# Carbon-biomass draw methodology
This is a game using estimated carbon biomass, not individual counts or real birth probabilities. There are 72 non-overlapping result categories; plants remain one group. Viruses are excluded.
The parent-group weights use the authors’ unrounded 2018 carbon estimates: 549.0807681063422 Gt C in total, including 0.05537256615 Gt C for humans. The human chance is about 0.0100846%, or 1 in 9,916. Digits support reproducibility, not that level of ecological precision.
Wild land mammals are split into 40 species plus a residual group, preserving the 2018 parent budget and using 2023 species wet-mass shares. This hybrid model assumes equal carbon-to-wet-mass ratios within that split; it is an approximation. Marine protists and krill subdivisions preserve their parent budgets. Scientific uncertainty and survey coverage remain.
A group photo illustrates the group, not the odds of the individual pictured species. There are 50 CC0/public-domain images, 20 CC BY images and 2 AI concept illustrations. Credits, license URLs and crop notices remain on the result and share image.
Sources:
- Bar-On, Phillips & Milo (2018): https://doi.org/10.1073/pnas.1711842115
- Author data and MIT license: https://github.com/milo-lab/biomass_distribution
- Greenspoon et al. (2023): https://doi.org/10.1073/pnas.2204892120
- Mammal author data: https://zenodo.org/records/7594807
Draw weights are identical to the Korean edition. Excluding the current result re-normalizes all remaining weights; the ID and OG image still show the original full-pool probability.