Sunday
Jan292012
by 
Bishop Hill
Bishop Hill All models are wrong
Jan 29, 2012 Blogs Occasional BH commenter Tamsin Edwards - a climatologist at the University of Bristol - has started her own blog. About time too!
It's called All Models are Wrong, the corollary of course being that some are useful. I predict that this will become a must-read.
Tamsin's first post is about the travails of coming up with a blog name that wasn't going to bring down the wrath of one side or other of the global warming debate. I think All Models are Wrong is fine, although I couldn't help but be reminded of this. :-)
Reader Comments (38)
1. Thank you!
2. Ha ha! And "Oi..."
:)
All models are wrong, but some are more wrong than others.
Richard Betts, being more optimistic, suggested:
sub-heading "...but some are more useful than others"
:)
It is useful to remember that models are not reality either in the fashion industry, mathematics or any branch of science although they might be useful for a variety of purposes.
Hopefully, with all the extra CO2 around, and therefore crop yields going up, those skinny models in the video will be able to afford enough food for them to be able to put on a bit of normal weight. I've never found famine sexy.
None of the models can be correct because the physics underlying them is broken.
1. Tyndall’s experiment was done at constant volume so because CO2 has higher CTE than air, much temperature rise was due to increased pressure. Do the experiment at constant pressure and the much lower heating is probably at the walls of the container. As the vibrational oscillation is quantised, you can't transfer it in dribs and drabs to symmetrical N2 and O2 by collision; thermalisation is indirect.
2. To shield upward IR by the radiometer thus removing the Prevost exchange offset then claiming ‘DLR’ is from a heat source is seriously dumb, There is no ‘back radiation, no strong positive feedback, an artefact of the modelling.
3. Hansen’s 33 K present GHG warming includes lapse rate warming so is also dumb; reduce it to ~9 K.
4. In 2010, experimental data showed the climate models use double real optical depth also 40% of low level clouds with bimodal droplet size distribution have different optical physics.Net AIE is slightly positive, much greater in the past and polar atmospheres. It explains present Arctic melting and much present warming. A side effect is that because data processing from albedo to optical depth is broken, you can’t trust satellite data.
So, the IPCC models are useless. This programme has been appallingly run. It’s time the farrago was ended. CO2-AGW may well be slightly negative due to self absorption near the Earth’s surface. The proof that it is low is that the N. Atlantic is cooling as the Arctic heads to the freeze part of its 50-70 year cycle.
Tamsin,
If temperatures continue bobbling along for the next few years, in much the same way that they have for the last 15 years, then very soon everyone will notice that the models have very obviously failed. And then, given that the models encapsulate the consensus view of climate science (as pointed out recently by Micky Mann), the whole CAGW facade will come tumbling down. Is your blog intended to get the retaliation in first?
Huh, so the planets NATURAL Co2 level is 350ppm. I did not know that??? :)
I'm glad I clicked on that link now here at home rather than in the office tomorrow morning - I think I would have got fired!
In fact I'm probably in trouble now as it is. This is a shared family computer!
Seriously though, best of luck to Tamsin with the new blog, I agree with BH that it is probably going to become a "must-read".
//"I repeat “wrong” is the wrong term".It WILL be misunderstood and misused//
Looks like TE decided wrong is the new right!?!...
Hilarious first post - Good luck Tamsin!!
All models suffer from exponential error. They just become more and more expensive random number generators.
Good for her! I have used that quote in any number of presentations or training classes on industrial statistics, a subject whose practitioners hugely admire George Box (the source of the quote) as an outstanding applied statistician, as well as a distinguished theoretician. It used to be quite a good way to make folks sit up and be clearer about what they were doing with statistics and why. The great advantage for the practical statistician is that you thereby sidestepped 'proving' a model must be right in favour of the more modest task of showing that it could, in the right circumstances, be useful.
The 'useful' bit is whether the model gives us good guidance for some practical decision, or provides a productive conversation piece for clarifying theories and identifying or designing good data collection plans to make further progress in understanding something - primarily by exposing hunches to test by observation or experiment, putting them in jeopardy if you like. The Normal distribution is a simple statistical model, and of course it is 'wrong', not least because it extends to infinitely large positive and infinitely large negative quantities that have no correspondence in reality, for example some measures are always positive - the length of a nail for example, yet the distribution might still do a great job of modelling the nail lengths coming off a stable production process and be of good practical value.
