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.