Laws of ux

Laws of UX

Heuristic lenses for designing and critiquing interfaces. A lens is a default, not a mandate: apply the ones that fit, name each lens you use so feedback is concrete and teachable, and stay anchored to the user's actual goal — when their context or research contradicts a lens, the context wins; say so explicitly.

Load references/laws-catalog.md for a law's full takeaways, examples, and origins — reach for it when a finding needs depth or the user asks why.

Designing / building

Pick the 3–5 lenses most relevant to the surface (a form → Tesler's, Postel's, Working Memory, Doherty; a landing page → Hick's, Von Restorff, Serial Position, Aesthetic-Usability) and let them shape the choices as you make them. Done when the delivered design names the lenses that shaped it.

Auditing / reviewing

Walk every group below. Cite a lens only where there's a real issue or a real win. Done when all six groups have been walked and every finding carries three parts: the law by name, a severity (critical / important / nice-to-have), and a concrete fix.

Decisions & choices (reduce friction to act)

• Hick's Law — Too many choices at once? Break into steps, group, defer. Decision time grows with number/complexity of options.
• Choice Overload — Long undifferentiated lists (plans, products)? Add filtering, a recommended/featured default, side-by-side comparison.
• Miller's Law — Are items chunked into ~5–9 groups rather than one flat dump? (Chunking guidance, not a quota — never cite "7±2" to justify an arbitrary limit.)
• Occam's Razor — Can any element be removed without losing function? Remove it. Simplest sufficient design wins.
• Tesler's Law — Irreducible complexity exists — is the system absorbing it, or is it dumped on the user (e.g. asking for data you could infer/autofill)?

Targets & interaction (make actions easy to hit)

• Fitts's Law — Are interactive targets large enough, well-spaced, and near the user's focus? Primary actions shouldn't be tiny or far.
• Postel's Law — Inputs liberal in what they accept (phone/date formats, casing, whitespace), strict/clear in what they emit and validate? Forgiving + clear feedback.
• Doherty Threshold — Feedback under ~400ms? Use optimistic UI, skeletons, progress, and perceived-performance tricks where work is slow.
• Flow — Does the path keep the user in focused momentum (clear feedback, minimal friction), matching challenge to skill?

Attention & emphasis (guide the eye)

• Von Restorff (Isolation) Effect — Does the single most important action/element stand out visually — with one emphasis, not competing ones, and not styled like an ad?
• Selective Attention — Anything important styled like/next to ads (banner blindness)? Any critical change that could go unnoticed (change blindness)?
• Serial Position Effect — Are key items first/last in lists & nav (primacy/recency), least-important in the middle?
• Aesthetic-Usability Effect — Is it visually polished? Pleasing design raises perceived usability and buys tolerance — and can mask real usability bugs in testing, so verify usability separately.

Grouping & layout (Gestalt — structure perception)

• Law of Proximity — Related things near each other; unrelated things spaced apart. Spacing communicates relationship.
• Law of Common Region — Use a shared boundary/background (card, panel) to bind a group.
• Law of Uniform Connectedness — Visually connect related controls (shared color/line/frame); connect with a line/arrow when showing relationships.
• Law of Similarity — Similar appearance implies similar function; make links/nav visually distinct from body text.
• Law of Prägnanz — Resolve ambiguity toward the simplest form; simple shapes are processed and remembered better.

Memory & cognition (don't overload the user)

• Cognitive Load — Cut extraneous load (decorative noise, irrelevant info); respect intrinsic load. Is every element earning its mental cost?
• Working Memory — Recognition over recall: carry context across screens (breadcrumbs, visited states, comparison tables) so users never hold info in their head.
• Chunking — Is content grouped into scannable, hierarchical modules (e.g. formatted phone/card numbers, sectioned forms)?
• Mental Model / Jakob's Law — Does it match conventions users already know from other sites/apps? Innovate on value, conform on patterns. On big changes, let users opt into the new version.

Motivation & memory of the experience (completion & feeling)

• Zeigarnik Effect — Use progress indicators, clear signifiers of more content, and (honest) head-start progress to motivate completion.
• Goal-Gradient Effect — Show proximity to the goal; people accelerate near the finish. Visible progress + a small head start works.
• Parkinson's Law — Keep task duration at or under expectation — speed it up with autofill and smart defaults before it expands to fill the time.
• Peak-End Rule — Invest in the emotional peak and the final moment (confirmations, success states, waits). Negatives are remembered more vividly — defuse them.
• Pareto Principle — Are you investing effort in the ~20% of features/flows that drive ~80% of value? Prioritize there.
• Paradox of the Active User — Users dive in without reading docs. Put guidance in-context (tooltips, inline hints) along whatever path they take.
• Cognitive Bias — Watch for biases (e.g. confirmation bias) in both users and your own design/research decisions; design safeguards against them.

Tensions & guardrails

• Lenses conflict (Jakob's familiarity vs. a novel differentiator; Occam's minimalism vs. Tesler's irreducible complexity). Surface the tension and make a reasoned call.
• The motivation lenses (Goal-Gradient, Zeigarnik, Peak-End, Cognitive Bias) can manipulate. Use them to help users finish their goals — a dark pattern is a critical finding, never a recommendation.