Sceptic Links Discussion

[Admiral_Bobski]

C-Bob, if the models were the only tool in the arsenal of policy making, then your point would be quite strong. I actually agree with you that sometimes, too many assumptions about future conditions are based on the (unsubstantiated) predictive skill of models, especially RCMs and single or short-scale runs...

Hi P3

I won't quote the whole of your post - I hope you don't mind

This is all getting a bit off-topic, but if I may quickly say that I have the impression that the policy-making (certainly in this country, if not others) is heavily influenced by the models "this is how bad it's likely to get" predictions. I'm not trying to say that the government and/or their advisors rely only on the models, but that the models are, shall we say, overutilised.

The original point of this topic of discussion, though, was not about how the models are used but how good they are.

I confess that it was I who took the discussion off-topic with my quote from the link you posted, P3, but there was the author's comment about the accuracy (or otherwise) of the models. My apologies.

CB

[parmenides3]

I think it can be reasonably well established that climate models can do some things well; they capture certain physical realities and process them in a way which strongly resembles real events. But only some of them.

I think it can also be reasonably well established that, in combination with ongoing observations and hindcasting, the models can give answers to question about some effects within a margin of error which is scientifically acceptable. But only some of them.

From these, there needs to be a certain amount of inference done - though teams work hard to reduce this as much as possible, for reasons that I'll mention in a minute.

What they can't do is anticipate every response to what looks like a given; ongoing warming and the feedback effects. In this, the predictions/projections/forecasts are, if you like, plot synopses rather than the finished story.

The question is; how much of the story do we need to know to make a judgement about what the synopsis tells us? And how often do we need the synopsis to be repeated before we get the point?

It would be best if we could read the whole story, but this is a technical impossibility; all we can ever hope from in a simulacrum is an approximation of the real; a virtual reality of a kind. The problem is that the storyline is a bit scary; what we're getting (imperfectly) are signs of bad things ahead. We can't model reality perfectly, we can't predict with certainty, but we still need to try to understand the processes, the dangers, and the limitations of the models themselves; so the work goes on. You have been arguing the case that the plot is too thin and there's not enough detail to work out what response if any is best; I am arguing the case that, imperfect as it is, it not only is adequate (for the big picture, at least), but it is also imperative, that we respond to what the models tell us.

It has occurred to me that some of the issues we are touching on are less about climate science in particular as about all science; questions about the independence of the observation, the loading of the experiment, acceptable degrees of uncertainty. These are all matters which fall more under the aegis of the Philosophy of Science than of CS alone. So, you have some important and valid questions, but they are more about the nature of science than about the reliability of the AGW hypothesis or of the models which are used to assess its impacts.

I am not sure where the burden of proof needs to be on the question of the viability of models as predictive tools; is it me who needs to show that they are good enough, or you who needs to show that they are not?

:)P

[Admiral_Bobski]

The question is; how much of the story do we need to know to make a judgement about what the synopsis tells us? And how often do we need the synopsis to be repeated before we get the point?

Well, if this were a murder-mystery novel then we could (and have) quite easily come up with a prime suspect. However, if we've skipped several pages, or several chapters, of the novel then there is a distinct possibility that we may have fingered the wrong man, having missed relevant information which points to the true culprit. The counter-argument to this is, of course, that if we have enough of the story then a few pages don't make any difference - we'll still arrest the right person - but it doesn't appear that we have read enough yet to be able to make the arrest.

I am not sure where the burden of proof needs to be on the question of the viability of models as predictive tools; is it me who needs to show that they are good enough, or you who needs to show that they are not?

I suppose this is really the crux of the whole argument: should the climate models be used as predictive tools? Generally, in science a model is used to clarify or test a theory - they are generally only used predictively when they are used to predict a particular thing which is testable (and by "testable" I mean able to be tested immediately or in the near future - making a prediction which can't be confirmed or refuted for a hundred years is fairly pointless in scientific research).

So, I'm not complaining about the models per se, but rather I am complaining about what they are used for.

:lol:

CB

[parmenides3]

On this, we are going to (largely) agree, again. There's a big difference between seeing a signpost and having a map.

The thing I object to is the way in which a piece of research is based on a model run which provides just one likely pathway, makes conjectural conclusions which are entirely conditional (ie; if the model run were right, then this would happen...), and then gets written up as if the finding is written in stone. the press are especially adept at this; hence 'possible (conditional) consequence' becomes 'will happen'.

I actually think is is against the interests of sensible AGW science to have such material presented this way; it doesn't make people aware of the real risks, but diverts them instead to a believe (or cynicism) about the details of some specific subject or other. It also means that when a serious piece of work reaches a strong conclusion which is robust, people tend to assume it has the same credibility as the nonsense reportage, which, in the case of AGW, should be a cause for concern.

:)P

[Gray-Wolf]

http://www.foxnews.com/story/0,2933,296060,00.html

more positives for you sceptics!

[HighPressure]

Hi CB

Well, if this were a murder-mystery novel then we could (and have) quite easily come up with a prime suspect. However, if we've skipped several pages, or several chapters, of the novel then there is a distinct possibility that we may have fingered the wrong man, having missed relevant information which points to the true culprit. The counter-argument to this is, of course, that if we have enough of the story then a few pages don't make any difference - we'll still arrest the right person - but it doesn't appear that we have read enough yet to be able to make the arrest. :blink:

The problem here is we do not know if one page is missing or half the book.

I suppose this is really the crux of the whole argument: should the climate models be used as predictive tools? Generally, in science a model is used to clarify or test a theory - they are generally only used predictively when they are used to predict a particular thing which is testable (and by "testable" I mean able to be tested immediately or in the near future - making a prediction which can't be confirmed or refuted for a hundred years is fairly pointless in scientific research).

So, I'm not complaining about the models per se, but rather I am complaining about what they are used for.

CB

I do think that when it comes to models people get prediction and projection mixed up, I would say that what climate models do is provide projections and not predictions! Projections are provided by a mathematical model given boundary calculations and formula, by varying these its possible to achieve several different projections. However these are not predictions as far as I understand it, but a tool to help human's ultimately make a prediction, which may or may not encompass several of the projections. My argument here is that models are correct given the boundaries for which they are set, but whether human predictions are is quite another matter!

[jethro]

There's no where obvious place to post this link so thought I'd pop it in here. I've had a quick read through, it doesn't appear to explain where the bottom heat came from, unless my early morning head is missing something glaringly obvious.

Another piece of the puzzle perhaps?

http://www.eurekalert.org/pub_releases/200...c-cdd092507.php

[parmenides3]

There's no where obvious place to post this link so thought I'd pop it in here. I've had a quick read through, it doesn't appear to explain where the bottom heat came from, unless my early morning head is missing something glaringly obvious.

Another piece of the puzzle perhaps?

http://www.eurekalert.org/pub_releases/200...c-cdd092507.php

Morning, jethro. This looks like an interesting piece; I'll do some follow-up and see what comes up, I'm not sure what makes this especially original, though, unless it's the conclusion that changes in the Antarctic drove NH changes; this appears to run counter to most ideas I have read. My early morning head hasn't gone away either, so give me some time...

:)P

[jethro]

Morning, jethro. This looks like an interesting piece; I'll do some follow-up and see what comes up, I'm not sure what makes this especially original, though, unless it's the conclusion that changes in the Antarctic drove NH changes; this appears to run counter to most ideas I have read. My early morning head hasn't gone away either, so give me some time...

:)P

Cheers P3, when I get a little more time I'll have a look around too.

D