The exponential decay model established by Allentoft et al. remains the field's most cited constraint on ancient DNA (aDNA) recovery, and subsequent replication studies have not meaningfully extended the theoretical survival window. Public de-extinction efforts (e.g., proxy-based approaches such as Horner's atavistic reversion work in Aves (2)) have instead pursued developmental reversion in living descendant lineages rather than direct ancient sequence recovery, on the working assumption that no viable template exists for organisms predating the Allentoft ceiling by tens of millions of years.
Division 01 does not dispute this ceiling. Rather, CSP treats surviving fragment remnants — however degraded, however sparse — as anchor points within a probabilistic reconstruction model, rather than as a complete sequence to be assembled. Where Allentoft's model predicts total information loss, our working hypothesis holds that structural and regulatory information can be partially inferred from what survives, given a sufficiently close living bracket.
Fragment remnants recovered from site material undergo standard extraction and library preparation (see Appendix A, restricted). Recovered fragments are aligned against a composite reference constructed from extant archosaur genomes. Gaps are filled using a proprietary inference model (internal designation: CSP-3; methodology withheld pending patent review) that weights regulatory region reconstruction over coding-region completeness, on the premise that viable phenotype expression depends more heavily on regulatory fidelity than on complete coding sequence recovery.
Nine successive generations of stabilized lines have been produced under CSP since program inception. Structural viability has improved generation-over-generation; phenotypic divergence from projected baseline morphology has narrowed accordingly. XXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXX
One consistent and unexplained finding recurs across all nine generations: regulatory regions associated with neurological and sensory development in the bracket reference genomes show a higher-than-expected rate of successful inference relative to musculoskeletal and integumentary regions, despite no methodological weighting toward this outcome. Division 01 has not identified a technical explanation for this pattern. See Division 02 correspondence for behavioral follow-up.
CSP represents, to our knowledge, the only method to date producing stable, multi-generational specimen lines from material predating the aDNA survival ceiling. Further characterization of the neurological-regulatory inference bias described above is recommended before any Stage 3 conditioning work proceeds on affected lines.