To be useful for decision-support in policymaking, the climate models must do at least as well, and preferably better, than more readily available methods - e.g. looking up an almanac, extrapolating cycles and trends, or even merely declaring that the future will be much the same as the past. The cost and complexity of the model is scarcely justified, for practical purposes, if cheaper methods can beat it or do as well as it.
But, as far as I know, and that is not very far I must hasten to admit, the GCM modellers have taken pains to avoid making predictions, coming up with such contrivances as 'projections' or 'scenarios' instead.
If they can't do useful predictions, beating cheaper methods, then they should be hustled out of the public square, and back into the relative obscurity of research labs - where their operators can exchange brochures of scenarios until the cows come home without much harming the rest of us.
I wonder too if the prominence of models has harmed the development of climate science itself. Complex, poorly understand software is a bit of a nightmare by definition - and may be no aid to clear thinking about the physics, amidst concerns over the stability of computational algorithms, and over the endless scope for wheezes and ingenuity on the part of programmers. Virtual worlds can be very alluring, and more tidy than the messy old real one.
Anyway, I very much look forward to visiting this new blog, where I hope I'll get insight into modern attitudes and capabilities of GCM operators, knock some of the rust off my own - no doubt terrribly dated - grasp of what these models get up to, and learn a lot of new stuff about them.
Coo!
So the more CO2 we produce the more the supermodels strip?
Where's the down side?
It's a great blog name but I'd have added two words "all models are wrong... until validated". And the models that have been validated must be updated or replaced when they no longer make accurate predictions.
In our annual business budget meetings we would construct from all available information and with suitably informed guesses set a budget, or forecast, for the coming year. As we formally approved the budget, the Chairman would remind us that 'all forecasts are wrong' in a paraphrase of the George Box quotation. The usefulness of the budget (business 'model' if you like) was that as you strayed above or below it, you looked at why, tried to understand the reasons for variance, and then took action to regain the planned progress or continue to exceed it. Much as with scientific modeling, the value is in understanding the variances and using them proactively to learn.
That's rather a long winded way of saying 'good luck' to Tamsin, I'll certainly be looking in.
models eh
elephant trunk wiggle.
Tamsin - What have you gotten yourself into. ;)
exceelent blog title, and your reasoning is sound, their are lots of highly educated people amongst the public/politicians, etc.. the whole mindset of the message, and 'confusing' the 'public' is so patronising, that you would think that intellegent scientists and self proclaimed communicators would realise this.
I remember when I first saw that video I thought
a) I was on the wrong side.. if you've met the 'sceptics' I have :( (sorry Josh,Andrew,Ben,etc)
b) so it's going to be getting hotter, doesn't seem too bad to me... !
c) I remember really regretting tweeting, when I used to 'model', at Reading Uni (forgot to say computer !!)
Peter Gleick thoughts about me/Bishop Hill and sceptics in general were absoultely fascinating..
I think his worldview (similar to Trenberths/Mann's - I think - has greatly influenced CRU) and explains a lot.
blog post tomorrow.
That Gleick is one policeman, isn't he?
At least he is being open about it.
@ Shub Jan 29, 2012 at 10:38 PM
Agreed. One thing I've noticed in all the Gleickenshpiels™ that I've read is that - if nothing else - he uniformly (albeit often uninformatively) wears his warmist/alarmist heart on his <even when I'm wrong, I'm right> sleeve.
I'd argue that models are illustrations of hypotheses but are not tests of them.
That Gleick confuses this distinction is damning. That he is so far from being unique in this regard, among climatologists, is inordinately more so.
Good luck Tamsin!
Bob Ward rather lamentably suggests toning it down to 'howskillfularemodels'.
Faced with coining a name, be it a blog, book title, band name or new species is incredibly hard when actually faced with the challenge. Tamsin's choice is provocative and strong. I'm rather surprised she didn't add a 'but' (one t) on the end, (if only in genuflexion to her employer), and wish her well. Judith Curry has set a very high bar. We need a fearless equal, badly.
