Challenging the sceptics

NERC has a bold new initiative... the NERC Climate Change Challenge. As they say Scientific evidence demonstrates clearly that human activity is changing the planet's climate. But there are still sceptics who dispute the data and its interpretation. If you don't believe the science, please tell us why and we'll respond to your challenge. Or you can have more detail. There isn't much there yet, so I think Lubos should pile in.

Since they lean on the Stern report, I thought it would be fun to ask them about high climate sensitivity. Maybe they will ask James as one of their experts to reply... :-). Ah no, I've just found the Team of Experts: Professor Alan Thorpe, Professor Julia Slingo, Professor Peter Cox and Professor Colin Prentice. Obviously busy folk, because they haven't found time to answer yesterdays inaugural question from... someone in Swindon.

I found out about this from the NERC internal magazine. I wonder how many sceptics read that, so I also wonder who else is likely to find out about this...

More like this

Well, as Colin Prentice couldn't be arsed turning up to his talk at AGU a few years back, I can't imagine that this was top of his list of priorities.

[I guess there is an under-team of people to supply the answers. Or maybe not: no-one is answering yet... -W]

I was an undergrad at the time, using a lot of his papers and was (rather sadly) looking forward to seeing his presentation. I'm still bitter about it now!

I also note that the "scientists predicted an ice age" canard has had an early showing!

[Yep, the nutters are showing up, there is even "its all caused by UV" -W]

Do we really need *another* forum for this same tired chit-chat? The naievety of those science types - If we could just get them to understand the science! This might do it! - never ceases to amaze.

[See Gavins comment -W]

John, are you suggesting that 'us science types' should stop trying to explain things to people? I agree that these kinds of fora (like this blog or Realclimate) often appear to have a naive idea about public education. But NERC (like RealClimate) has a primarily science-based mission and explaining things to people is a good thing in and of itself, regardless of the overall effect. You may be confusing this with attempts to influence policy through PUS (Public Understanding of Science), which I would agree seems pointless. But in my experience faciltating direct access to scientists from the public ends up helping both - (given of course that some scientists should never be allowed out in public). Would you prefer that access only be granted to journalists? (grin)

[Hmm, I'm rather with John on this. It looks like its going to fail on the "noise" grounds: there are too many wackos out there. Also (unless they get off their bums and actually answer some questions) its going to fail on the lack-of-response-grounds, which will be silly. Since they have had nothing but softballs so far (with the possible exception of mine...) its unimpressive -W]

Gavin, I think that what John is suggesting is that if you enjoy futility you will try and explain things to denialists. They have no interest in learning. Talking climate science with Fred Singer or Pat Michaels is pointless. That's not what they get paid to do.

They simply are not interested beyond the opportunity to manipulate you and score points with the public. On the other hand, if you aim at countering their propaganda, you, and Real Climate, have something going. I often point out the asymmetric nature of the debate, and you can judge whether I am right from the furious response that is generated.

Beyond that there is value in what Real Climate set out to do, and to some extent is succeeding at, providing a place where the public (and kids doing reports for school, etc.) can go and get good information.

As for myself, I've picked a slightly different niche, deconstruction of tactics used by the Exxon 50, using factual fisking and good (ok that's a value judgement) humor as often as possible.

I find Realclimate very useful, as a compendium of information and explanation. I've even been thanked for directing someone to it. The fact that denialists I debate then ignore the science I throw at them, only goes to show them up. I got one of them so riled up by repeatedly pointing out he had no science backing his view that he started insulting me.
(Note for other denialists etc- all he could come up with was some wibble about entropy. Now, I'm not a thermodynamicist, buta chemistry degree gets you familiar with the basics, and he wasnt even quoting them.)

I quietly hope that the real scientists dealing with public comments are keeping track of IP numbers used for comments, to know when someone's trolling rather than honestly asking naive questions. I don't envy you all the task of dealing with the public, especially the anonymous public.

By Hank Roberts (not verified) on 20 Dec 2006 #permalink

Well, they patted back your softball :-)

Can hardly expect them to talk about this sort of nitty-gritty in detail when the scientists actually involved in the research are so reticent...

[I've patted it back and explicitly referenced your rejected paper. Still it is good to see some answers being given -W]

It can't succeed. Inspired by you,WC, Gavin and the RC crew & various others, I've been posting on a general weather forum for some months now. Two things are apparent; it seems you have to deal with each 'sceptic' or doubter one at a time; they just don't read the background material, and also, when you do give the Thomas's all the evidence, supporting material, etc., etc., the ultimate response inevitably boils down to a version of 'I don't really believe it'. As has been attested many times, changing people's minds when they are already made up is almost impossible. I give the NERC about two months before they throw their hands up in despair and say 'we can't cope with the responses'.

