" Might I ask what strategies you would use to maximise robustness
under those conditions?"

You can't as the model is only a conjecture or a theory. Robustness would come from observing things or measuring things or conducting controlled experiments (which you obviously observe and measure too)

I agree with rhoda. Meaningless padding.
I am a retired NHS consultant physician and now work very part time as an NHS bureaucrat. 'Robust' is a word very much favoured in this world. When I hear talk of 'Robust care bundles' I struggle to keep quiet!

Rhoda, Martin A, Rob Burton

I wonder how an immunologist and an Oxfordshire housewife came to share the same delusion. Science is very rarely black or white, right or wrong, there are usually infnite shades of grey. It's the first time I've formally written it out, but my definition, or something like it has been in use in biology, astronomy and other such science for as long as I can remember. Martin A, you are a lucky man to be sheltered in your laboratory from such uncomfortable truths.

Observation and measurement is possible on all scales, but I still wonder how you expect to do replicated controlled experiments on a whole planet. Perhaps you have access to the Long Earth? I had hoped for positive suggestions, not the blinkered assumption that anything not done in the lab is not science.

Rhoda, it would help if you learned more about climate models before deriding them.

A model builder starts with the known knowns. The first is geography. The spherical world is divided into 50km squares. Each is given appropriate phtaical properties, desert, ocean, mountain etc. Predictable changes such as seasonal variations in insulation are put in place. Initial conditions are set; temperatures, water contents etc taken from past weather data or other sources.
Second is physics. The appropriate physical laws, as described by their mathematical formulae, are included. These calculate the energy content and energy flows associated with each grid square and its neighbours at appropriate intervals.

A model run entirely on past data will always give the same result and its properties can be fine tuned to improve its accuracy as a hindcasting tool.

To run the model as a forecasting tool the next thing to add are the known unknowns.Stronger or weaker solar cycles, variations in volcanic activity, variations in aerosol pollution, ENSO variations, differences in the rate of CO2 production,etc all affect the degree of forcing and the temperature. They need to be included, but their effect cannot be predicted in advance, so they are randomised on the basis of past experience.

Finally the model is run repeatedly. Runs with greater forcing show higher end temperatures. Runs with lower forcing show a lower temperature rise.

Which are better? Those whose forecast come closest to observed conditions. We are currently looking back from 2014 at model runs from 2007.We know that the known unknowns produced lower than average forcing during that period and warming slowed.
The models which happened to have low forcing gave results matching reality. Those with average or higher forcing forecast higher temperatures than occurred.

Just to complicate matters, the best models so far will probably not be the best in 30 years time. As the know unknowns increase or decrease forcing the models which best track that forcing will give the most accurate temperature forecasts. Which will they be? Nobody will know until the time rolls around.

Until then the results from the whole group of runs give an idea of the range within which the future temperatures are likely to be. No right or wrong, just probabilities.

Now, I have other work. Goodnight.

Martin

> Discussing things we have imagined is
> unlikely to lead anywhere, so best avoided.

If you, the Bishop or other 'sceptics' think the Daily Mail has a position in the normal scientific process then you are as guilty of discussing figments of your imagination as you suggest I am.

Objecting to the word 'robust' is just like objecting to 'consensus' or equally to 'acidification' or the various other words that 'sceptics' like to redefine to suit their purposes. It has little purchase outside your world and so is mainly a harmless game. The word might well be thrown around more than strictly necessary but then a lot of prose can be stripped of verbiage and lose nothing in meaning.

If you think the science is not robust, you are free to explain why. The authors using the word robust, clearly disagree with you though, so their use of the word is appropriate.

Chandra

"That is a polite way of saying that the Mail doesn't know what it is talking about."

You're not surely suggesting it does?!!!?

;-)

For what it's worth,

My definition of "robust" within the climate debate is analogous to a doctor saying "trust me".

Tis just the way my head works.....

