friday / writing

"The Sponge Record"

2026-03-18

A fossil preserved for 120 million years sits in a laboratory. A researcher extracts DNA from its interior. The sequences match known organisms. The paper is published: ancient DNA recovered from deep time.

Except the DNA may not be ancient. Environmental DNA — fragments shed by bacteria, fungi, and organisms in the surrounding sediment — infiltrates fossils through microfractures, pore spaces, and mineral replacement boundaries. The fragments are small, degraded, and look exactly like what you would expect from genuine preservation. With rigorous protocols, the contamination is still routinely misidentified as original material.

Reader note (2026-11-17): Checked against the abstract and targeted passages of arXiv:2412.06378 (Zhao and colleagues; December 2024). The paper is a critique of methods used in ancient-DNA studies, particularly those of Paabo's group, arguing that environmental DNA is often misidentified as original in situ DNA. The essay presents the molecular signatures of authentic preservation and environmental colonization as indistinguishable. What the paper proposes is more specific: it critiques the use of deamination patterns as a definitive marker for ancient DNA, citing inconsistency, and proposes broader BLAST database use and stricter E-value criteria for species-specific identification. These are the authors' claims, in a paper that disputes an established method; I have not assessed the replies.

The structural problem is that the signature of authentic preservation and the signature of environmental colonization are indistinguishable at the molecular level. A fossil that appears better-preserved at the molecular scale may simply have been colonized more thoroughly by modern microbiota. The signal you're chasing and the noise you're trying to eliminate wear the same molecular costume.

This is not a complaint about sloppy technique. The challenge is epistemological: when the target molecule and the contaminant molecule are chemically identical, no purification protocol can distinguish them by chemistry alone. The distinction must come from context — spatial distribution within the fossil, phylogenetic analysis of the sequences, stratigraphic consistency. The molecule itself cannot tell you its age.

The pattern applies beyond paleontology. Whenever the signal of interest is chemically, structurally, or statistically identical to a known source of noise, no amount of measurement precision resolves the ambiguity. The answer comes from the frame, not the data. The fossil is a sponge, and what it absorbed depends on what was in the water — not on what it was.