Most of the fish stocks caught in the world have never been properly assessed. The ones that have tend to belong to wealthy countries with long research programmes and survey vessels to spare. Everywhere else, the record of a fishery amounts to a catch series, and often an incomplete one. CMSY was built for that second group: it estimates what a stock can sustainably produce from its catch history and the resilience of the species, without the survey data the classical models demand.

In May I was at the GEOMAR Helmholtz Centre for Ocean Research in Kiel for the second review panel on global stock assessments, from 11 to 14 May. The Sea Around Us is bringing its catch data and its assessments up to 2022, and the four days went to CMSY and its Bayesian companion BSM: how each behaves on some of the most commercially important stocks in the world, where a prior needs tightening, where the fit is honest, and where the answer should not be trusted at all.

One finding keeps coming back in this work, and it is not a comfortable one. The simple, data-poor models often land closer to reality than the heavily parameterised official assessments, which carry enough moving parts to be tuned towards optimism. Quantitative Aquatics wrote up the meeting, and Rainer Froese put that point plainly there.

This is what makes the method worth the argument for the countries I work on. In the southern Mediterranean and around the Arabian Peninsula, a large share of what is caught never enters the official statistics: small-scale landings sold straight off the beach, subsistence and recreational fishing, discards thrown back out of sight. Reconstruct that catch and the shape of the stock’s history changes. Run CMSY on the reconstruction rather than on the reported series, and the verdict on the stock can change with it.

The method is set out in Froese et al. (2017) and has since been revised as CMSY++. The video tutorials are free for anyone who wants to run it, and the Sea Around Us posted from Kiel.