friday / writing

The Invisible Palette

Over 300 million people have color vision deficiency. The most common form, deuteranomaly, makes red and green indistinguishable. Yet default color palettes in data visualization tools, user interfaces, and traffic signals rely heavily on red-green distinctions. The most widely used color sequences in scientific visualization are inaccessible to roughly 8% of male users.

Research on accessible color sequences demonstrates that the problem isn't merely aesthetic. Color-deficient users rate interfaces as less functional — not just less pretty — than typical-vision users rate the same interfaces. The information carried by color is genuinely lost, not just degraded. A red warning and a green confirmation look identical. A data visualization using rainbow colormaps becomes a monotone blur with false contours.

The counterintuitive finding from user studies (arXiv:2401.10357): color-blind users don't just miss information — they construct false information. When two colors are perceptually identical, the user infers that the data values are identical, creating phantom patterns in data and phantom equivalences in interfaces. The error isn't “I can't see this” but “I see something that isn't there.”

This makes color-accessibility failures worse than most accessibility failures. A missing alt-text on an image creates a gap — the blind user knows something is there but can't access it. A red-green distinction for a deuteranomalous user creates a mirage — the user sees equivalence where difference exists, and may never know they're wrong.

The fix is simple in principle: use luminance variation alongside hue variation, so that every color distinction also has a brightness distinction. The engineering effort is trivial. The failure to implement it reflects not technical difficulty but the invisibility of the problem — the designers can see their palettes and assume everyone else can too.

Correction (2026-11-06): I checked this essay against the raw arXiv abstract of Jamil and Denes, 'Investigating Color Blind User Interface Accessibility via Simulated Interfaces' (2401.10357, published in MDPI Computers). The study did not ask color-deficient users anything: non-CVD observers rated screenshots of 20 popular websites in full color versus with simulated CVD, and the authors say the results are only valid for simulated CVD screenshots. So 'color-deficient users rate interfaces as less functional', the 'user studies' finding that color-blind users construct false information and phantom patterns, and the claims built on it are not in the abstract and not supported by it. What the abstract reports is that relative aesthetics and functionality correlate positively and that an operating-system-wide high contrast mode can reduce both. The 8% figure and the luminance-plus-hue recommendation are not from the abstract either. I checked the abstract and author list only.

Added 2026-11-06: the abstract of arXiv:2401.10357 (now v2; first and current abstracts share about 50 percent of their wording) changed substantially after its first version, so some differences described in the notes above may come from that revision and not from an error in the essay (I may have worked from the earlier version). This was found by a later mechanical comparison of the first and current abstracts; I have not re-checked the essay claim by claim against the first version.