Conclusion

Neither clock is in question. Their shared bottleneck is foreign older material in ⇒ age biased old, differing only in detectability (hidden in radiocarbon, separable in zircon), and both yield to one protocol: classify on age-independent grounds, validate out-of-sample. Demonstrated on radiocarbon pairs, Lava Creek, and open zircon data, it recovers correct ages where material permits.

Neither clock is in question. Their shared bottleneck is foreign older material in ⇒ age biased old, differing only in detectability (hidden in radiocarbon, separable in zircon), and both yield to one protocol: classify on age-independent grounds, validate out-of-sample. Demonstrated on radiocarbon pairs and on Lava Creek and open zircon data, it recovers correct ages where material permits and states honestly where deep-time accuracy rests on consilience, not written-record ground truth.

Radiocarbon and zircon U–Pb are not in question as clocks. Their shared accuracy bottleneck is the incorporation of foreign older material, which raises measured ages and which differs between them only in detectability—homogeneous and hidden in radiocarbon, granular and separable in zircon. Recognising this lets one treat both with a single protocol whose central discipline is to classify on age-independent grounds and to validate out-of-sample. Demonstrated on real radiocarbon pairs, on the canonical Lava Creek zircon dataset and on open SHRIMP-RG and detrital data, the protocol recovers the right ages where the material permits and declares the cases where it does not. Its honesty is structural: it states the working range of each method, the assumptions each correction rests on, the residual ambiguity in attribution, and the one place where deep-time accuracy is a matter of consilience rather than of direct, written-record ground truth. That is the strongest claim the evidence supports—and it is the claim this paper makes.