CONCEPT

The Encoded Color Line

Du Bois's prophecy that the color line would remain the master problem of the century, now instantiated in machine-learning systems that reproduce racial inequality through mathematical proxies while offering the alibi of objectivity—the line that did not vanish in the twenty-first century but went into the code.

The color line, Du Bois announced in 1903, was the problem of the twentieth century. He meant something structural: not individual prejudice but a self-reproducing arrangement by which advantage and disadvantage were sorted along the axis of race and made to seem natural. The crucial insight was systemic: the line did not require individual malice to persist. It was maintained by institutions, laws, habits, and economies that converted historical injustice into present fact and present fact into future inevitability. Racism, for Du Bois, was less a feeling than a machine. The encoded color line is what happens when that machine is distilled into a scoring function. The most common defence of algorithmic decision-making is that it removes human prejudice: the loan officer might be a bigot, but the model is just math. Du Bois's systemic analysis exposes the sleight of hand. If the historical loans, hires, and

Keep reading with YOU ON AI