Scarcity, then choice

For most of human history, color was hierarchy made visible. Lapis lazuli ground into ultramarine was reserved for painting the Virgin Mary's robe because the pigment cost more than gold. If you could afford the dye, you could afford the authority.

William Henry Perkin, an eighteen-year-old chemistry student trying to synthesize quinine in 1856, accidentally produced mauveine, the first synthetic aniline dye. Within a decade, anyone could wear purple. Color stopped being a gate and became a choice.

Choices need organizing. By 1963, Pantone had built a standardized matching system that turned subjective sensation into reproducible specification. Color could be owned, not just worn. Tiffany & Co. trademarked a robin's-egg blue, Pantone 1837, named for the company's founding year. The color preceded the product. It preceded the logo. When LVMH acquired Tiffany for $15.8 billion in 2021, they changed the advertising, the creative direction, the celebrity partnerships. They did not touch the blue. The blue was the acquisition.

What optimization actually produces

Between roughly 2015 and 2022, a generation of direct-to-consumer brands made the opposite bet. Glossier, Casper, Hims, Outdoor Voices, Brandless. They all converged on nearly identical palettes: soft pink, muted lavender, pale sage, warm neutrals. These weren't independent creative decisions. They were the output of the same Instagram conversion metrics, the same consumer research pointing toward "approachable" and "friendly."

Individually, the palettes performed. Collectively, they produced what the design press called "blanding." Optimization converges on the median. The median, in color, is invisibility.

Color is the fastest signal a brand sends, faster than copy, faster than product photography. When every competitor occupies the same narrow band of the visible spectrum, that speed advantage evaporates. Color chosen by committee, by data, by what tests well, stops being color. It becomes camouflage.

The algorithm's ceiling

Run the same prompt four times in Looka, Brandmark, or Canva's brand generator. Ask for a skincare brand targeting millennial women. You will get four different logos, four different type systems, four different layouts, and almost the exact same color palette. Muted. Soft. Safe.

The algorithm synthesizes the statistical center of what has already worked. Layout follows spatial rules. Typography follows hierarchical logic. Copywriting follows linguistic patterns. All of these can be systematized.

Color resists this. The same Pantone blue reads as clinical next to white, electric next to orange, melancholy next to grey. It reads as trustworthy in Chicago and cold in São Paulo, differently to a twenty-two-year-old in Lagos than to a fifty-five-year-old in Oslo. Research from the University of Oregon and the University of Cincinnati found that simply changing a grocery brand's logo to blue-green tones made consumers judge the company's ethical behavior more favorably, even when the company's actions were ambiguous. Color constructs the perception of values in real time, below conscious awareness.

An algorithm can generate a palette. It cannot tell you whether your brand should feel like dried terracotta or wet clay. That distinction is emotional, cultural, and irreducibly specific. It requires someone to sit with the discomfort of choosing something that cannot be validated by data. Only by conviction.

Coca-Cola has held red for over a century, not because red tested better than blue, but because red was a declaration that the brand knew itself and refused to negotiate. Pepsi has cycled through more than a dozen identity shifts in the same period, chasing relevance through reinvention. The market reads the difference instantly. Confidence is the one thing a generative model cannot hallucinate into existence.

The fragility underneath

Cadbury tried to trademark its signature purple in the UK and lost a protracted legal battle with Nestlé in 2013. The court ruled that a corporation cannot own a color in the broad sense Cadbury wanted. Consumers disagreed. Cadbury purple remains Cadbury in the collective imagination regardless of what the law permits.

Color's power does not come from the brand that claims it. It comes from the millions of people who agree, without being asked, to associate a wavelength of light with a feeling. That agreement is cultural, not proprietary. It can be earned. It cannot be manufactured. And it cannot be automated. Every brand that adopted millennial pink in 2017 borrowed a feeling without earning it, and the feeling dissipated the moment it became ubiquitous. The moment a color becomes a trend, it loses the specificity that made it meaningful.

What's left after generation

AI has compressed the cost and time of every other identity element toward zero. The logo is minutes away. The type system is a slider. The brand voice is a prompt. What remains is the one decision that still costs something psychologically: the willingness to commit to a color that says who you are at the risk of alienating everyone it doesn't speak to. Every other element of brand identity can now be produced without confronting the fundamental question. Color still demands the answer. Not because machines lack the capacity to suggest hues, but because a color chosen without conviction is just decoration, and consumers feel the difference before they can name it.

The twelve thousand snails that died for a Roman emperor's robe were never about purple. They were about paying a cost that proved the claim was real. The next time you scroll past a brand and feel nothing, look at the palette. You will see exactly where the conviction ran out.

If color, conviction, and the things branding can't automate are your kind of problem, subscribe to The Threads of Culture newsletter.

Keep Reading