Discussion > Do computer models provide 'evidence'?
If you have an ensemble of computer models based on a common set of assumptions and they all fail to match reality (even after you've finagled every parameter you can)in the same direction to varying degrees:
You ought to at least consider that you have a problem.
You ought to consider that as they all fail the same, the assumptions on which they are based are incorrect.
If you don't at least look at the problem, or conceal that you are looking, you are being a scientist any more, but a partisan whose priority is to protect the narrative rather than discover the truth.
And I don't have to consider the output of any model where I don't trust the behaviour of the people who run it.
rhoda
"I didn't say models weren't useful, I said they weren't evidence."
Playing with words again. It is the same thing in this context. If in the above example of modelling your idea or theory, you draw a conclusion from the results of the model, you are using the model output as evidence for or against the theory. If you stop work when it becomes clear to you that the evidence points one way or the other even before the model is complete and validated, that means that the evidence was strong enough for you to act. It might not have been proof of the theory, but it was sufficient. Maybe my list was too subtle for you. The thing is that you want to deny anything as evidence that is not provable. It is not evidence but proof you are after.
Splitpin it makes no difference which computer we are talking of. Vast increases in computing power make things possible that once were not. That is not a complicated idea to understand. ANH1, ditto. It is the ratio of processing power that matters. The Cray 1 we had at uni in the 80s is like a toy compared to an iPhone.
rhoda, do you attend the appropriate conferences, read the papers, talk to the modellers? From what you write one might think you are deeply knowledgable about climate modelling but your self description as an oxfordshire housewife makes me doubt that. How many modellers and separate teams have you discussed your concerns with? Have they all said that they consider there to be no problems with models in general or their own models in particular? Have they stopped work on their models because they think they are perfect?
As someone who claims the power to recognize positive feedback on the basis that temperatures change and then change a little bit more (that must be difficult to spot) your perceptiveness is likely to be in high demand by modellers.
Raff
Raff,
Increases in computer power are not as relevant to this discussion as my other questions. Who writes the specification for the program that attempts to model the climate? Who decides what are the relationships between the variables? What sort of quality assurance testing is performed?
You give me the impression that you think that somehow modern computers can think for themselves.
ANH1
Raff, there have been dozens of people like you here, I suspect you think you're using up energy or something by indulging in circular conversations and never actually addressing any of the points raised to you. We see you doing it, you're not unique, we've had many like you. We have infinite patience, because in the end, you all leave.
Something being useful and something being proof are not the same things, in this context or anywhere. If I draw conclusions from an incomplete model, then it's a shaky conclusion. If I use that conclusion to inspire further realms of study, to discount a theory, or simply to generate discussion, then that means it was a useful model, even if shaky. If I use that conclusion to claim that the models are the only reliable way to predict future climate (to quote the Met Office) then that is completely stupid. They are not the same thing, for you to argue they are is trite and ridiculous.
Now since you enjoy arguing, and it seems would argue about the colour of shite if it was a 'sceptic' saying it was brown, you will reply to this self-evident fact by providing a straw-man and then knocking it down, perhaps a bit of pigeon-holing, or a sneaky ad-hom to distract. If the point really stumps you, there is always the troll strategy of ignoring it completely, and saying something inflammatory to someone else on the thread who will rise to the bait, or perhaps abandoning this thread altogether and moving to another. These tactics are as familiar to us as the back of your hairy hand is to you.
Now, I've countered every point you've made. You have been unable to demonstrate any flaw in my arguments. You call this word-play - I prefer to call it the inexorable application of logic. Care to try some?
TheBigYinJames
A signal with positive feedback has characteristics. More tendency to change rapidly with time, both upwards and downwards. We don't really see that in thew global mean temps. We don't generally see that out of the window either, while we can see negative feedbacks, like those cumulus clouds out there right now. What I see is consistent with negative or very weak positive feedbacks. No magic powers are involved. Signals of phenomena with positive feedbacks don't usually stay flat for 17 years. They might, but horses, not zebras.
And the modellers? I don't have any idea what they say privately. Are you saying it is different from what they and their bosses say publicly? That model output can be used for policy decisions? If so they are being dishonest. If not, I was right to question their approach. Any one of them can come here and correct me. Why do they let Raff carry the load of putting out all the logical fallacies, obfuscation and misrepresentation of his opponents' arguments?