Entropic Man

You say "just to complicate matters, the best models so far will probably not be the best in 30 years time. As the know unknowns increase or decrease forcing the models which best track that forcing will give the most accurate temperature forecasts. Which will they be? Nobody will know until the time rolls around."

I don't truly know your private stance on this "settled" science but your statement does, at least, imply that it is no such thing.

Yet nearly thirty years ago they were stating the AGW position as an axiom.....!!!? So in thirty years time we will have even greater "certainty"?. How does that cognitive phenomenon work? I don't know myself but would be grateful for guidance.

To cite the Rumsfeld quote (which I think is rather wonderful by the way) also is to suggest the "unknown unknowns".

My point I would suppose is that we are embarked on absolutely ruinously expensive policies which ARE killing many many people...........and for....what?

a "belief".

The precautionary principle?.

I hope you can see what I'm trying to get at here?

Andy

Or to a politician saying "trust me"?

Or to a politician saying "trust me"?

Ah......no, that's an easy-peasy one. His/her/LGBTG/etc lips are moving.

That IS axiomatic.

:)

Rhoda, Martin A, jones

Rhoda

You are missing the point. The " skill" of a model is the ability of a model to accurately simulate a process given correct input. When one runs a climate model to forecast future conditions the know unknowns cannot be forecast, which means that forecast can give an outcome between probable upper and lower limits, but not exactly.
If you can do a mystic Meg and tell the forecasters exactly how many volcanoes will eruprt in the next 50 years, the behaviour of the sun, how China will manage its pollution and how much CO 2 we will release each year; then you would get more precise forecasts.
Regarding initial conditions you can start with anything. You then have to wait 10,000 virtual years and several weeks of wasted computer time for the model to settle to equilibrium. Better to start with a set of known conditions, say 1950, and make all your runs from there. It is also good practice. Starting all your runs from the same initial conditions controls for one possible variable.

Martin A

Are you sure that your concern with robustness is not a rationalisation allowing you to reject evidence you would prefer not to believe?

Jones

"The science is settled" is a sceptic straw man. Read you Karl Popper and you will discover the concept of falsifiability, by which a hypothesis can never be proved correct. It may possibly be disproved.
Martin A will tell you the same. His knowledge of immunology is probably correct, but never certain.
If you encounter anyone saying "The science is settled", they are probably a PR hack, a reporter, a politician or a sceptic. It's not something a working scientist would normally be expected to say, or think.

That should also clarify my own position. I rgard the current climate paradigm as a good match to reality, and a good basis for policy decisions, but definitely not aettled in the sense that a layman would use the term.

Think of climatology as a partially completed jigsaw. Unknown unknowns acting on the current climate would show as missing areas of the jigsaw. Many individual pieces remain to be filled in, but the lack if any big holes tells us that no big unknown unknowns are currently acting on the system.

Back tonight.

Thank you for that EM.

I'm not entirely sure you read my post correctly but it's also quite possible my use of words confused too. If so apologies. However.

You state "Unknown unknowns acting on the current climate would show as missing areas of the jigsaw". Is this your own understanding of unknown unknowns? That would appear to be a known unknown but I may well also be misunderstanding it myself so don't mind me.

I am having some trouble accepting that you are not being willfully sophistic in suggesting that a proponent of science being settled is the sceptic position.

I would certainly ask that you quantify that particular assertion.

Thank you most kindly for the tutorial on the philosophy of science. I really did feel the teacher in you at that point. Honest.(The written word unfortunately leaves out huge swathes of actual meaning in case my words evoked the thought I was only being sarcastic. Most communication is non-verbal with the written word actually conveying even less than that. Sorry, that was the teacher coming out in me!).

May I ask you please your own view on what would prove to be falsifiability criteria given Popperian principles?

I thank you for your response. It was civil.

