Visual Design Research Consolidation

Research State Snapshot

• What this package is: a canonical synthesis of existing repository findings.
• What this package is not: a new literature review, new experiment, or new theory branch.
• Highest-confidence areas: perception is conditional; spacing is relational rather than universal; typography is multi-layered; color is contextual; hierarchy must align with task; accessibility is a structural constraint; semantics and layout ownership matter more than stylistic abstraction.
• Lowest-confidence areas: universal grid ratios; quantitative white-space bands; cross-medium motion laws; artist-style transfer rules; exact predictive weighting across typography, color, motion, and structure.
• Largest remaining unknown: how to turn the repository's validated mechanisms into practical predictive models for complex real interfaces without collapsing them into simplistic constants.

Executive Summary

Strongest consolidated position

The repository does not support a single style doctrine such as minimalism, maximalism, intuitive-by-default, whitespace-as-good, or component-reuse-as-good.

It supports a stronger model:

Effective visual design is the deliberate externalization of structure so that perception, attention, memory, prediction, and action align with the user's task, capabilities, and context.

This means visual design is best understood as a conditional control system, not as decoration and not as a library of timeless ratios.

Repository-wide recurring findings

1. No important visual variable is universally monotonic. More spacing, more contrast, more salience, more boldness, more density reduction, and more progressive disclosure can all help or harm depending on task and context.
2. Perceptual organization is the bridge layer. Grouping, separation, figure-ground, hierarchy, and cue competition connect low-level sensation to higher-level comprehension and action.
3. The active bottleneck governs value. The best intervention is the one that improves the process currently constraining performance.
4. Local improvements do not guarantee system improvements. Better glyph recognition may not improve reading. Better salience may not improve decisions. Cleaner spacing may reduce identification errors while weakening grouping.
5. Meaningful structure must exist across visual, semantic, and operational layers. A hierarchy that only works visually is fragile. Accessibility failures are theory falsifiers, not minor implementation bugs.
6. Design quality is relational. Goodness depends on user, task, environment, consequence, familiarity, and time.
7. Historical and artistic systems remain useful mainly as heuristics or observational taxonomies. They often mix real mechanisms with pedagogy, convention, and taste.

Consolidated operating model

The current repository supports this stack:

Physical rendering and environment
    ↓
Perceptual availability
    ↓
Grouping, segmentation, and discrimination
    ↓
Attention allocation and search
    ↓
Working memory and mental-model formation
    ↓
Interpretation and prediction
    ↓
Decision and action
    ↓
Learning, familiarity, adaptation, and habit

Every future interface should be evaluated at each layer, because success at one layer does not ensure success at the next.

Evidence Status Model

Scientific evidence

Repository claims count as scientific-evidence-backed when they derive primarily from psychophysics, vision science, cognitive psychology, neuroscience-adjacent perception work, ergonomics, human factors, HCI experiments, or standards-backed accessibility constraints.

Current examples:

• crowding depends strongly on eccentricity and configuration;
• grouping is multi-cue and competitive;
• attention is influenced by salience, goals, history, and learned expectation;
• working-memory limits are conditional rather than fixed constants;
• typography performance has threshold ranges and bottleneck dependence;
• familiarity changes subjective fluency and practiced performance;
• usability and legibility are task-, user-, and environment-relative.

Engineering best practice

These are stable implementation conclusions supported by evidence plus repeated production constraints.

Current examples:

• prefer native semantic HTML unless measurable needs justify replacement;
• keep intrinsic layout inside components and contextual layout in parents;
• preserve meaningful DOM order;
• use stable spatial organization before adaptive relocation;
• treat tokens as decision layers, not arbitrary variable bags;
• gate quality by accessibility, semantics, and consequence, not aesthetics alone.

Heuristics

These remain useful but should not be treated as laws.

Current examples:

• greater separation between groups than within groups;
• keep important state visible;
• use progressive disclosure to defer low-priority complexity;
• use distinctive landmarks at decision points;
• use a restrained number of attention peaks;
• keep a broad typographic safe range rather than one ideal value.

Opinions

The corpus contains some design preferences and project-level value judgments, but they are not the dominant basis of the theory.

