Knowing that the nose shape is highly uncertain while the overall cranial proportions are well-determined is useful information for anyone interpreting the reconstructed face.

Knowing that the AFA3D reconstruction diverges from the manual ones in certain systematic ways, possibly due to the modern reference database, is useful information for anyone evaluating that method’s applicability to ancient specimens.

The ground truth problem means that facial reconstruction will always be approximate and uncertain.

But uncertainty, properly characterized and communicated, is not a failure of the reconstruction enterprise.

It is an honest statement of what is and is not knowable — which is exactly what the best science aspires to.