Human Track — Taste & Judgment Catalog

What the eye must judge — the design that doesn’t reduce to a rule an agent can run. This is the catalog Wedge actually reads and looks at: where an agent can produce a candidate but not tell whether it’s good. Extracted per Design Two-Track Extraction. The mechanical core of anything here lives in the Agent Track. Source tags: [RUI] = Refactoring UI; [UPOD] = Universal Principles of Design.

Developing the eye

• Rebuild interfaces you admire from scratch, no peeking at devtools. The friction of matching the original is what surfaces the invisible tricks (tighter heading line-height, letter-spaced caps, doubled shadows). This is the single highest-yield calibration drill. [RUI]
• Hunt for “decisions you wouldn’t have made.” On any design you like, ask what the designer did that you’d never have thought to do (inverted datepicker, button inside the input, two-color headline). Collect these. [RUI]
• Run the elimination exercise to train comparison. Given a system, the act of seeing that two of three options are obviously wrong is the trainable perception; the scale itself is agent-work. [RUI; UPOD: Comparison]
• Become a type snob on purpose. Sustained attention to typography on good sites converts “I can’t say why” into fast good/bad calls. [RUI]
• Trust your eyes over the numbers at the finish. Systematic palettes, proportions, and scales get you 90% there; the last S/L tweak is made by eye. [RUI]
• The aesthetic-usability effect is real and it’s a trap. People rate attractive things as more usable, so beauty buys forgiveness — but it can also mask genuine usability faults from you. Judge both, separately. [UPOD: Aesthetic-Usability Effect]

When to break the rule

• Dense UIs are legitimate — deliberately overriding “start with too much whitespace” for dashboards. The test is whether density is a choice, not a default. [RUI]
• Grids are overrated; don’t be a slave. Break to fixed widths the moment fluid columns make something wider-on-medium-than-large or shrink past a usable minimum. [RUI]
• >75 characters per line “can sometimes work” — a judgment call you only make knowing you’re in risky territory. [RUI]
• Justify only to mimic print (magazine/newspaper feel), and even then left-align is usually fine. [RUI]
• MAYA — Most Advanced Yet Acceptable. Push novelty exactly as far as the audience will tolerate and no further; where that line sits is pure read-of-the-room. [UPOD: MAYA]
• Satisfice. Knowing when a solution is good enough to ship vs. worth more polish is a judgment the rules can’t make for you. [UPOD: Satisficing]

Aesthetic coherence & feel

• Choose a personality and make every choice serve it. Font, color feel, border-radius, and copy tone must cohere toward one character (serious / playful / elegant). Often a gut call; calibrate against the sites your audience already uses, but don’t ape direct competitors. [RUI]
• Border-radius: consistency is a rule, amount is a feel. Mixing square and round looks worse; how much round communicates personality. [RUI]
• Color “psychology” in practice = what looks right to you. Use it to explain a choice after the fact, not to derive one. [RUI]
• Restraint with realism. Once you can simulate light, the temptation is to over-tune toward photo-realism; knowing when to stop is taste. [RUI]
• Hierarchy is what makes something “feel designed” — and noticing when it’s off (noisy, flat, competing) is a perceptual skill, not a checklist. [RUI]
• Form follows function — but “faith follows function” warns the inverse fails: functional-looking ≠ functional. Judge whether honest form actually fits, vs. mere styling cues of competence. [UPOD: Form Follows Function, Faith Follows Function]
• Wabi-sabi / Gloss & Contour bias. Humans are drawn to glossy and to curved-over-angular, and find beauty in imperfection and age. Whether to lean into polish, softness, or deliberate roughness is an aesthetic stance. [UPOD: Gloss Bias, Contour Bias, Wabi-Sabi]
• Symmetry vs. dynamic balance. Symmetry reads as stable/formal; asymmetry reads as energetic. Which one fits the feel is judgment. [UPOD: Symmetry]
• Back of the dresser: finish the parts no one inspects (code, empty states, edge layouts) — a craft standard you hold yourself to, not a checkable rule. [UPOD: Back of the Dresser]

Visual perception (gestalt & attention)

• Gestalt grouping is perceived, not measured. Proximity, similarity, closure, common fate, good continuation, figure-ground — an agent can space things, but whether the eye reads the intended groups (and where figure/ground flips ambiguously) needs a human look. [UPOD: Proximity, Similarity, Closure, Common Fate, Good Continuation, Figure-Ground]
• Surface area drives perceived weight. Why bold and solid icons “feel heavy” — train the eye to feel imbalance before you can fix it with contrast. [RUI]
• Hue has perceived brightness independent of HSL lightness (yellow reads lighter than blue at equal L). The formula is agent-work; judging “rich vs. dull / warm vs. muddy” after a hue rotation is the eye. [RUI]
• Where attention lands first — focal point, entry point, reading gravity. Verify the eye actually goes where you intended; only looking tells you. [UPOD: Entry Point, Inattentional Blindness]

Problem-framing & audience

• Start with a feature, not a layout. Frame the real problem before any pixels; the shell can’t be designed until a few features exist. [RUI]
• Design the smallest shippable version; be a pessimist. Don’t imply functionality you can’t build; work in short build-test cycles rather than imagining every edge case. [RUI]
• Detail comes later — design in grayscale first to force spacing/contrast/size to carry the hierarchy before color rescues it. [RUI]
• User-centered vs. user-driven. Users report problems well and prescribe solutions badly; deciding how much to weight what they ask for vs. what they need is judgment. [UPOD: User-Centered vs. User-Driven Design]
• Mental model is the whole game. The user’s model is built only from what you show; judging whether your surface actually induces the right model (and surfacing it when “synced/cloud” breaks) is interpretive. [UPOD: Mental Model]
• Design the remembered experience (peak-end). People judge an experience by its peak and its end, not its average; deciding which moment to make the peak and how to end well is a taste/empathy call. [UPOD: Peak-End Rule]
• Desire lines tell the truth. Watch where users actually go and pave that, rather than enforcing the path you drew. [UPOD: Desire Line]

Examples to study

• Real-world light cues: a raised door panel vs. an inset cabinet panel — the source of every UI shadow decision. [RUI]
• Well-designed sites’ typography — inspect and “steal from people who care.” [RUI]
• Asset sources worth knowing: Unsplash (free photography), Hero Patterns (subtle backgrounds). [RUI]
• (Add tsumugu surfaces here as worked good/bad pairs once the reader and dictionary chrome stabilize — see Front-End Web Design.)

Sources

• Wathan, A. & Schoger, S. Refactoring UI — raw/sources/design/Refactoring UI.pdf.
• Lidwell, W., Holden, K. & Butler, J. Universal Principles of Design, 3rd ed. (2023) — raw/sources/design/Universal Principles of Design.pdf.