Sources and record of changes

SPRFMO

South Pacific Regional Fisheries Management Organisation
Jack Mackerel Candidate Management Procedures
Sources, Review Status, and Record of Changes

1 Sources used to create this site

Source Exact saved version used Material used here
SPRFMO/jmMSE isolated 500-draw refinement worktree 77968e4a9febcad5315a7c00419b6d1690e3e102; the release check also reports whether tracked changes remain Primary source: completed 500-posterior reference and nine-OM robustness runs for the eight CMP comparison cases
SPRFMO/scw17 0eb97c9c0329f5ec3338e3cf885f6e1f184c4780 Secondary source: Papers 03, 04, 08, and 09 plus Quarto conventions
output/jm_candidates_500.slick SHA-256 eac4d67ed02884b833397b601f5406e17c08cfe31cb507db2637a88ff41eddf5 Combined 500-posterior reference and nine-OM robustness results for eight CMPs, including VB/VB[2025], VB/VB[MSY], and pooled 2041–2050 low-catch probability

The named package is SC14-MSE-2026-RC1. Its automatic agreement report provides the complete run list, file list, current repository revisions, and the result of the numerical comparisons.

The current registry derives from jmMSE, not the SCW17 test table. MP43 (HS-20) and MP44 (PR-20) are the focal analytical cases. MP29 (HS+20) and MP32 (PR+20) are sensitivity cases retained to show the effect of the earlier annual-limit specification. MP45–MP48 are additional comparison cases. This analytical designation is not a formal recommendation or adoption decision. All eight procedures were refined against the same 500 posterior draws before the current reference and robustness evaluation.

2 Information that must accompany each CMP

Every CMP version must link to:

  • the method used to calculate its indicator, harvest control rule, and catch allocation;
  • all parameter values and annual data requirements;
  • the exact saved version of jmMSE used for evaluation;
  • OM and robustness scenario definitions;
  • definitions of the performance measures and links to the saved results;
  • review findings and known limitations; and
  • the decision that changed its status.

3 Types of change

Class Examples Required review
Wording only Spelling, navigation, or clearer background explanation Documentation review
Same calculation, revised code Software change shown to give the same answers Code review plus a repeat-calculation test
Scientific Parameter, index, window, HCR, constraint, allocation, or fallback change Renewed scientific review and normally MSE evaluation
Status Test to shortlist, recommendation, adoption, or retirement Cited decision record

4 Current status and remaining work

Item Status Evidence or remaining action
Assign analytical roles and stable identifiers Complete for the current analysis MP43/MP44 are focal cases; MP29/MP32 are sensitivity cases. Legacy MP identifiers remain the stable links to code and saved results. Formal recommendation or adoption remains a separate decision.
Complete common 500-draw tuning and robustness evaluation Complete All eight CMPs were refined against the same 500 draws and evaluated under the reference OM and nine robustness OMs.
Check performance at small trigger changes around each final value Complete Every CMP was evaluated one grid step below and above its selected trigger using the same saved 500 draws: 0.0625 for hockey-stick rules and 0.03125 for power-ramp rules. All eight local profiles were monotonic; see the saved summary. In plain language, this checks how much the result changes after a small move on either side of the chosen trigger.
Record full run objects with the Slick file and fingerprints In progress The named release now has a run index, a file list, and an automatic comparison report. Publication remains incomplete until the reporting revision is committed and the tracked changes in the analysis worktree are resolved.
Finalize operational specifications Partial Candidate controls and tested data windows are documented. Authoritative units, annual cutoffs, missing-data rules, rounding, and the formal exceptional-circumstances provisions still require approval and completion.
Save worked examples and automatic repeat-calculation checks Partial A saved retrospective worked example and reproducible report/Slick build scripts exist. Independent automatic repeat-calculation tests against saved expected advice values remain to be added.