Orthogonality: Why Should Good Dimensions Not Quietly Replace One Another?

Orthogonality originally describes perpendicular relations in geometry. More broadly, it means that different dimensions can vary separately: adjusting one does not quietly substitute for another. Its value is not isolation, but making the source of a change identifiable.

On a camera, aperture mainly controls incoming light and depth of field, while the focus ring determines which distance appears sharp. Both still contribute to the same photograph. But if turning the focus ring also changed exposure dramatically, the photographer would struggle to tell what produced the result. Well-designed controls keep the two questions as separate as practical.

Orthogonality does not mean having no relationship; aperture and focus remain parts of one imaging process. Nor is it the same as modularity, which concerns how parts can be separated and recombined. Orthogonality concerns whether dimensions of change are independent. In discussion, data analysis and AI evaluation, treating fluent answers and factual reliability as one scale shows a lack of orthogonality. A good conceptual framework does not make the world simple, but it prevents one change from quietly replacing another judgement.