Examples:

• visual redesign should remain highly flexible;
• generic cards are usually unstable abstractions;
• expressive quality matters and should not be reduced to utility.

Hypotheses

These remain active but unvalidated in integrated form.

Examples:

• externalized cognition as the unifying theory of composition;
• transfer alignment as the law governing typography's higher-level effects;
• relational legibility envelope as a predictive composition model;
• product legibility as a gated multi-stage interaction profile;
• color outcome models based on task-specific perceptual spaces.

Topic Synthesis

1. Perception, Human Vision, Neuroscience, Eye Tracking

Included topics: Perception, Human Vision, Neuroscience, Eye Tracking

What we currently believe

Perception is not a simple feed-forward pipeline. Visual organization is probabilistic, competitive, and conditioned by fixation, eccentricity, task, prior selection, learned expectation, and scene structure. Eye movements are part of the system, not just a measurement of it.

Supporting evidence

• gn-100-perception-autonomous-research-report-v1.md rejects linear and universal-spacing interpretations.
• composition-science-visual-density-crowding-and-perceptual-separation.md shows that identification depends on fixation-relative geometry and global configuration.
• project-atlas-typography-autonomous-research-report-v1.md and rp-atlas-typo-transfer-001.md show readers compensate through fixations, regressions, and inference.
• composition-science-phase-2-visual-hierarchy-and-wayfinding.md explicitly treats saccades, fixations, scan paths, and peripheral vision as core composition variables.

Competing viewpoints

• Simpler pipeline models remain attractive because they are easier to operationalize.
• Some design guidance still treats perception as mostly salience-driven rather than task-conditioned.

Confidence

High for conditional and fixation-relative perception. Moderate for any single unified model spanning all media.

Open questions

• Which eye-movement signatures best predict interface weakness?
• How can practical design reviews diagnose fixation-relative risk without lab eye tracking?
• Which findings transfer most cleanly from controlled arrays to complex interfaces?

Implications

Design cannot be judged from flat appearance alone. Interfaces must be tested as temporal perceptual systems under realistic viewing conditions.

Practical design guidance

• Evaluate what is visible near fixation and what must remain legible in the periphery.
• Reduce reliance on users noticing subtle relationships without guided attention.
• Treat scan cost, regressions, and resampling as evidence of design debt, not user failure.
• Do not convert laboratory thresholds directly into pixel rules.

2. Attention, Visual Hierarchy, Decision Making, Cognitive Load

Included topics: Attention, Visual Hierarchy, Decision Making, Cognitive Load

What we currently believe

Attention is a priority-allocation system shaped by salience, goals, history, reward, action opportunity, and expectation. Hierarchy is the ordered probability of attention, not merely a visual style. Every unnecessary decision consumes limited cognitive capacity.

Supporting evidence

• composition-science-phase-3-evidence-review-attention-wayfinding.md identifies attention as a multi-factor priority system and proposes hierarchy alignment, strategic visibility, and salience competition.
• composition-science-phase-2-visual-hierarchy-and-wayfinding.md frames hierarchy as an attention-allocation mechanism.
• project-atlas-autonomous-research-report-001-relational-legibility.md shows salience is competitive and task-conditioned rather than universally good.

Competing viewpoints

• Classic saliency-first thinking assumes the visually strongest item will usually be the correct attractor.
• Some minimalist design traditions over-index on removing stimuli rather than aligning priorities.

Confidence

High.

Open questions

• Can hierarchy misalignment be measured automatically?
• How many competing attention peaks are tolerable before search and decision costs rise sharply?

Implications

Hierarchy must be evaluated against intended behavior, not against surface neatness.

Practical design guidance

• Align intended, perceptual, behavioral, and task hierarchies.
• Emphasize few things at once; every added accent weakens the rest.
• Keep high-consequence choices visually obvious and semantically clear.
• Reduce avoidable decisions through defaults, visibility, and sequencing.

3. Gestalt Psychology, Grouping, Composition, White Space, Negative Space

Included topics: Gestalt Psychology, Composition, White Space, Negative Space

What we currently believe

Composition works by creating perceptual objects and relations. White space is not an independent good; it is one cue among many in a competitive grouping process. Negative space is meaningful when it changes grouping, separation, emphasis, perceived extent, or rhythm.

