3 ms·

The problem I'm having, and I have a masters in statistics, is that this data is generated after the fact. If you compare the predictions of the 2000-2010 perio

The problem I'm having, and I have a masters in statistics, is that this data is generated after the fact. If you compare the predictions of the 2000-2010 period with the "after-the-fact" predictions of the 2000-2010 period ... There's a huge difference. According to the prediction EVERY year between 2000 and 2010 was between a 3 and 5-sigma event. In the after-the-fact model this is simply ignored, and the data follows the graph much better and ... nothing special happened apparently. Sorry no. This is an indication that your predictions are worthless. And of course, after such a massive fuckup, it's going to take a bit of work to convince people otherwise. But none of that is happening, I wonder why not. The worst of it all is that I used to make a similar argument, first about 1990-2000, then about 1995-2005. They all had the same result. Suddenly, and without explanation they were far, far cooler (compared to the prediction). Not quite 5-sigma years, but 3 sigma years were in there. 2000-2010 is the worst of the lot, but there isn't any decade-long period where one can say ... well climate scientists ... well done. You were right. The IPCC AR's, predicted temperature anomaly versus observed temperature anomaly. Given these circumstances, how can we give any credence to the idea that the IPCC predictions about the future will match reality ? The joke is that in order for their 95% prediction to actually be true in 95% of cases they now have to be correct for 20.000 years. They don't even predict out that far. It's a joke, but it's also true. You'd think it would give people pause. And all this is still ignoring the many fundamental theoretical problems climate science has. When they say "it's worse than ever before" ... well they're right, in terms of many variables. Ok ... but that means their methods are invalid. Statistics can fill in expected values in an observed population of values. And yet their claim is that the observed populations are VERY different from the ones seen before. Ok, fine claim. You can even prove that to some extent with statistics. And they've done that. After that, of course, you cannot use statistics to predict these values anymore ! This alone could explain why their predictions are so horribly bad. I get the utilitarian argument. We don't have any better way of making predictions about the climate. That's great, but that doesn't make our best prediction adequate, nor is it even value-free. In some ways, it's just a political argument : "this looks worrying, let's fix it". It should be treated as such.

rusk 10y ago
You claim to have a "masters in statistics" but that doesn't really help me in understanding the main points of your argument. In a nutshell what you are saying is that a lot of the statistical models developed have changed over the last 20 years or so to accommodate the data that has actually emerged. That each of these models if they were applied today would diverge significantly in their predictions from the actual situation we're seeing today? I find this a stunning indictment of mainstream scientific consensus, especially considering many of the actors in that field probably have Masters in statistics and other fields, PhDs and professorships. But, we do see this kind of skew in other fields, in particular economics comes to mind, where an overarching consensus on what is and is not good seems to promote the viewpoints selected for discussion, while neglecting counter-theses.
- candiodari 10y ago
  > I find this a stunning indictment of mainstream scientific consensus How exactly ? Climate science is thoroughly in the humanities at this point, very much not in exact science, and this is hardly the only wrong-but-we-can't-do-better consensus that exists. People believe this and it provides them with things to do, with the "ability to predict". Except, it's not correct use of math, with the predictable result that the outcome doesn't match reality. We "believe" a lot of things for political reasons. For instance, starting in 1995 or so papers suddenly started showing that all races are equal in everything, from iq to running speed. It's becoming controversial to state that Africans need supplemental UV at high latitudes, despite the medical consequences of not doing so. Looking at differences in their performance is systematically removed from all textbooks. Now look at reality. You have watched the olympic games recently ? I assure you the impression you get from those races is more or less correct. As for this specific one, you can simply look this up. That's supposed to be the beauty of science, except, of course where there's "scientific consensus", whatever that means. And you know, worst case is you learn something. For the lazy, google it (this looks reasonable): http://clivebest.com/blog/?p=2208 http://clivebest.com/blog/?p=2208 (but obviously download the IPCC AR0 and check the data in a table as well) Obviously there is scientific consensus in the statistics community too that you can't do what climate models do without destroying your credibility, but coming out and saying that ... isn't happening. And sadly, I know why. Even if you disregard results, you know how the IPCC arrives at their 95% certainty interval ? They take 20 highly non-linear climate models, assume normal distribution of the results per-year (think about that for 5 seconds), and come up with confidence intervals. To claim this won't provide accurate results ... does that really require data ? At what point can you just call bullshit without further study ? One statistics paper humorously provided a proof that with this process, you can create any outcome you want. And for the conspiracy nuts : the most recent IPCC AR leaves out actual numbers entirely. Because apparently (yes really) they're too easy to criticize.

rusk 10y ago

You claim to have a "masters in statistics" but that doesn't really help me in understanding the main points of your argument. In a nutshell what you are saying is that a lot of the statistical models developed have changed over the last 20 years or so to accommodate the data that has actually emerged. That each of these models if they were applied today would diverge significantly in their predictions from the actual situation we're seeing today? I find this a stunning indictment of mainstream scientific consensus, especially considering many of the actors in that field probably have Masters in statistics and other fields, PhDs and professorships. But, we do see this kind of skew in other fields, in particular economics comes to mind, where an overarching consensus on what is and is not good seems to promote the viewpoints selected for discussion, while neglecting counter-theses.

candiodari 10y ago
> I find this a stunning indictment of mainstream scientific consensus How exactly ? Climate science is thoroughly in the humanities at this point, very much not in exact science, and this is hardly the only wrong-but-we-can't-do-better consensus that exists. People believe this and it provides them with things to do, with the "ability to predict". Except, it's not correct use of math, with the predictable result that the outcome doesn't match reality. We "believe" a lot of things for political reasons. For instance, starting in 1995 or so papers suddenly started showing that all races are equal in everything, from iq to running speed. It's becoming controversial to state that Africans need supplemental UV at high latitudes, despite the medical consequences of not doing so. Looking at differences in their performance is systematically removed from all textbooks. Now look at reality. You have watched the olympic games recently ? I assure you the impression you get from those races is more or less correct. As for this specific one, you can simply look this up. That's supposed to be the beauty of science, except, of course where there's "scientific consensus", whatever that means. And you know, worst case is you learn something. For the lazy, google it (this looks reasonable): http://clivebest.com/blog/?p=2208 http://clivebest.com/blog/?p=2208 (but obviously download the IPCC AR0 and check the data in a table as well) Obviously there is scientific consensus in the statistics community too that you can't do what climate models do without destroying your credibility, but coming out and saying that ... isn't happening. And sadly, I know why. Even if you disregard results, you know how the IPCC arrives at their 95% certainty interval ? They take 20 highly non-linear climate models, assume normal distribution of the results per-year (think about that for 5 seconds), and come up with confidence intervals. To claim this won't provide accurate results ... does that really require data ? At what point can you just call bullshit without further study ? One statistics paper humorously provided a proof that with this process, you can create any outcome you want. And for the conspiracy nuts : the most recent IPCC AR leaves out actual numbers entirely. Because apparently (yes really) they're too easy to criticize.