Keep up the good work, and Merry Christmas.

By Fergus Brown (not verified) on 21 Dec 2006 #permalink

May I humbly point out that the pat contains a (to me) important point that I raised here a while ago, that the pdfs are skewed to the high side. If there is anything that worries me about A&H it is that the Bayesian analysis is not nearly as skew as the inputs, but maybe that is the point of the whole thing.

[The PDFs have to be so skewed, since they are bounded below by zero. But the degree of skewness? That is the entire discussion. If you knocked out the entrainment datapoints, the skew (from CP.net) would reduce a lot. AT has replied to my reply, and I have pursued... -W]

CP is not the whole ball game, all of the pdfs are skew, in fact a lot of them are inferred from observations not models ( see page 45). Further, why could the sensitivity NOT be negative (well, ok that would be weird physics, but that is the point)

[OK, in principle it could be, but in practice one of the few things anyone will agree on is that its not. In fact there are pretty good observational lower bounds at about 1.5 oC.

And true, CP.net isn't all of them, but its the big ones that people have been quoting recently and they are probably *wrong* -W]

Right, now you are making James' argument, which I agree with. Given imperfect data you cannot start with a prior far removed from reality.

However, my argument is different, and I WILL note that A&H (you got me doing it too) do not consider CP or models (at least all the time). CP is a bit of a red herring here.

The best I can do is the following. Any distribution can be described by its moments (mean, st. dev., skew, kurtosis) . Does the Bayesian analysis reduce the effect of the higher moments relative to the lower ones, or simply wipe them out?

If this is true (and it looks to me at first glance that it is) and you start with a number of distributions skewed to the same side (does not matter to which one), then a pdf from Bayesian analysis underestimates the probability on the skewed side.

>[The PDFs have to be so skewed, since they are bounded below by zero. But the degree of skewness? That is the entire discussion. If you knocked out the entrainment datapoints, the skew (from CP.net) would reduce a lot. AT has replied to my reply, and I have pursued... -W]

You are doing it again - having a go at CPDN. Knocking out the entrainment variation reduces the histogram distribution as shown in Stainforth et at but that isn't a PDF. Introducing more parameter variations than the 6 used in Stainforth et al would likely increase the range/Std Deviation/Skew of the distribution wouldn't it? Surely increasing the number of parameters varied is a more appropriate thing to do than arbitrarily knock out the entrainment parameter simply because you don't like the model being sensitive to that parameter. If the model is particularly sensitive to the entrainment parameter then surely it is really important to vary it not to ignore it.

At least CPDN is doing something about the sensitivity to this parameter and finding some better values to use.

Also please compare like with like. Stainforth does not have PDFs so please use Piani et al which has a 5:95 range of 2.2 to 6.8K and it is quite clear it is only using 20C warming observations. (So if you use other evidence as well the range will be reduced further.)

Eli, Piani et al suggests a reason for the distribution being skewed in that way.

Yes, I am having a go at CPDN. Why should they be immune?

Histogram/PDF is a fine distinction of no importance to this discussion.

knock out the entrainment parameter simply because you don't like the model being sensitive to that parameter - why are you saying this? You *have* been paying attention and you know full well that Palmer has shown there are good reasons for thinking that the CPDN values are unphysical.

At least CPDN is doing something about the sensitivity to this parameter and finding some better values to use - unfortunately CPDN may well have mislead a lot of people about the probability of high values of CS - just look at the Stern report for example.

Don't you find the disparity between the blanket coverage of the high CS values from CPDN and the near-silence of J+J's stuff rather significant?

>why are you saying this?

Do you really know whether all models have this sort of sensitivity to some parameter or other?

Also you use the same model.

>At least CPDN is doing something about the sensitivity to this parameter and finding some better values to use - unfortunately CPDN may well have mislead a lot of people about the probability of high values of CS - just look at the Stern report for example.

That seems a very different reason for having a go at them. So are you saying one reporting problem means you are allowed to have a go at them on any issue even if they haven't done anything wrong (and in fact are making progress on that subject)?

Maybe you are just bored with having a go at them for the same old thing all the time?

I think histogram/PDF is important to this discussion but you have obviously made up your mind that it isn't. Wonder if Mike Atkinson pulling you up on that has anything to do with that unexplained decision.

I do find the disparity significant. Though if it suited your argument, would you be pointing out that one is peer reviewed and the other isn't?