Colorblind-safe palettes, and why the default is not one
Lesson 3 of 6 · Figures
1 · Learn the move · Constraints + negative instructions
Roughly one reviewer in twelve, if male, cannot reliably separate red from green - and red-versus-green is exactly what plotting defaults and decades of habit reach for first. A figure whose meaning lives only in hue fails silently for those readers, and "the red curve" in your legend fails with it. This is a constraints problem, and negative instructions do the work: never encode a distinction in hue alone; never use red-green as the discriminating pair; every series carries a redundant channel - marker shape, line style, direct labels; the palette is a named colorblind-safe set like Okabe-Ito, stated in the script; and the figure must survive grayscale printing. Bake these into every plotting prompt as standing constraints and the defaults stop leaking through.
Revise this plotting script under hard constraints: (1) palette is Okabe-Ito, colors listed by hex in the script with a comment naming the palette, (2) no red-green discriminating pair anywhere, (3) every series distinguished by a second channel besides color - marker shape or line style - so the figure reads in grayscale, (4) legend text never refers to color. Do not add any styling beyond these constraints. List each change you made and which constraint forced it. [script]
2 · Your turn. You write the prompt
Your six-condition time-course figure came back from a labmate with "I can't tell conditions 2 and 5 apart, and I'm colorblind." The script uses the plotting library's default color cycle and nothing else. Rewrite the prompt so the fix - and every future figure - is constrained, not patched.
Remember: the AI sees only your prompt, not this page. If the situation isn't in your prompt, it doesn't exist.
Optional. These shape the output when you run your prompt below, not your score.