Supporting evidence

• gn-100-perception-autonomous-research-report-v1.md concludes that composition acts by creating perceptual objects.
• composition-science-research-library-v0-4.md rejects universal spacing bands and treats grouping as probabilistic.
• rep-atlas-0002-relational-legibility-envelope.md treats group coherence and individual discriminability as distinct constraints.
• project-atlas-applied-case-study-001-account-settings-form.md shows how uniform spacing can flatten hierarchy.

Competing viewpoints

• Traditional whitespace advice often assumes more empty space is generally better.
• Some grid-heavy systems treat composition as primarily a spacing-and-alignment exercise.

Confidence

High that grouping is multi-cue and probabilistic. Low for universal whitespace ratios.

Open questions

• Which grouping-cue combinations transfer best into dense application UIs?
• How should negative space be measured relative to semantic and action structure?

Implications

Spacing tokens alone cannot encode composition quality. Composition must model relationships, not only gaps.

Practical design guidance

• Use space to express relational differences, not cosmetic regularity.
• Distinguish local pairing, subgrouping, sectioning, and action separation.
• Use common region, alignment, labeling, and contrast to reinforce spacing when space alone is weak.
• Avoid uniform rhythms that erase semantic levels.

4. Layout Systems, Grid Systems, Information Architecture

Included topics: Layout Systems, Grid Systems, Information Architecture

What we currently believe

Layouts should externalize information structure and reduce navigation overhead. Grids are useful coordination systems, not universal perceptual laws. Information architecture governs whether users can construct a coherent mental model; layout is one expression of that structure.

Supporting evidence

• architecture-as-a-foundation-for-composition-science.md and composition-science-architecture-human-scale-and-proportion-research.md tie navigation, thresholds, progression, and human scale to digital composition.
• component-library-foundations-research-report.md separates semantic structure, composition patterns, and product compositions.
• 2026-07-21-semantic-durability-research-execution-package.md treats semantic durability and meaningful sequence as prerequisites for layout flexibility.

Competing viewpoints

• Some systems treat grid fidelity as a proxy for quality.
• Others assume layout can be fully decoupled from semantics through CSS alone.

Confidence

Moderate-high for semantic and structural primacy. Low for any one grid doctrine.

Open questions

• Which grid properties actually improve performance, versus simply improving production consistency?
• When should dense comparison views use different semantic models from narrative views?

Implications

Layout systems should be judged by how well they support grouping, scan efficiency, meaning, and adaptation across contexts.

Practical design guidance

• Use grids for coordination, not as a substitute for hierarchy.
• Preserve logical reading order and focus order.
• Prefer parent-owned layout relationships over component-owned page geometry.
• Treat IA shifts as structural changes, not as pure visual restyles.

5. Typography, Typography Genome, Readability, HTML Semantics

Included topics: Typography, Typography Genome, HTML Semantics

What we currently believe

Typography is a layered communication system spanning rendering, perception, discrimination, sequence recovery, semantic interpretation, navigation, and action. Glyph robustness is measurable, but local gains do not reliably transfer upward unless they affect the active bottleneck. Semantics matter because typography operates inside structured meaning, not as isolated shapes.

Supporting evidence

• project-atlas-visual-information-transfer-foundations-v1-2.md defines the layered channel model.
• project-atlas-typography-autonomous-research-report-v1.md rejects universal monotonic rules for spacing, size, weight, and width.
• rp-atlas-typo-transfer-001.md proposes the Transfer Alignment Principle.
• component-library-foundations-research-report.md and 2026-07-21-semantic-durability-research-execution-package.md show that semantic structure conditions typographic effectiveness in interfaces.

Competing viewpoints

• Strong form: typography quality can be reduced to information efficiency.
• Strong form: one best font or one best setting exists for a population.
• Strong form: beauty is fluent communication.

All were rejected or substantially weakened by the repository.

Confidence

Moderate-high.

Open questions

• How to diagnose the active reading bottleneck quickly in product settings?
• Which typographic improvements yield measurable task gains in UI contexts, not just reading labs?

