The folks down in Oregon had developed a really good model for explaining the returns of Oregon coastal coho. It had 4 different, temporally successive, variables. The problem was that the model was not good at predicting the future because the last variable (essentially measured at the coho arrival to the bays) could over-rule the other 3. The first 3 said "bonanza" and #4 could still crash it.

The lesson that should be learned, but won't be because of inertia and demand for mixed stock fisheries, is that you fish on known numbers with those numbers based on updates of the actual returning fish.

That won't happen because we all (AK, BC, WA) want that ocean fishery. So we take our best shot knowing (or should know) that a last minute change in marine conditions could blow things out of the water.

As to Todd's question about this year having age-4 adults instead of age-3, a competent sampling program would have caught this. Sockeye managers read (at least used to) hundreds of scales a day, the day after collection. The technology exists to do a much better job of knowing what is going on with the stocks. It simply costs money that is desired to be spent elsewhere.