Fulcrum wrote: ↑08 Aug 2019, 17:33
AngusF1 wrote: ↑08 Aug 2019, 11:00
Fulcrum wrote: ↑07 Aug 2019, 16:32
I can't comment on whether using a Gaussian distribution is appropriate or not, having no exposure to the data.
However, if he'd used a non-Gaussian distribution, in all likelihood the distribution would be positively skewed, meaning the worst case temperature scenario would probably have a delta higher than 8 degrees for the same level of significance.
The level of significance must have been fairly small to obtain that "worst case" right-tail event, considering the values for mean warming I am aware of (1.5C - 2C by the latter half / end of this century).
The point is IMO, if you are ignorant of a phenomenon, you can't just say "suppose it's statistical according to this specific model which I pulled out of my backside and can take any value I want!", which is precisely what the professor is doing.
If you are ignorant of a phenomenon you get to say... nothing, and stick to what you are certain of, ie the 1.2 degrees directly caused by CO2. Anything further is pure speculation.
"Suppose" is a pretty well worn phrase in statistical analysis. You make an assumption, run tests, reach conclusions about your original assumption. It does not imply he is ignorant.
I am not claiming that the professor's use of statistics and "suppose that..." reasoning makes him ignorant. I am taking him at his own word that he is ignorant, then criticising his use of an unjustifiable and inapplicable statistical distribution to claim that the real model (which he admits to not knowing) could be anything, especially his particular emphasis on alarmist worst-case scenarios made possible at the extreme end of his unjustified application of a arbitrary statistical distribution.
On the ability to accurately predict the actual real-world temperature change due to a doubling of CO2, the professor states
"...it's still really not possible today, and it may not be possible any time soon. Probably not, probably, actually, to really make a definitive statement on this." So, he openly and readily admits ignorance, which is good of him and perfectly fine. He then goes on to explain the reasons for the difficulty, mainly revolving around the non-linear feedbacks in the climate system which make it so difficult to model.
What I'm criticising is his subsequent application of a Gaussian distribution to the known expected rise due to CO2 alone, to represent the range of possible models. This is a total abuse of statistics, because the "real climate model" is not a statistical phenomenon at all. It's a concrete mathematical model played out in real time all over the earth every day, which we happen to be ignorant of.
To illustrate my point, why a Gaussian distribution? What about the range of possible climate models is normally distributed? (Nothing, it doesn't even make sense to say this, it's a completely non-applicable concept.) Could they logically follow a Guassian distribution, which has a nonzero likelihood of any possible outcome? (No.) If a Gaussian distribution is applied, what should the standard deviation be?* (No ability to sensibly determine, all possible values make zero sense.)
*[The Standard Deviation is one of two statistical parameters which describe a normal distribution. It determines how wide the variation is. The professor places the other parameter, the mean, at the amount of warming caused by CO2 alone.]
Fulcrum wrote: ↑07 Aug 2019, 16:32
Like others have said, there is a high price for being wrong (under prediction). I can understand the pressure scientists must feel to have some estimate for 'worst case scenarios'.
Certainly. This is where scientists need to stand firm, resist the pressure to engage in alarmism and present what they know for sure instead of engaging in speculation. Speculation is fun. It costs nothing, you can say whatever you like, and there is no skin in the game or responsibility for being wrong. After all, I only said it "could" happen, not that it "would".
By contrast, scientists seem to be relatively quiet in speculating about planet-killing asteroids, alien invasion or exotic super-viruses. They know these scenarios are possible but are hardly urging us to immediately develop an interstellar Space Force, or ban most international travel with a 1-month quarantine before entry. I wonder what the difference is.
Fulcrum wrote: ↑07 Aug 2019, 16:32
Like you said, if you're ignorant, perhaps its best to not comment.
I know nothing about statistical mechanics, quantum mechanics and absorption spectrums, so I leave it up to the scientists to determine the warming caused directly by increasing the CO2. They say it's 1.2 degrees and I believe them because it's validated by experiment and fundamental physics. This is real science as it's a closed loop process: hypothesise, experiment, validate, repeat.
What I can do, is spot the difference between genuine science and speculation. Currently the climate models have no predictive ability and are not backed up by experiment (ie, they are all wrong), which is readily admitted by the scientists themselves. So, I categorise them as speculation. This is fine as an academic pursuit, but they are not suitable as a trigger for action in the real world, much less for turning our civilisation on its head.