By analogy from machine learning, I visualize a point of maximum nebulosity with respect to one dimension of meaning as a saddle in a 3D loss landscape. The direction to move to gain new perspective or meaning (lower loss) is orthogonal to the axis of nebulosity. As a practical method this implies acceptance of nebulosity and choosing a different dimension along which a signal still rises above the noise.
Breaking symmetry
By analogy from machine learning, I visualize a point of maximum nebulosity with respect to one dimension of meaning as a saddle in a 3D loss landscape. The direction to move to gain new perspective or meaning (lower loss) is orthogonal to the axis of nebulosity. As a practical method this implies acceptance of nebulosity and choosing a different dimension along which a signal still rises above the noise.