The anti-slop manifesto

The first draft can be artificial. The final judgement cannot.

A few principles for making AI-assisted work clearer, more useful, and less likely to waste a stranger's time.

Swamp mascots rallying around clearer writing and recycling bloated drafts

Say the thing

Do not spend three paragraphs approaching a point that fits in one.

Name the audience

Content for everyone is usually useful to no one in particular.

Show your work

Replace inflated certainty with evidence, examples, and decisions.

Own the output

A human publisher remains responsible for every sentence.

What we reject

Volume pretending to be value

More text is not more insight. Faster production does not excuse weaker judgement, and a polished tone does not make an unsupported claim useful.

  • Introductions that merely announce that a topic is important.
  • Thought leadership assembled from frictionless generalities.
  • Confident factual claims nobody checked.
  • Design and code generated without regard for users or maintainers.

What we want

Human standards applied to AI speed

The useful future is not pretending the tools do not exist. It is using them while preserving taste, responsibility, expertise, and respect for the person on the other side.

  • Shorter routes from question to useful answer.
  • Specific language grounded in real products and people.
  • Clear constraints and honest uncertainty.
  • Editing as a required part of generation, not an optional polish step.

Less discourse. Better drafts.

Put the principle into practice on the thing you are working on.

Score a draft