Implications

Typography should be personalized, task-weighted, and consequence-aware rather than universally optimized.

Practical design guidance

• Design typographic systems as safe operating ranges, not fixed ideals.
• Strengthen glyph-level distinction when context is weak or errors are costly.
• Measure fatigue, regressions, and compensation, not just reading speed.
• Preserve semantic markup and meaningful structure so typography has coherent targets.

6. Color Theory, Contrast, Perception, Meaning, Artist Style Analysis

Included topics: Color Theory, Contrast, Artist Style Analysis

What we currently believe

Color should be modeled by mechanism and function rather than by traditional doctrine alone. Contrast is multidimensional. Historical color systems remain useful as observational and pedagogical frameworks, but they mix perceptual mechanisms, physical models, composition heuristics, and aesthetics.

Supporting evidence

• project-atlas-perceptual-color-genome.md establishes perception, attention, emotion, meaning, structure, context, material, and time as separate color branches.
• project-atlas-color-vocabulary-and-mechanism-taxonomy.md formalizes layered definitions and rejects the use of undifferentiated "contrast."
• rp-atlas-color-004-munsell-vs-modern-perceptual-color-spaces.md concludes Atlas should not choose one universal color space.
• project-atlas-autonomous-research-itten-seven-contrasts-v0-1.md reconstructs Itten's framework as mixed pedagogy rather than seven equal mechanisms.

Competing viewpoints

• Traditional hue-harmony systems imply stable universal relationships.
• Some engineering approaches reduce color choice to accessibility contrast thresholds alone.
• Some artistic traditions treat complementarity, warm-cool, or area balance as universal laws.

Confidence

High for contextual and multidimensional color. Moderate for practical UI prediction outside pairwise difference tasks.

Open questions

• Which measurement stack best predicts interface-scale outcomes?
• How should color-area, motion, and time-dependent changes be modeled in apps?
• Which learned meanings are durable enough to depend on across domains?

Implications

Color systems should separate physical measurement, perceptual appearance, semantic use, and design intent.

Practical design guidance

• Specify the contrast type you are using.
• Use color structurally for grouping, navigation, status, and priority, not only for decoration.
• Do not assume numerical color distance predicts functional UI difference.
• Treat artistic color doctrines as hypothesis generators and heuristics unless supported by mechanism-level evidence.

7. Visual Balance, Rhythm, Familiarity vs Intuition, Artist and Architectural Patterning

Included topics: Visual Balance, Rhythm, Familiarity vs Intuition, Artist Style Analysis, Architecture

What we currently believe

Balance and rhythm are real compositional effects, but current repository evidence supports them more strongly as emergent outcomes of structure, spacing, repetition, and expectation than as standalone measurable laws. Familiarity is a major source of perceived intuitiveness, but not the whole phenomenon.

Supporting evidence

• composition-genome-v1.md includes rhythm, balance, symmetry, asymmetry, repetition, and variety as composition systems.
• intuitive-is-just-familiar-research-report.md shows that intuition-like fluency often arises from familiarity, but structural compatibility and visible mappings also matter.
• architecture-as-a-foundation-for-composition-science.md and composition-science-architecture-human-scale-and-proportion-research.md show repeated spatial patterning, thresholds, compression, expansion, and human scale as transferable compositional mechanisms.

Competing viewpoints

• Formalist traditions often treat balance or rhythm as largely self-justifying aesthetic laws.
• "Intuitive" is often used as if it names inherent object quality.

Confidence

Moderate for balance and rhythm as emergent structural properties. High that familiarity materially affects fluency and preference.

Open questions

• Can balance or rhythm be operationalized without oversimplifying them?
• When do familiar but inferior patterns deserve replacement despite switching cost?

Implications

Design should separate practiced fluency from intrinsic structural quality.

Practical design guidance

• Use repetition and rhythm to stabilize expectations and scan flow.
• Do not treat user comfort with a pattern as evidence that the pattern is intrinsically good.
• When changing familiar systems, justify migration by benefits large enough to repay retraining and error costs.

