Skip to content
Prompt engineering

The 5-line prompt myth

'Short prompts are better prompts' is the most expensive folk wisdom in AI design. What the run logs actually show about brief length, and where the long brief's value really lives.

Intermediate2 min read (computed · recorded 9)updated 2026-09-10promptslengthevidenceai

Where the myth comes from

The myth has a true ancestor: for open-ended generation (a poem, an idea, a tagline), short prompts outperform — the model has room and the task rewards surprise. Somewhere along the way that finding was generalized into 'briefs should be short', and it quietly destroyed a generation of UI prompts, because a website is not a poem: it has twenty decisions that must be made consistently, and an unasked decision gets a default — usually the model's favourite generic one.

  • Length correlates with specificity, and specificity is what UI fidelity scores: the run logs show block-listed briefs at 88–93 fidelity and 3-line briefs at 70–80, consistently, across models.
  • A long brief full of genre praise ('modern, clean, premium, sleek') fails exactly like a short one — length without obligations is just a longer wish.
  • The real variable is obligation density: how many decisions the brief settles that the model would otherwise make by probability. Block list, palette, never-list, counts — four dense lines beat forty vague ones.
  • The 5-line brief's best use is iteration: a short brief to explore directions fast, then the winner gets the full obligation brief for the fidelity pass. Short for breadth, long for build.
obligation-density.md
Vague line:   "a modern landing page with hero and pricing"
               -> model defaults: gradient hero, lorem pricing

Obligation:   "hero: product name, one verb, one button.
              pricing: three named tiers, real figures,
              what each includes. palette: paper, ink,
              one green. no gradients, no fake testimonials."
              -> the defaults are replaced by decisions.

The evidence habit

The myth survives because nobody logs. Motif scores every run and the scores are the argument: briefs win on fidelity when they settle decisions, and the scoreboard is public in every prompt's run log. If you believe short prompts are better, run the same brief at two lengths on the same model, score both against the same rubric, and keep the number. The myth does not survive contact with a run log.

Words this guide uses that the glossary defines: Fidelity score · Run log

Practise the lesson

Theory sticks when you ship it. These original Motif assets put this guide's lesson to work — open one and copy it into your own page.