As it is, they don't even chuck out the worst-performing models from the ensemble. For political reasons is my guess although when I asked that here of a Met Office person I got no reply. Oxfordshire housewife though I am, I share this concern with proper scientists such as Roy G. Brown at Duke who explains his problem with the models elsewhere.
rhoda
ANH1, if you want your questions answered, address them to modellers. But be prepared to get the cold shoulder unless you leave your "Do computer models provide evidence? Don't be silly." attitude at home. Also you might omit boasting about being an 'operator', 'ops manager', 'project manager' or 'contractor' as it will impress noone. Even people calling themselves 'analysts' usually make me wonder a little.
TBYJ: "I didn't say models weren't useful, I said they weren't evidence."
Raff: Playing with words again. It is the same thing in this context.
TBYJ: "Something being useful and something being proof are not the same things, in this context or anywhere."
No indeed they are not, but that is not what I said, as you well know. Now it seems you quietly accept that it really is evidence, as I have said all along, and change the words to cover your dignity. Proof is quite different.
rhoda, "A signal with positive feedback has characteristics. More tendency to change rapidly with time, both upwards and downwards. We don't really see that in thew global mean temps." Still digging, I see. What do you mean by 'rapidly'? With respect to what? To the rate of change without feedback perhaps, but neither you nor I knows what that is. So how can you judge what is 'more' rapid because of feedback or just normal speed? You have no way of telling whether you are seeing feedback or not. You cannot even tell if the cumulus cloud you observe is a 'feedback' unless you can say for sure that cumulus cloud cover increases as a result of warming. How would you know that? Got a model of some sort?
And it is interesting that "they don't even chuck out the worst-performing models from the ensemble". Would you have more confidence in modelling or the modellers or 'they' if you knew that the ensemble had been pre-filtered to present only well performing models?
Raff
Raff, Finally you are stumped so you have nothing left but to go for plain insults and nastiness. I wrote down what jobs I have done to show that I have some experience in IT, not boasting, just stating facts. I never said I was a contractor, although I was at one time. Obviously you have no answer to my questions, this doesn't surprise me one bit.
ANH1
If models can't get the broad brush strokes right at one end or the fine detail at the other, how can they be right in the middle?
http://hockeyschtick.blogspot.co.uk/2014/07/new-paper-finds-striking-mismatch.html
New paper finds a 'striking mismatch' between data & models for the present & last interglacial
TinyCO2
Some time back I posted the following. I'll post it again here.
_________________________________________________________________________-
I think I have noticed a trend - although the sample is rather small to draw any firm conclusions.
1. A CAGW Believer turns up and poses questions/comments (evidently having read something like "How to Talk to a Climate Skeptic"). Their knowledge of physics generally seems to have be limited more or less to rote learning. They seem adept at quoting things they have read but their comments don't seem to show any fundamental understanding of the points under discussion. They tend to say things like "go and read some science" - something it is quite hard to imagine anybody who has studied, say, physics or chemistry saying.
2. It becomes apparent the Believer seems surprised that the lucidity of their comments has not convinced readers into accepting that their sceptical views are erroneous.
3. The CAGW Believer becomes noticably more and more more prolific.
4. The CAGW Believer starts to show signs of frustration, dropping any pretence at discussing rationally, and turns to mocking people's names or other personal attributes.
5. Finally, the CAGW Believer disappears for one reason or another.
Martin A
6. It reappears under a different name.
I studied electrical circuits around 50 yrs ago and then failed to do any work in them so I've forgotten most of what I learned. However enough of my knowledge remains to identify a complete and utter bu****itter talking about positive feedbacks and everything else. It is Chandra the eternal unwelcome guest, arguing the usual half digested, and barely understood "facts". He's hi-jacked the thread under the name raff, in typical Chandra fashion. He brings no knew information merely taking the responses, or comments, and turning them to suit his agenda, while taking care to ignore comments that he doesn't understand/can't answer.
He's not being deliberately provocative when he says model outputs are "evidence" it's what he thinks, so he's more to be pitied than scorned for his lamentable ignorance. Many on here, myself included, will find it hard to believe that an intelligent person could possibly believe "evidence" in physics can exist outwith the natural world, physics is, after all, simply an explanation of the natural world. But put yourself in Chandra's shoes and try to understand his complete lack of any knowledge of physics (and for all I know chemistry, biology, maths and certainly feedbacks in the climate system) in those shoes bereft of any knowledge or experience you could, or might just, believe that a computer could produce "evidence". Why not? Again, emboldened by the belief that you're parrotting the views of really clever climate scientists you might assume that someone like ANH1 who tries to give you some background on his experience in the use of computers is an idiot to be sneered at and rebuked for "showing off". Instead of being respected for the undoubted expertise he brings to the table.
But that's our unlovable Chandra for you.
geronimo
g - there are similarities but I don't think they are the same person. I think raff is a C programmer - hence his sneering contempt for someone who was once a computer operator. And (common with many C programmers) his lack of comprehension of the relation between the physical world and abstractions of it.
