Item — HN Reader

The context of that prediction were the extreme records set by sea ice in 2012, when it did almost disappear. That made some scientists think that we were seeing a nonlinear phase shift and that it would not recover. It did recover, and resumed its linear downward trajectory.

But if your takeaway from that is that scientists are overconfident and wrong, that is insane, because the important thing to know about sea ice is that at its yearly minimum, its volume is a third of what it was in 1980. No matter how you cut it, it's almost gone.

https://twitter.com/ZLabe/status/1314620374601003008?s=19

2 comments

He did not say the scientists are overconfident and wrong.

He said they hurt the public trust in climate science when their predictions do not reliably come true.

What hurts public trust in climate science is primarily a massive and well funded propaganda campaign that's been going on for 50 years.

Some scientists are going to sometimes be wrong about things. It is impossible for them to be scientists otherwise.

Neither of us denied any of that.

OP just said that conservative predictions are better than aggressive ones, as they're more likely to come true and thus build trust.

Do you see any evidence that the consistent under-shooting of IPCC consensus predictions has built trust?

It is not possible to have an accurate mean prediction, and also to have no predictions above the mean.

I never said they did, or that they should focus on trustbuilding instead of aiming for accuracy.

I just pointed out that you were arguing with something other than what OP said.

> But if your takeaway from that is that scientists are overconfident and wrong

my takeaway is that every years hundred models are given birth, every year later the model fitting the data best survive, and two year later when that last model prediction fail, a new model from the previous that that predicted the change better replaces it, in a never ending cycle of bullshit.

the models go both way: without a predictive model that can hold water, how do you know which parameter to tune to resolve the climate crisis?

You’re describing how models improve over time with additional data and better knowledge about the atmosphere.

Or models over fitted to past data with 0 predictive capability that need to be re-tuned every year or two.