"Please tell me what you propose in place of models. Crossed fingers?"

It is perfectly understandable that if one believes in ghosts then one's frightened of them and expect them to appear at any time. It is also perfectly understandable that if one believes in ghosts one assumes everyone else does. Hence the totally understandable question that assumes bad things are going to happen and models can tell us what they are and when they'll happen.

Try taking a step make and try to understand that humans have coped with all sorts of climatic conditions and climatic change over thousands of years without the aid of computers. We still have people who believe in ghosts, but technology and education, and belief in ghosts is confined to really daft people now. But for large numbers of people there is still the need to have something to frighten them. Global warming is the current scare. Whereas the superstitious in the past relied upon clerics shaking bones or examining birds entrails today's superstitious rely upon scientists and computers as a replacement for clerics and bird's entrails.

I don't need a model to tell me about the future because I don't believe there will be anything like the catastrophes forecast from the models, because the models are crap at forecasting. You are making the mistake of assuming that everyone has the same basic beliefs as you do. I don't.

However, if you were to ask me what I would do if I really believed that CO2 was going to cause untold disasters in the future, it would have been to abandon funding climate science once we'd accepted the science and diverted the money and much more into research into fusion reactors.

By the way your confusing "clues" with "evidence", evidence is what provides proof, clues do not.

For me the whole business of climate modelling was massively tainted by the many failings of the climateprediction.net public relations exercise. If that project is any where close to typical then I will continue to have real difficulties with the way model results are portrayed. I would not regard them as evidence of anything other than the biases of the modellers.

http://www.climateprediction.net/

"In February 2006, the project moved on to more realistic climate models. A BBC Climate Change Experiment[17] was launched, attracting around 23,000 participants on the first day. The transient climate simulation introduced realistic oceans. This allowed the experiment to investigate changes in the climate response as the climate forcings are changed, rather than an equilibrium response to a significant change like doubling the carbon dioxide level".

http://en.wikipedia.org/wiki/Climateprediction.net

Again from wiki

The first results of the experiment were published in Nature in January 2005 and show that with only slight changes to the parameters within plausible ranges, the models can show climate sensitivities ranging from less than 2 °C to more than 11 °C (see [20] and explanation[21]). The higher climate sensitivities have been challenged as implausible. For example by Gavin Schmidt (a climate modeler with the NASA Goddard Institute for Space Studies in New York).

The program churned out a vast range of results some of which were "acceptable" but some were wrong because either the program contained bugs or it produced unphysical results.

http://climateaudit.org/2006/04/18/earths-climate-crashes-in-2013/

The modellers obviously had to sift out the wrong results but how do we know those left in are any more correct than those taken out?

Then we find that the modellers present the results that they found "acceptable" but which really just reflect their own personal view and biases.

http://ccgi.newbery1.plus.com/blog/?p=70

Rhoda, Ridley's reported formulation is all about semantics. He's quibbling about the meaning of words, redefining 'model' and 'evidence' as he would prefer them rather than as they are used. Models are perfectly able to provide evidence. They are used all the time for that purpose, even by Ridley.

Ridley has in the past written about the economic benefits he expects from climate change and even used Prof Tol's research as evidence in his argument. Tol's study used the results from 14 models and predicted, I'm sure you know, that there would be positive results from warming up to 2 degrees. I don't remember Ridley saying that he actually had no evidence for what he wrote because it was all based upon models. Nor have I seen him withdrawing what he wrote when it became evident that demons had written Tol's paper, not Tol himself, and that the actual updated result was negative for all levels of warming.

In other words Ridley is happy to use model output as evidence when it suits him. His words are worthless.

"What would we prefer to bad models? Good models."

Have you ever watched Miss Congeniality? Your sentence reminds me of the banal statement of desires from the contestants. What do you want? Peace and happiness on Earth. Or was it motherhood and apple pie?

We all want better models, you no more than me or Gavin Schmidt. There are many teams working towards that, just as there probably are in economic models. But what climate models have over economic ones is that they are based on physics. Yet people like Ridley are all too willing to believe favourable output from economic models which have no such foundation and yet they reject the physical. Did you object to the economic models Tol/Ridley promoted?