Wasn't bitbucket a C programmer too?
splitpin
An interesting article at Hockey Schtick, the original is paywalled at Wall Street Journal, some very interesting comments too and all relevant here.
http://hockeyschtick.blogspot.fr/2014/07/wsj-confessions-of-computer-modeler-any.html
Re Raff, definitely a different entity from Chandra and BitBucket, and BBD for that matter, equally convinced that he is right and he is dealing with idiots though. BB was as I recall a C programmer but TBYJ will know for sure Every C programmer I have ever met, there have been quite a few, seem to think the only way is C. Anyone with some experience of life knows there is more than one way to kill a cat.
SandyS
Raff fits the general Martin model. They come here to 'educate the savages' and quickly realise they have the lowest IQ here and begin spitting and screaming. I once described it as the belief that saving the planet and all the fluffy animals therein is so fundamentally good that anyone who questions the science behind it whatsoever must be either mad or bad. Once they realise we're not mad, we must be bad, and we get treated as such.
TheBigYinJames
I'm still in occasional touch with BitBucket, I believe he has done a little C++ in his time (who hasn't?) because when we were doing the lunar insolation modelling, that's what he wrote his model in, but I believe he's not coding professionally these days. While he followed the "Martin Sequence" while he was here, my conversations afterwards with him led me to believe that was coincidental, and I feel a bit bad about asking him to leave in the end. (The fact that he did leave when we asked him to further strengthens my belief he was not a traditional shill troll like Chandra or Raff who would rather take a cheese grater to their genitals than acquiesce to our requests)
If anyone has anything they'd like me to pass on to BitBucket, let me know :)
TheBigYinJames
As there are presently 100+ computer models of the “climate”, and not one of them agrees with any of the others, which model are we to go to for the evidence?
With typical fudge-it skill, however, the catastrophists decide that a better model could be obtained by taking an average of all the models, no matter how wrong they may be; now, we have an “average” model – which, again, is completely at odds with reality, yet this is the one providing the “evidence”. To repeat an argument I used in one of the BH blogs: A survey of scientists was conducted as to what they believed was the direction of “Up”. One scientist pointing nearly vertically, close to right angles to the horizontal; all the others pointed at different angles, to about 45° below the horizontal.
Conclusion: with such a divergence of opinions, it has to be concluded that no one scientist can claim to be correct; the actual direction of “Up”, therefore, has to be the mean of all the directions indicate by the whole of those present. The consensus, thus derived, is that “Up” is anywhere at an angle of 30.325° above the nominal horizontal.
Further work is being undertaken to determine what “Horizontal” is.
Radical Rodent
As there are presently 100+ computer models of the “climate”, and not one of them agrees with any of the others, which model are we to go to for the evidence?
Jul 19, 2014 at 1:19 PM | Registered CommenterRadical Rodent
Is that correct RR?
I thought they were pretty well unanimous that runaway warming is taking place.
I was astonished on reading a papers on a new climate model to find that its 'validation' consisted of verifying that their results agreed with previous climate models. It's hard to start enumerating the depths of misconceptions about correctly modelling a complex physical system that that reveals.
Martin A
…despite rising sea levels, air and ocean temperatures, acidifying oceans, shrinking ice sheets and glaciers, degrading permafrost, changing species distributions…Raff, please tell us when in history these (or the converse) have NOT been happening.
…our understanding that rising CO2 levels are the most likely cause of all that…Is it, now….? Curious, but CO2 levels continue to rise, yet temperatures, sea levels and ocean “acidification” are not; indeed, the entire “greenhouse effect” is getting on increasingly shaky ground.
As for predicting the future, one of the best ways is to look at the patterns of the past (for example, this is how astronomy works – what were the movements of celestial bodies in the past? So, unless something drastic happens (even astronomy accepts that in the vast expanse of nothing at all there could be something), this is where they will be in the future). With that hypothesis, I will predict, here and now, for all to hear, that the Earth in 1,000 years’ time, barring celestial catastrophe, will be more or less as it is now, with a margin of +/- 2°C in temperature, +/- 30cm in sea levels, and +/- pH0.5 ocean “acidity” – though, as the ocean pH is presently in the region of 8.3, it would be more accurate (but less scary) to refer to it as alkalinity. Also, pandas and many other species will be extinct; other species will have been discovered (though, interestingly, none will ever be proved to have evolved in the interim). If I am proved wrong, you can then gloat.
Radical Rodent
TBYJ
that's what he wrote his model in
That explains a great deal.
