Work

Retrospective Drift in the Mineral Record: code and data pipeline

Year: 2026
Type: article
Abstract: Version: 3 (19 November 2026) Authors: Kriger, Boris — Information Physics Institute; Institute of Integrative and Interdisciplinary Research. ORCID 0009-0001-0034-2903 Description:Scripts and results for version 2 of B. Kriger, “Retrospective Drift in the Mineral Record: Separating Survivorship from Direction in 4.5 Billion Years of Mineral Evolution” (2026). The article uses the Mineral Evolution Database as a test bed for the Retrospective Drift Theorem (https://doi.org/10.13140/RG.2.2.26532.64648): a record observed conditional on persistence acquires an apparent directed drift whose size is fixed by the persistence filter. This package reproduces the analyses new in version 2 and the re-implemented baseline quantities. Quantities carried over unchanged from version 1 (Figures 3–7, Tables 2, 4 and 8) are not recomputed here. Contents: complexity.py: formula parser computing Shannon chemical complexity per formula unit after Krivovichev; stage_test.py: the 1.80 Ga complexity jump with multi-stage localities and late-stage phases removed, step scans and cluster bootstraps over localities (D7); classes_jump.py, paired_redox.py: chemical decomposition of the jump, including sulfates against sulfides at shared localities (D8); lith_ratio.py: occurrences per unit volume of matching Macrostrat lithology, North America, Phanerozoic and full record (D9); lith_ratio_map.py: the same test under four denominators (column volume, surface units, subsurface units, mapped outcrop area of the State Geologic Map Compilation) and in fractional form, the share of a mineral class over the share of its host lithology (D9, step 19); proterozoic_pulse.py, pulse_geography.py: composition, geography and concentration of the 1.9–1.8 Ga Macrostrat pulse; fig_age_quantity.py, figs_new.py, fig_d9_denominators.py: Figures 2 and 8–11; results/: the JSON outputs quoted in the article. Inputs (not redistributed; retrieval described in README.md): the MED snapshot of 3 February 2020 (209,993 dated mineral–locality pairs, 4,786 species, 26,122 localities), object med_data_cache in R/sysdata.rda of the dragon package (Spielman & Moore 2020, https://doi.org/10.3389/feart.2020.585087), CRAN archive dragon_1.2.1; rock units from the public Macrostrat API, retrieved 18 November 2026; SGMC map-unit areas from the Macrostrat API (source 133), retrieved 19 November 2026; Natural Earth 1:50m country boundaries. Requirements: Python 3 with numpy, pandas, matplotlib, pyreadr and shapely. Run the steps listed in README.md. Random seeds are fixed. The Macrostrat API is live, so rock quantities may differ slightly between retrievals. All scripts in this package reproduce the published values on the 18–19 November 2026 retrievals. Keywords: mineral evolution; Mineral Evolution Database; survivorship bias; persistence conditioning; retrospective drift; Macrostrat; macrostratigraphy; host-matched ratio; geologic map; mineral complexity; Shannon information; Nuna supercontinent; sulfidic ocean License: MIT Related identifiers: Is supplement to: https://doi.org/10.13140/RG.2.2.26532.64648 (the theorem) References: https://doi.org/10.3389/feart.2020.585087 (dragon / MED snapshot) References: https://doi.org/10.1029/2018GC007467 (Macrostrat, Peters et al. 2018) References: https://doi.org/10.3133/ds1052 (SGMC, Horton et al. 2017)
Language: English
Open Access status: green
OpenAlex ID: W7233941167
FWCI:
Cites:
Cited by:
Related to: 10