Models are tools.

geronimo, you say "I don't need a model to tell me about the future because I don't believe..." but where does that belief come from? You are forecasting and as such you must be using a model, whether mental, physical or virtual. The difference is that you are unable to express that model in any rigourous way. That is not much use when trying to assess risk and determine a course of action.

Martin, models are a way of investigating the climate system, just like your models were a way of representing and investigating your systems. Even if you had had no way of validting your models, I expect you would have produced them nevertheless. You would doubtless have given them less credence but they would still have given useful insights into the system. Models are tools, climate models no less.

What you are effectively saying is that unlike Rhoda, who at least wants better models, despite rising sea levels, air and ocean temperatures, acidifying oceans, shrinking ice sheets and glaciers, degrading permafrost, changing species distributions (all of which you claim not to know), and despite our understanding that rising CO2 levels are the most likely cause of all that, nevertheless trying to model this system and taking notice of what these models say is more harmful than ignoring it and hoping for the best (i.e. crossing fingers and guessing that it will all be ok). The crossed fingers bit makes you feel foolish because it is foolish.

Unless your own model (mental presumably) of the world can show that the effect of changing CO2 levels has a probability distribution that is infinitely narrow (in other words, your model tells you exactly what will happen with no doubt), there will always be a tail of uncertainty. In other words there is a non-zero probability that temperature will change as a result of extra CO2 by 4 degrees (say), or -4 degrees. The latter is comparable to entering a new glaciation. What does the former equate to? How big is the risk? Should we try to understand it and mitigate it? You seem just to know the answer already without studying the problem. To me that does seem foolish.

"Even if you had had no way of validting your models, I expect you would have produced them nevertheless."

Why would you?

It would be a complete and utter waste of time and energy.

Anyone want to buy a second hand polar-bear model?

Arguing about what the meaning of the word 'evidence' is doesn't come into it. That's just semantics. Raff argues a straw man. What would we prefer to bad models? Good models. But we'd still know that we are looking at the prejudice of the programmer, even if he happened to be right.

Jul 9, 2014 at 10:18 AM | rhoda

Avoids answering the question in the same way as Chandra, and again schoolboy (13?) argument.

Jul 9, 2014 at 4:31 PM | Nial
Not to mention large amounts of research money leading to huge waste in renewable energy.

Sandy, which question did I avoid? There is none there except mine. The title of the thread is "Do computer models provide 'evidence'?" - not evidence but 'evidence'. The word and its meaning are obviously central. Martin quotes Ridley questioning the word, claiming that models cannot provide evidence and yet he (Ridley) uses model output as evidence in articles he writes - where the evidence fits his objective. Ridley is clearly talking nonsense. You should be able to recognise it by the fact that he is inconsistent.

Rhoda, this thread is largely related to what Ridley has written. If you don't bloody care about what he writes or the inaccuracy thereof, why comment?

Nial, if doing a job properly requires the estimation the value of something, that is generally what is done. Even if the model produced cannot be validated (and all estimates use models of some form, if only mental), people still do their best to produce a result and they give confidence levels. That is how it is.

Martin, you are rewriting your own words. You didn't just say,"We simply don't know. That's all there is to it", you said, "I have said before, if you believe the output of an unvalidated model, you are putting yourself in a worse position than saying "We simply don't know. That's all there is to it". We'd suggest not using models and being honest about our lack of understanding. In other words you did not state your position but that of some other posited person. Possibly that is too subtle a distinction for this forum.

However, if you really characterise your own position as "We simply don't know", then I still cannot believe you mean "we know nothing". That would imply that you think the probability of any outcome is equal - that two decades from now global temperatures have an equal chance of being zero Kelvin or infinite. I imagine you would agree that those are impossible extremes, so I have to assume you would consider some values more likely than others - you have some sort of probability distribution in mind.

