Thursday, Friday, Saturday

  • forage fish seminar - Patrick and Jay with interesting stuff. Sounds like compensating benefits due to quality-response (larger fish) and stock effect (easier to find) will be more minor in this fishery, but demand response be compensating.
  • knit hooks
  • externalization example working (based on Reed.R). Still unsure how I feel about externalization vs fully literate code. Redundant typing is bad but isn’t that why common tasks are functionalized?
  • pmc-0.5 posted to CRAN (use check --as-cran in tests)
  • R updated 2.14.2, packages updated.
  • markdown notes of Reed vs extra noise

Rules-of-thumb that are not robust

  • Precautionary.md considers deriving an optimal policy by assuming that a threshold (in the RickerAllee model) is higher than it really is, which should lead to a conservative policy. This does lead to fewer crashes than the optimal model derived with the correct uncertainty, but, surprisingly, when faced with a higher noise than anticipated, this does not out-perform the optimal model.
  • This pattern persists in the case where the additional noise comes from the implementation step, rather than as additional growth noise (a conservative estimate of the threshold point doesn’t help).

Progress as told by the commit logs.