Very cool visualization and great way to display it.
As someone who refuses to trust authority and wants to understand things for myself before making a decision, global warming is very frustrating because nobody will answer my questions without personal attacks or appeals to authority. I don't have an agenda either way, I just have a very inquisitive mind. Have any of you felt the same?
Some of the questions I have that never seem to get an adequate answer:
1) How are the models validated? Is it like backtesting a trading strategy? Come up with a hypothesis that seems to fit historical data, then let it run with actual data, and see how accurate it is? If so, how have the models held up?
2) How do they account for confounding factors and how do they separate causal correlations from mere correlations?
For example, at 95% of fires firemen were present. Firemen and fires are strongly correlated. But nobody would say firemen are the cause of fires.
Cholesterol was thought to be a causal factor for heart problems because it is strongly correlated but they later found it is not a causal factor. Something else causes the heart disease and cholesterol raises when heart disease is present. They can use it as a predictor of heart disease but it is now understood that cholesterol doesn't actually cause the problem.
3) It seems to me that for a model to be trusted it must have predictive capability, and it must fit a physical model of our current understanding. How do the various models hold up with these criteria? It seems like climate is still a very complex field that we don't fully understand.
As someone who refuses to trust authority and wants to understand things for myself before making a decision, global warming is very frustrating because nobody will answer my questions without personal attacks or appeals to authority. I don't have an agenda either way, I just have a very inquisitive mind. Have any of you felt the same?
Some of the questions I have that never seem to get an adequate answer:
1) How are the models validated? Is it like backtesting a trading strategy? Come up with a hypothesis that seems to fit historical data, then let it run with actual data, and see how accurate it is? If so, how have the models held up?
2) How do they account for confounding factors and how do they separate causal correlations from mere correlations?
For example, at 95% of fires firemen were present. Firemen and fires are strongly correlated. But nobody would say firemen are the cause of fires.
Cholesterol was thought to be a causal factor for heart problems because it is strongly correlated but they later found it is not a causal factor. Something else causes the heart disease and cholesterol raises when heart disease is present. They can use it as a predictor of heart disease but it is now understood that cholesterol doesn't actually cause the problem.
3) It seems to me that for a model to be trusted it must have predictive capability, and it must fit a physical model of our current understanding. How do the various models hold up with these criteria? It seems like climate is still a very complex field that we don't fully understand.