So to repeat myself, as you have stated that your own mental model of the world cannot show that the effect of changing CO2 levels has a probability distribution that is infinitely narrow (in other words, that your model tells you exactly what will happen with no doubt), there is inevitably some sort of probability distribution to what your model actually predicts. And hence there will be tails of uncertainty. There is hence a non-zero probability that temperature will change as a result of extra CO2 by 4 degrees (say), or -4 degrees. The latter is comparable to entering a new glaciation. What does the former equate to? How big is the risk? Should we try to understand it and mitigate it? You don't know the probability of a 4C rise and it is difficult to see how you can argue that the risk is zero. How do you suggest we determine that probability - whether we have a problem - without models or paleo studies? Or do we just say, "its too difficult, let's ignore it"?

Raff
Sandy, which question did I avoid?

rhoda
Arguing about what the meaning of the word 'evidence' is doesn't come into it. That's just semantics. Raff argues a straw man. What would we prefer to bad models? Good models. But we'd still know that we are looking at the prejudice of the programmer, even if he happened to be right.

Allows you to avoid answering any question and is a common feature on threads like this as well as disappearing off onto another one leaving questions in limbo.

Sandy, it is really unclear to me which question you think I avoided and which threads I have disappeared from leaving questions in limbo. From activity today on this thread I'd say that Nial, Rhoda, geronimo and Martin have 'disappeared' from this thread - is that what you are referring to? I agree that Martin for one has outstanding issues to answer, but the same issues apply to you all really, not just him. See my 11:07 - the logic is that we need to assess the likelihood and effects of a significant rise in temps and the rhetoric here indicates that you all think models are not to be used. In other words, the only tools at our disposal are to be rejected because the evidence they produce cannot be proven beyond doubt. That is beyond my comprehension, but you do seem to believe it - so what do we use instead?

> Nial, if doing a job properly requires the estimation the value of something, that is generally what
> is done. Even if the model produced cannot be validated (and all estimates use models of some
> form, if only mental), people still do their best to produce a result and they give
> confidence levels. That is how it is.

If you're basing your results on a model that cannot be validated the confidence level is 0.

You might as well use chicken entrails.

In the real world of Engineering 'evidence' that something occurs is confirmed by measurement. How many aircraft manufaturers sign off air worthiness based on model predctions? You can't get a PCB through emc testing by showing simulation results, you have to test it.

How much effort has climate science committed to measure the CO2 hypothesis?

Rhoda, you at least seem to accept that models would be useful if 'better'. What criteria would you use to judge them to be so?

Nial, do you accept that there is a non-zero probability of temperatures changing significantly due to CO2? If not, what 'model' (mental, physical, virtual, etc) are you using to give you such confidence. If you do, how do you think we should respond - by ignoring it or my modeling it?

> Nial, do you accept that there is a non-zero probability of temperatures changing significantly due to CO2?

No (depending on your definition of 'significanly').

The 'model' I'm using to come to this conclusion is a comparison of historical measurements of CO2 and temperature.

Raff
Having quoted what Rhoda said twice and my interpretation of what that means it should be clear that putting up strawmen and arguing semantics avoids answering all difficult questions.

Malcolm The Maiden put avoiding the issue more clearly when he alledgedly said to Henry II "I came in peace and not to answer difficult questions".

Nial, can you describe or illustrate the probability distribution function produced by your model?

Sandy, as I said once before, the title of the thread is "Do computer models provide 'evidence'?" - not evidence but 'evidence'. The word and its meaning are obviously central. So duscussing the word is not evading the 'question'. I don't know why you cannot get this, but it is your problem, I guess.

Nial
That's my position and has been for a number of years, I did have a slight wait and see caveat in the early days of Global Warming™, but having waited I'm in my default position.

The Climate isn't changing merely behaving as it has always done, there isn't a single parameter outside previously experienced limits.

How can it be said to have changed? /rhetorical

From that position we can plan for events that historically have caused Britain most problems, in my view these are
1. Extended cold - the LIA for example
2. Severe storms as in the LIA for example
3. Mismanagement by politicians - this is up to the individual.

Yeah sure Martin. But why not answer the question? Your mental model (which is a model of sorts just like a climate model - and you seem to be happy to trust its output despite it not being 'validated') cannot say exactly what effect changing CO2 levels will have, so even there there is a chance that temperature will change as a result of extra CO2 by several degrees. How big is the risk that this will happen? Shouldn't we try to understand it? How do you suggest we do this?