8. Product Design, Industrial Design, Affordances, Feedback

Included topics: Product Design, Industrial Design, Affordances

What we currently believe

Product and interface design are both legibility problems. Affordance is relational, not merely visual. Good design coordinates discoverability, interpretability, executability, feedback closure, and consequence comprehension. Minimalism is not a universal law.

Supporting evidence

• product-genome-project-atlas-v1.md frames product design as the human-object-action loop.
• product-genome-research-execution-package-run-02.md operationalizes product legibility as a gated profile.
• intuitive-is-just-familiar-research-report.md shows that fluency and habit cannot repair dangerous mappings, hidden state, or poor feedback.

Competing viewpoints

• Visual simplicity is often mistaken for usability.
• Affordance is often collapsed into appearance or visual suggestion alone.

Confidence

Moderate-high.

Open questions

• How should product-legibility metrics map onto purely digital controls?
• Which dimensions normalize across physical and digital products?

Implications

Interface design should borrow human-factors rigor from industrial design rather than relying on style alone.

Practical design guidance

• Make action possibilities, consequences, and system response explicit.
• Add deliberate friction when consequence is high.
• Differentiate controls by operational importance, not just by aesthetic grouping.
• Evaluate first-use, practiced use, and failure recovery separately.

9. Motion, Progressive Disclosure, Time

Included topics: Motion, Progressive Disclosure, Time

What we currently believe

Motion is a hierarchy and attention variable, but the current repository contains stronger theoretical than empirical treatment. Progressive disclosure redistributes complexity rather than removing it. Time is a first-class design variable because perception, learning, adaptation, movement, and state change unfold over time.

Supporting evidence

• composition-genome-v1.md includes motion as a visual-encoding mechanism and progressive disclosure as a navigation mechanism.
• project-atlas-autonomous-research-report-001-relational-legibility.md concludes progressive disclosure trades visible clutter for discovery, navigation, and memory costs.
• product-genome-project-atlas-v1.md treats time as a product-design layer.

Competing viewpoints

• Motion is often treated as purely expressive.
• Progressive disclosure is often treated as an automatic simplification win.

Confidence

Moderate for progressive disclosure tradeoffs. Low-to-moderate for generalized motion laws.

Open questions

• Which motion patterns reliably improve hierarchy without inducing distraction?
• How much hidden complexity can users tolerate before overview collapses?

Implications

Any reduction in visible complexity must be balanced against increased search, recall, and navigation costs over time.

Practical design guidance

• Use motion to signal state change, hierarchy change, or causal continuity, not ornament alone.
• Prefer revealing low-priority detail rather than hiding high-consequence state.
• Evaluate disclosure patterns with no-JavaScript, keyboard, and low-vision scenarios.

10. Accessibility, Semantic Structure, Component Libraries, Design Systems

Included topics: Accessibility, Component Libraries, Design Systems, HTML Semantics

What we currently believe

Accessibility is a structural validity constraint. Component systems should encode durable meaning, intrinsic behavior, and controlled boundaries, not freeze presentational assumptions. Native-first, category-specific architecture is the strongest current position.

Supporting evidence

• component-library-foundations-research-report.md rejects a universal custom-element strategy and default Shadow DOM.
• 2026-07-21-semantic-durability-research-execution-package.md treats accessibility conflicts as falsification conditions for universal semantic models.
• project-atlas-autonomous-research-report-001-relational-legibility.md identifies accessibility as a structural falsification test.

Competing viewpoints

• Universal component abstraction as the route to reuse.
• Shadow DOM or tokens as primary redesign mechanisms.
• Accessibility as something that can be fully encapsulated inside components.

Confidence

High.

Open questions

• Which semantic-core models remain stable across product families?
• Where should encapsulation boundaries fall for rich, content-heavy structures?

Implications

Component libraries should be derived from evidence-backed content and interaction boundaries, not from shared surface appearance.

Practical design guidance

• Keep native semantics by default.
• Treat contextual layout as parent-owned.
• Avoid generic "card" abstractions unless they claim only surface behavior.
• Require manual review of reading order, focus order, and structural meaning.

