The first step when conducting a stock assessment is to pick up the previous assessment and reuse and modify the analyses. This step is a particularly important challenge at SPC because of the high complexity of the analyses, staff turnover rate, and the number of years between assessments.
An important quality indicator for SPC assessments is how well it has been 'parked'. In other words, how easily it can be picked up by the next scientist.
In 2023, we organize the assessments in directory trees on Penguin that resemble how things have been done in the past. The subtle but important changes are that:
-
We use a standard directory tree instead of the old way of arbitrary directory names. This makes it easier to find things and check which parts have been completed and properly parked.
-
R scripts use relative file paths instead of the old way of using
setwd
and absolute paths. This makes it possible for teammates to run the R scripts. -
Some folders on Penguin are GitHub repositories. This helps track incremental development and improves reproducibility.
For the 2023 assessments, we consider rules #1 and #2 compulsory. They introduce no additional overhead and bring important benefits for the scientists and SPC as a whole.
Recommendation #3 is optional and scientist can decide whether and when it is practical to organize specific analyses on GitHub. The manifesto elaborates on this point.
Finally, we apply the Arni test to each part of the stock assessment-related analyses, identifying where quality improvements can be made.
The score card helps to measure progress, where we can expect incremental improvements each year.