A question for Richard Betts, please. The Daily Mail published an article here:
http://www.dailymail.co.uk/sciencetech/article-2093264/Forget-global-warming--Cycle-25-need-worry-NASA-scientists-right-Thames-freezing-again.html
that has gotten some attention in the States. It refers to:
"According to a paper issued last week by the Met Office, there is a 92 per cent chance that both Cycle 25 and those taking place in the following decades will be as weak as, or weaker than, the ‘Dalton minimum’ of 1790 to 1830.
But of course, they didn't publish a link to this ground-breaking paper. The MET office web site does seem to have a sort of press release.... but I can't find the paper anywhere. Are you aware of any of this? Is there such a paper? Is there a link for the paper if it exists?
By the way, the solar cycle plot in the mail article, complete with copyright notice, appears to be lifted from David Archibald's post at WUWT: http://wattsupwiththat.com/2012/01/25/first-estimate-of-solar-cycle-25-amplitudesmallest-in-over-300-years/
Thanks.
In essence, it comes down to whether someone has the self-confidence to admit that the Hansen and Trenberth model is based on incorrect physics aimed at political ends rather than establishing the truth.
The worrying thing for those further down the food chain is that they know solar effects will probably dwarf CO2 and as the funding dries up, the true believers will be cast into the wilderness.
Thus Curry and Spencer have adapted by learning from the many expert commentators who have pointed out the major errors in the Hansen Trenberth physics. The move away from the fantasy back radiation and cloud cooling physics to hide it is welcome.
I don't know what the likes of the propagandists like Ward and Romm will do with so many new targets popping up. Gleick, who unlike the former, appears to be a genuine zealot instead of an opportunist, has been honest.
Such models of friendship are precious and rare,
but the friendship of models, is not...
M. Flanders, D Swann
My mistake. The 'but' is there.
Gleick, who unlike the former, appears to be a genuine zealot instead of an opportunist, has been honest.
Zealot yes but writing reviews of books not read by the reviewer is not a sign of honesty.
I just saw Jonathan Jones' final response to Tamsin et al on her blog name thread:_
@nmrqip: @richardabetts Yep. Lying “to avoid being misunderstood” never ends well @PeterGleick @flimsin
When the dust has settled on all of this and we choose an epitaph for the "climatology"'s tombstone - Jonathan's phrase has to be a prime contender.
Thanks for all your comments. Too many here and at AMAW for me to reply individually! I obviously touched a nerve ;)
Firesign Theatre: "Everything You Know Is Wrong"
Models are tools. One tool can't be the best tool for every task.
The zeal of modelers to apply their tool to all climate science situations is a form of bias.
John
Tamsin Edwards' website is very promising. His introductory remarks suggest that he has a high level understanding of models and that he intends to explicate the issues for the common man. What more can one ask for?
Theo, Tamsin is a woman ;) I wish her blog every success. I hope she realises what she is getting into though. It's one thing to make the odd comment or two on someone else's blog and to play on the twitosphere, it's another thing entirely to put yourself up front and out there for all to see. She has my respect. Good luck to her.
Thanks Theo and LC :)
Pharos writes:
"Bob Ward rather lamentably suggests toning it down to 'howskillfularemodels'."
My suggestion would be "letuscountthewaysinwhichclimatescientistsmisunderstandmodeling." That is, it would be if it had a snowball's chance in Hell. So it won't.
LC,
Thanks for letting me know that Tasmin is female. And I understand the cautions that you express about her new undertaking. My advice to Tasmin would be to ignore or ban the Trolls and Fanatics. I call Judith Curry "Saint Judith" for the patience she shows to the Trolls and Fanatics. Given that Tasmin has selected a rather narrow topic, I believe that her blog just might take off and contribute something unique. There is reason for hope. I recognize a large number of people who have left messages on her introductory note and all of them strike me as serious people who will contribute judiciously.
Tamsin,
You are quite welcome. Frankly, I am just beside myself with excitement about your new blog. I promise to be nice.
Theo,
Very pleased to hear it (the excitement and the promise to be nice)! Thanks for the advice too.
Only one model can be correct, but all can be wrong.