11. Information Density, Wayfinding, Mental Models, Landmarks

Included topics: Information Density, Wayfinding, Decision Making, UX Research

What we currently believe

Dense displays are not bad by definition. Density becomes harmful when it degrades grouping, search, discrimination, or decision quality. Wayfinding succeeds when local decisions reinforce a coherent global model and important landmarks are visible, stable, and meaningful.

Supporting evidence

• composition-science-phase-3-evidence-review-attention-wayfinding.md identifies decision-point landmarks, strategic visibility, and externalized memory.
• composition-science-architecture-human-scale-and-proportion-research.md frames navigation overhead as a subtraction from task capacity.
• composition-science-visual-density-crowding-and-perceptual-separation.md rejects element count as a sufficient measure of clutter.

Competing viewpoints

• Simpler-looking interfaces are often assumed to be cognitively simpler.
• Dense analytical views are often criticized on aesthetics rather than on task fitness.

Confidence

High that density and clutter are not reducible to count alone.

Open questions

• Which density measures best predict failure in dashboards, registries, and comparison views?
• When should analytical density outweigh narrative clarity?

Implications

Information density must be matched to task, expertise, and consequence rather than minimized by default.

Practical design guidance

• Keep destinations, status, and progress visible when they reduce uncertainty.
• Use distinctive landmarks at decision points.
• Tune density separately for novices, experts, and high-frequency tasks.
• Distinguish overview loss from genuine simplification.

12. UX Research, Engineering Method, Additional Emergent Disciplines

Included topics: UX Research, Cognitive Psychology, Human Factors, Research Methodology, Additional disciplines

What we currently believe

The repository's strongest methodological principle is that design recommendations must not exceed the evidence supporting them. Cross-disciplinary synthesis should preserve contradictions and boundary conditions instead of flattening them into slogans.

Supporting evidence

• composition-genome-v1.md explicitly defines an evidence hierarchy.
• rep-atlas-0002-relational-legibility-envelope.md and product-genome-research-execution-package-run-02.md convert theory into gated and conditional models rather than universal scores.
• Nearly every recent report preserves invalidated hypotheses, contradictions, and confidence levels.

Competing viewpoints

• Fast product work often rewards premature simplification.
• Design systems often promote universal rules before they are validated.

Confidence

High.

Open questions

• How should the repository encode validated versus provisional guidance in component outputs?
• What minimal experiment set is required before promoting a design rule into implementation defaults?

Implications

Future interface work should be evidence-weighted, consequence-aware, and explicit about uncertainty.

Practical design guidance

• Tag every design rule as evidence-backed, heuristic, opinion, or hypothesis.
• Keep invalidated ideas visible so they do not re-enter through folklore.
• Prefer falsifiable design claims over taste-based declarations.

Recurring Patterns Across Disciplines

Pattern 1: Conditionality beats universality

Across typography, spacing, color, hierarchy, product legibility, and semantics, the project repeatedly rejects one-size-fits-all laws.

Pattern 2: Relational structure matters more than isolated values

Spacing ratios, contrast relationships, semantic order, action consequence, and task alignment are more predictive than absolute values alone.

Pattern 3: Competing constraints are normal

Grouping competes with discrimination. Visibility competes with clutter. Stability competes with adaptation. Simplicity competes with capability.

Pattern 4: The observer is part of the system

Age, expertise, familiarity, language, impairment, fatigue, and prior learning change the effective interface.

Pattern 5: Surface neatness can conceal deep weakness

Uniform spacing, attractive hierarchy, high speed, or strong brand coherence can hide semantic, accessibility, or consequence failures.

Pattern 6: Externalized cognition is a strong candidate unifier

The best recurring explanation across architecture, interfaces, typography, and product design is that good design moves burden from internal cognition into the environment.

Pattern 7: Measurement must be multidimensional

Single scores repeatedly fail because severe weakness in one critical dimension can be masked by strength elsewhere.

Pattern 8: Historical systems contain real observations plus contamination

Color masters, style traditions, and familiar patterns remain useful, but they often mix mechanism, craft, and convention.