Andy

I' m interested in exploring your concept of robustness further. In studying climate, comparison and modelling have to substitute for replicated cotrolled trials. Might I ask what strategies you would use to maximise robustness under those conditions?
Apr 3, 2014 at 3:19 PM | Unregistered CommenterEntropic man

I think you are missing the point. I would no more try to maximise "robustness" than I would try to maximise "novelty". I regard it as a linguistic term, not a technical one that can be defined and measured. As such, I think it holds roughly the same meaning in climatology as in other disciplines, but subject to the practitioners own perceptions and standards. Others may describe as "boilerplate" or just plain BS. Its use certainly adds nothing except the authors opinions of their own work.

Martin, I think it reasonable to think that those who caused your ire by using the word 'robust' did indeed consider their science to be robust in the common sense of the word (which your ad hoc definition seems to capture well). All you are doing with your attack on the word is to try to suggest that they are instead deliberately lying - that they know the science not to be robust. Such accusations of dishonesty in scientists are of course the bread and butter of BH and other 'sceptic' blogs. The Bishop is doubtless quite happy for you to make such claims - care to identify some lying scientists by name?

Martin, I think it reasonable to think that those who caused your ire by using the word 'robust' did indeed consider their science to be robust in the common sense of the word (which your ad hoc definition seems to capture well). All you are doing with your attack on the word is to try to suggest that they are instead deliberately lying - that they know the science not to be robust. Such accusations of dishonesty in scientists are of course the bread and butter of BH and other 'sceptic' blogs. The Bishop is doubtless quite happy for you to make such claims - care to identify some lying scientists by name?

Martin, you claim that scientists are deliberately misrepresenting their work as 'robust' when they know it not to be so. That is an accusation of dishonesty (aka lying) in any language. So come on, name the accused.

Robust = Best ever washing up tablet etc.

Our next best ever will be along shortly.

Welcome back EM.

So settled science is now a sceptic word is it? Was it the sceptics telling the world the "science is settled" or was it the sceptics that were saying that the science was far from settled - even Trenberth came out and said the science is not settled at the height of this particular sound bite. Science is the business of getting asymptotically closer to the truth all sceptics know that. It is. a sign of insecurity if you have to project your own words onto your opponent when it becomes obvious you're wrong - actually it's infantile behaviour

Science is a journey with no end in that a hypothesis can never be proved to be correct, the mechanism used by scientists is to make forecasts for the physical world based on the hypothesis. We don't need the world in a laboratory for that.

So how are the scientific forecasts doing? Let's just take two forecasts. According to the Met Office the global temperature should have risen by 0.3C between 2004 and 2014. They didn't in any other science that would make the science look a little less certain. For the Met Office it makes it more robust.

The second is the accumulated heat - it's not there. For climate science the defence for this is to state that it is there, we just don't know where. After floundering around they've come up with the solution that it must be hiding in the deep oceans, which it has apparently managed to do without passing through the Argo buoys. Worse yet, while all this heat has been tunnelling into the deep oceans it's stopped accumulating in the atmosphere.

The famous "pause" that has been denied -accepted and denied again on the basis of a paper by a couple of novice scientists who work out of a web site interpolating temperatures in the Arctic and Hey Presto! Newbie and Rookie's have restored all the warming while literally thousands of climate scientist being paid for their efforts couldn't figure it out.

I'm continuously reminded of the dead parrot sketch in Monty Python's Flying Circus as scientists step forward after another failed prediction and say, it hasn't failed it's just sleeping.

The problem with this prediction business based is that that the prediction based on your hypothesis can be right a thousand times and wrong only once. In climate science it appears to be the opposite a thousand wrong predictions with one right one among them confirms the hypothesis.

And yes we all know that when making a projection from a climate model there are a range of possibilities, but surely when everyone of the projections are at, or below the 5% range you would think, wouldn't you, that some doubts about the way the models have been put together would begin to creep in to the climate science community? Far from it the evidence has gotten more "robust" not less.

So I give you: Robust: Impervious to evidence.

Robust = programmed in Fortran