Turn 20 UX Laws Into One-Line Instructions Claude Can Follow

Named UX laws only help an AI coding tool if you translate them into direct instructions, so this video compresses 20 classic laws into 20 one-line commands you can hand to Claude. In 31 seconds, @adam_ha_yes reads out the instruction form of each law while the on-screen list fills in the law names, from Hick's law to the Pareto principle. The pitch is that a markdown file of these lines gives Claude concrete design rules instead of vague requests like "make it look good."

The format: a law name on screen, an instruction in the audio

The video runs a split presentation. A numbered list from 1 to 20 builds on screen with law names such as Hick's law, Fitts's law, Jakob's law, and the Doherty threshold. The narration never says the law names. It speaks each one as an actionable rule: "Reduce choices per screen. Make targets large. Follow familiar patterns. Group related information." That translation is the whole point. Claude does not need the citation, it needs the behavior.

The first ten: interaction cost and attention

The first half of the list covers how users act on a screen. Hick's law becomes "reduce choices per screen." Fitts's law becomes "make targets large" and its neighbor, minimize target distance, becomes "place key actions nearby." Miller's law becomes "break content into chunks." The Doherty threshold gets a hard number: keep interactions within 400 milliseconds. The Von Restorff effect turns into "highlight the primary action," and the serial position effect and peak-end rule become "put essentials first" and "end flows memorably."

The second ten: structure, defaults, and simplicity

The back half shifts to visual structure and error handling. The Zeigarnik effect maps to "show visible progress." The Law of Prägnanz maps to "simplify complex interfaces." The narration adds rules for defaults and failure states: "Use sensible defaults. Prevent errors proactively. Make errors recoverable." On screen, the named laws continue with the law of similarity, uniform connectedness, Tesler's law, Postel's law, Parkinson's law, Occam's razor, and the Pareto principle, matched in the audio by lines like "maintain pattern consistency," "connect related elements visually," "reduce task completion time," and "reveal complexity gradually."

Why this works as a Claude prompt

Each line is short, imperative, and testable. "Make targets large" and "interactions within 400 milliseconds" are instructions Claude can apply and you can check, which beats asking for "good UX." The on-screen call to action, "Comment 'UX' for the full MD instructions," points at the intended delivery mechanism: a markdown file. In Claude Code, that maps directly to a CLAUDE.md or rules file that loads into every session, so the design rules apply without repeating them in each prompt.

Key Takeaways

• Translate design principles into imperative one-liners before giving them to Claude. The law name is trivia, the instruction is what changes output.
• The 20 laws span three areas: interaction cost (Hick's, Fitts's, Miller's, Doherty), visual grouping (proximity, similarity, uniform connectedness, Prägnanz), and memory or sequencing (serial position, peak-end, Zeigarnik, Von Restorff).
• Concrete thresholds beat adjectives. "Interactions within 400 milliseconds" gives Claude a target that "fast" does not.
• Store the list in a markdown instruction file so the rules persist across sessions instead of living in individual prompts.

Resources

• Laws of UX — Reference collection covering all the laws named in the video, including Hick's law, Fitts's law, the Doherty threshold, and the Pareto principle.
• Claude Code memory docs — How to put persistent instructions like these 20 rules into CLAUDE.md and rules files.