Fundamental Principles Repeated Across the Repository

Visual Design Principles

1. Design for the active bottleneck, not for abstract perfection.
2. Express relationships relationally: within-group, between-group, primary, secondary, reversible, dangerous, persistent.
3. Preserve agreement across visual hierarchy, semantic structure, focus order, and task flow.
4. Externalize memory, orientation, and state whenever doing so reduces decision cost.
5. Treat salience as a scarce resource; every added peak creates competition.
6. Optimize for safe operating ranges, not universal maxima.
7. Separate grouping from discrimination and verify both.
8. Prefer stable location and identity before adaptive relocation.
9. Use native semantics and controls by default; replace only with evidence.
10. Evaluate designs as time-based systems including learning, habit, fatigue, and recovery.
11. Match density to task, expertise, and consequence instead of minimizing it by reflex.
12. Promote a rule only when its evidence, limits, and contradictions are explicit.

Anti-patterns

• Treating more whitespace as a universal cure.
• Converting research into universal pixel, ratio, or token rules without boundary conditions.
• Making everything prominent.
• Hiding high-consequence state behind progressive disclosure.
• Using generic surface abstractions where semantic roles differ.
• Assuming accessibility can be retrofitted after visual design.
• Treating familiarity as proof of quality.
• Using single summary scores that hide catastrophic failure modes.
• Letting components own page-level layout and spacing relationships.
• Replacing native controls without measurable justification.
• Treating color distance, type ranking, or grid fidelity as sufficient quality metrics.
• Confusing expressive motion with informative motion.

Design Evaluation Checklist

1. What is the user's primary task, and which layer is most likely to bottleneck it: perception, search, interpretation, action, memory, or learning?
2. Does the perceptual hierarchy match the task hierarchy?
3. Are related items grouped strongly enough without compromising individual discrimination?
4. Are high-consequence actions, states, and distinctions visible and interpretable without relying on memory?
5. Does the interface remain structurally correct in DOM order, focus order, and accessible naming?
6. Is salience concentrated on genuinely important elements, or is everything competing?
7. Are spacing, typography, contrast, and density operating inside safe ranges rather than at stylistic extremes?
8. Does progressive disclosure reduce complexity, or merely move it into recall, search, and navigation?
9. Would an unfamiliar, low-vision, fatigued, or interrupted user still succeed?
10. Does expert speed depend on compensation for weak design rather than on genuine legibility?
11. Are layout relationships parent-owned where they should be, or trapped inside components?
12. Are native controls and semantics being preserved unless a stronger option has been proven?
13. Are color and contrast choices specified by function and mechanism, not by vague aesthetic vocabulary?
14. Is motion conveying causality, hierarchy, or state, rather than adding noise?
15. What evidence tier supports each major design choice: scientific, best practice, heuristic, opinion, or hypothesis?

Missing Research

High-priority gaps

• Practical diagnostics for active bottlenecks in real interfaces.
• Cross-disciplinary quantitative models that join grouping, attention, typography, and action without collapsing their differences.
• Interface-native validation for color measurement stacks.
• Better operational measures for compensation cost, fatigue, and cognitive load.
• Motion research specific to hierarchy, transition comprehension, and distraction tradeoffs.
• Grid and layout evidence that separates production convenience from perceptual benefit.
• Stronger research on dense analytical interfaces, not only simplified consumer flows.

Medium-priority gaps

• Artist-style analysis translated into mechanism-level variables rather than admired examples.
• Robust models of balance and rhythm.
• Cross-script and multilingual extension of typography findings.
• Better population models for aging, low vision, and impairment in composition systems.
• Longitudinal validation of familiarity, migration cost, and interface habit disruption.

Structural repository gaps

• A canonical registry separating validated principles from provisional heuristics.
• Cross-reference maps from component guidance back to evidence packages.
• Shared measurement templates for layout, color, typography, and accessibility evaluation.

Final Position

The repository's current visual-design theory is already more rigorous than most production design doctrine, but it is not a finished universal system.

Its strongest contribution is not a style language.

It is a way of thinking:

• conditional rather than absolute;
• relational rather than scalar;
• structural rather than decorative;
• consequence-aware rather than taste-led;
• and explicit about evidence, contradiction, and uncertainty.

That should govern every future interface produced from this project.