Vines 2014 — 19% of 516 ecology dataset requests delivered

Vines, Albert, and colleagues attempted to retrieve raw morphological data underlying 516 papers in ecology and evolutionary biology, sampled at two-year intervals from 1991 through 2011. They received only 19% of the requested datasets. Among authors who responded, the odds of a dataset still existing fell by approximately 17% per year after publication.

The study established the empirical method (direct contact for raw data) that subsequent measurements (E-0002, E-0003, E-0004) replicated across other disciplines and decades. The 17%-per-year decay is the time dimension of the baseline established in C-0002: data does not survive at a constant rate; it survives less, year over year, on the architecture most papers in the cohort were stored on.

The paper's interpretation focused on operational explanations (researcher behavior, institutional policy). The architectural reading developed in C-0001 reinterprets the same finding as a structural consequence of the storage substrate the cohort was using.