the change is bad—it is evidence the change is real. That friction happens every time: the technology is too useful to keep resisting, and the reflex fades faster than it predicts.

The current reflex

AI-generated code triggers the same pattern. People hate it. It is impersonal—a machine wrote it, not a human thinking about your problem. It is prone to errors—hallucinations, incompleteness, security risks. The accusation is that the AI simply generated garbage and the work was incomplete.

The hate is real. The complaints are valid. Plenty of AI code is shipped without enough scrutiny. The risks are genuine.

The hate is also not evidence the technology is bad. It is evidence the technology is powerful enough to change how work happens. That friction is where innovation lives.

What actually matters

But there is a distinction most of the hate skips over: the code is auditable.

“AI slop” implies code nobody read. Code that passed no testing. Code that made it to production unchanged because the authors thought automation meant no oversight. That code exists, and it is bad.

But open-source software is the opposite condition. Every commit is public. Anyone can read it. Anyone can fork it. Anyone can report bugs and demand fixes. You do not have to trust the process. You can verify it yourself.

This is not a defense of AI code. It is a clarification: the mode of distribution matters as much as the origin.

A private vendor shipping closed-source AI-generated code with no oversight is reckless. An open-source project where the code is public, licensed, and auditable by anyone who reads it is not “slop” just because the first commit was written by a machine.

The proof

Six open-source projects. 576 commits. Written by Claude-based agents. Licensed under MIT and Apache-2.0 so anyone can audit the contents.

Plus a closed-source marketplace—1,796 commits, 21 agents, running live at scriptreeapps.com. Built with Next.js, Apache OFBiz, Gitea, PostgreSQL, Stripe Connect, and Caddy. Not a demo. Not a toy. A functioning platform.

The hate is misdirected because it conflates who wrote it with whether it works. Those are separate questions. The software here answers the second one plainly: look at it. Read the code. The commits are public. You can run them yourself. You do not have to take our word for it.

The joke

Call it slop if the reflex satisfies you. “AI-generated code is garbage.” The receipt says otherwise, but the reflex does not need receipts. The reflex just needs to feel right.

That is exactly how people felt about answering machines. That is exactly how it always goes. The hate is the sound of a habit dying, not the sound of a prediction coming true.