I think the CSAS model is a good approach. I think where the rubber meets the road is often with the certainty that different scientific studies have. It seems that very few people ever ask, what is your confidence in your result?
Typically, for example, we've got some models that predict that removing seals from the west coast of Canada would increase the abundance of salmon, and many people will rally behind that conclusion without ever asking, how confident are you in those results? The people doing the models and those who are familiar with how they're parameterized would tell you, there's about a 30% to 40% chance that the model is right.
For many people, if you're going to make a big decision like that, you want to have a confidence of over 80%. On the other hand, if what you're putting up and what's at stake is something one might not value, perhaps the life of the seal for example, you can say that 30% to 40% odds are amazing when you look at how much the fishery is worth.
For other people, it's too big a gamble, as one that would take perhaps 30 to 40 years to discover it may have been a failed gamble.