SandyS
Good point, Martin. Perhaps my analogy would have been better if all had agreed that “Up” was above the horizontal, but all at different degrees – that is more or less how the climate models are; while the rises may differ, they are all upwards, and all significantly greater than reality, which is, curiously, quite obviously incorrect, in the minds of the AGWistas.
Radical Rodent
BYIJ - please convey our greetings to BB.
He never revealed where he was; presumably he did not wish to. But it was somewhere:
- At high altitude (so I initially assumed perhaps somewhere in Colorado)
- Where gasoline is incredibly cheap - much cheaper even than in the USA .
- Where power cuts happened sufficiently frequently that many buildings had their own generators.
I imagined that that information would pin down where he was but my curiosity was never sufficient to try to solve the puzzle.
Martin A
The only evidence that a model – whether 0s and 1s in a computer, or a real, solid model – can give is whether or not your ideas are correct; in doing that, they might induce further ideas to improve or amend them. As none of the climate models come anywhere near reality, the conclusion HAS to be that the ideas behind them are wrong, and a radical rethink is needed. But, no, we have apologists like Raff, and… so many others I cannot be bothered looking their names up, who insist that the models are so true that the evidence they present is close to scientific certainty and reality is wrong.
(No, Raff is not another incarnation of Chandra – Chandra was a junior researcher for a lesser politician (which might explain his/her dearth of input, as the politician no longer has an office, and Chandra has been thrown out into the real world). Raff, however, comes over as a student, sitting in the Uni library (or in digs, with the uni shut for summer) desperately bored, and looking for a larf, winding up the proles.)
Radical Rodent
“I thought they were pretty well unanimous that runaway warming is taking place.” Martin A.
http://www.drroyspencer.com/wp-content/uploads/CMIP5-90-models-global-Tsfc-vs-obs-thru-2013.png
The models are all steady upward trends, which suggests that the dominant input is greenhouse gases. They mostly show the wiggles during the 80s and early 90s, but apart from that they don’t resemble each other or reality. Year to year wiggles are often 180º out of phase. So the models can not model reality at the yearly level.
Time and again the Medieval Warm Period is shown to exist and be at least as warm as the present.
http://hockeyschtick.blogspot.co.uk/2014/07/new-paper-finds-medieval-warm-period.html
If the modellers could model climate at the millenial or centenial scale THEY should have shot Mann's hockey stick down in flames. I’ve never seen that blue moon but perhaps Raff or Entropic Man can post a link? Of course, given the speed with which the thermometer record is changing, Mann will be able to claim that his reconstruction was right and the reality record he centred it on was wrong.
Why should models ever be treated as evidence when they can’t match anything but the recent past on which the assumptions the programmes are based upon were made?
TinyCO2
Why should models ever be treated as evidence when they can’t match anything but the recent past on which the assumptions the programmes are based upon were made?
Jul 19, 2014 at 3:31 PM | Unregistered CommenterTinyCO2
It's nonsense. As I've said before, it's the fallacy of "testing on the training data". The Met Office's claim that the ability to reproduce (the recent) past is an example. I don't have the reference to hand but Richard Betts some time back referred me to their paper from which this claim originates.
If you make the assumption that things are independent when there are even tiny nonzero correlations, you generally get completely misleading results. Climate models, through "parameterisation" apart from anything else, depend heavily on the recorded recent history of climate.
Of course they match the recent past. They were constructed to do so. If they had not done so, the Met Office's Fortran programmers would have been told to get a grip and go back and sort things out.
The fact that they don't match what was, in 1998, the near future is sufficient evidence that they are unfit for purpose. The unanimity of the models confirms that they share wrong assumptions (not to mention that they also share code).
Martin A
Martin A, I knw WE can see it. What I can't understand is why anyone else (warmists) would see it differently.
TinyCO2
Computer power can only affect two things with regards to modelling - speed of computation, and resolution of computational unit. A bigger computer will allow you to model more cells faster than before, but if the uncertainties in the input data, fudge factors and underlying equations remains the same then all that happens is you calculate the wrong answer faster.
Raff seems to be arguing that because modelling in other domains are subject to the realities of incompleteness and commercial pressure to conform to a pre-defined answer, then climate models should be given a similar bye. That unless I hold my commercial modelling activities up to this gold standard, then I must allow climate modelling to be sloppy and shonky too.
This is, of course, tosh.
Fudging models is wrong wherever it is done. If it does anything, it further proves my point that models cannot be evidence, because there is too much scope for tampering, whether advertant or inadvertent. Modelling is - by its very nature - a simplification of a more complex problem, and thus can never be used as proof of the larger complex phenomenon.
All the rest of his post was a mix of ad-hom and distraction.