Monday, 21 October 2013

Timescale for testing scientific hypotheses: the longer the timescale of hypothetical change, the longer it takes to test the hypothesis


I learned from medicine that it is much harder, sometimes in practice impossible, to test therapeutic claims for drugs and treatments which are supposed to affect long-term and unpredictable processes.


For example, it was easy to show that lithium had a calming effect in mania, but very difficult to show that when lithium is taken long-term ('prophylactically') it can (probably, potentially, and with severe disadvantages and side effects) reduce the frequency of manic episodes.

But the 'mood stabilizing drugs' which have been marketed so successfully to replace lithium were adopted within a timescale of years, not decades, and therefore before there had been any possibility of testing their effectiveness in the treatment of such a long-term and variable condition as mania.

Therefore we know for sure these drugs were in fact (not as a matter of opinion) adopted without evidence, on the basis of hype.


Another medical example was the supposed effects of diet changes on the incidence of coronary heart disease.

These dietary changes (low cholesterol, high polyunsaturates, low fat overall etc) were recommended and implemented at a population level long before there had been any possibility of testing whether they a.) worked or b.) were overall beneficial - since this would have taken decades to establish.

Therefore we know for sure that these dietary changes (and others such as the eating of 'five a day' fruits and vegetables) were in fact adopted without evidence and on the basis of hype.


The necessary and sufficient ground for rejecting the claims of the advocates of 'man-made climate change' is that there is no way they could possibly know whether their claims are true.

(Or to put it the other way about: it is im-possible - i.e. not-possible, i.e. literally beyond possibility - for the climate change activists to know what they claim to know.)

In other words, these people are claiming to possess knowledge truly vast in its scope: they claim to be able to understand, predict and influence the climate of the earth. It is not a matter of whether they are right or wrong in their claims - but that their claims are pre-evidential; they have not even reached the level of being able to discuss whether they are right or wrong.

They are just ideas, notions, suggestions - not even formulated sufficiently to count as formal-hypotheses. 

Because of the very large temporal (year-by-year) variation in climate/ temperatures, the timescale over which one can be confident of measuring a climate trend is of the order of many years - in fact, a few decades.

In other words, because temperature varies so widely according to time and place, we could not know whether we really understood the causes, really could predict the future, and really could engineer global climate without many decades of measurement and testing of hypotheses.


It follows that the timescale of 'a few decades' would also be the absolute minimum time over which it could be tested whether hypotheses/ models of climate change were valid.

Thus it is necessary to: 1. put forward a specific hypothetical model of climate change, 2. use the model make precise predictions of future climate, and 3. measure future climate over a period of decades to compare the predictions of the model with what is observed.

Only if future observations matched with predictions would it be reasonable to assert that the model of climate change was accurate; and only if the model was accurate in these predictions would it be reasonable to believe that understanding of the causes of global climate had been achieved; and only if changes (against trend) had been mapped against outcomes would it be possible legitimately to claim that by changing the magnitude of these causes there was any genuine possibility of really controlling the earth's climate.


But this necessary process, lasting a few decades, was not followed. There never was a decades-long period between the construction of hypothetical climate models and their testing.

Therefore, we know for sure that the theories of anthropogenic climate change, the focus on carbon dioxide, the vast multi-trillion dollar government programs to change the world economy to prevent global warming -  all of these and the rest of it was in fact (and not as a matter of opinion) adopted without evidence, on the basis of hype.

Without this necessary period of testing; claims to be able to understand the causes of climate, predict future climate changes, and influence the future global climate have zero validity.

As the TV astronomer Patrick Moore used to say when he had reached the limits of solid knowledge: We just don't know



Whether there has been global warming in recent decades (in the past, from here backwards) is a matter of evidence and interpretation.

But, whether there will be (in the future, from now onwards) global warming - and whether future climate change is controllable by human action - is unknown, a matter of sheer assertion, purely conjectural.

And this is not just a matter of opinion. To disagree is proof of either scientific incompetence or personal dishonesty (or both).


dearieme said...

Globalwarmmongering is notable for the huge ratio of imposed costs (to the world's population) to achieved benefits (to the small cadre of "researchers" who have gained tenure, promotion, fame and power).

Anonymous said...

When talking about the timescale over which information has to be available in order to monitor a process that occurs over decades, we shouldn't so easily discount the centuries of data on, for example, past temperature and CO2 levels we have obtained simply through digging through peat bogs. Or the millenia of past data if we start to look at proxy data through ice cores. Suggesting that we have not observed the processes operating because of the long timescale ignores all of the work done on the bountiful historical data available. You asked also for a working model, that must then be tested against future predictions, by predicting and then waiting. I would expect that testing by seeing how well a hypothesised model can reproduce historical conditions is an equally valid test, so long as the traps of circular logic are avoided.

Though it says nothing about climate change as a whole, it is also possible to make some predictions about future climate from a basic knowledge of physics. As one example within climate studies the link between CO2 and the greenhouse effect was not sourced from a line on a graph of past change, it was postulated as early as the 1900's because of an understanding that CO2 molecules refract infra-red light. No complicated links or narratives necessary, more CO2 in atmosphere = more heat from sun retained.

I don't wish to argue over specific points in this debate, merely to point out that suggesting the process is unobservable and therefore not testable in principle ignores both historic evidence (giving us a much longer view, a long window of observation) and arguments based upon a knowledge of elementary processes, that are themselves provable and reproducible in a lab environment. That is not to say that anyone explicitly knows the future. If anyone did there would not be error bars on the estimates. But one can make a pretty good guess when given some understanding of the mechanisms and some data on how a system has behaved historically.

I'm afraid I cannot comment on this from the point of view of pharmaceutical testing, I do not know how well they understand their biological mechanisms, or how much usable historical data exists for such studies.

Bruce Charlton said...

@Anon - (Please use a pseudonym at least!) - in a word: NO.

Why should the normal, classical, successful ways of doing science be re-written for climate research?

Because the problem of predicting and controlling the world climate is so simple and obvious as not to require testing or confirmation?

Or because the people who do climate research are so clever and honest compared with the best scientists of the past?...

I have been an epidemiologist and know from experience that back-fitting models to fit past data is fraught with perils and self-deception (and dishonesty and incompetence); and the resulting models should always therefore be tested prospectively in whatever ways are possible, with predictiosn being monitored and feedback loops being checked at every stage.

This is necessary because the human body is extremely complex and hard to predict and hard to influnece. And so is the climate of this planet.

Jonathan C said...

In global warming, the debate almost never mentions that all the interventions are founded on not one, but two extraordinary claims:

1) Humans cause substantial global warming;
2) Substantial global warming would be very, very bad; so bad that it would be worth spending many billions to prevent it.

Claim (1) seems to be the focus of almost all the debate, and claim (2) is rarely mentioned at all, almost as if the opposition has ceded the point. This has always puzzled me, because claim (2) strikes me as the more unlikely of the two claims.

I think everything you have said in this post applies to both claims, not just the first one. We have no scientific basis for either claim.

But insofar as past performance is the best predictor of future performance, the geological evidence suggests that previous times of high temperature and high CO2 were times when earth teemed with life and food was plentiful.

Bruce Charlton said...

@JC - Actually the claim 2 you state contains the assumption that human can control the earth's climate in the direction and to the extent they want. There isn't one scrap of evidence that this is true, and it is massively implausible.

EVEN IF it were correct that humans had caused global warming by CO2 emissions, this does not have any implication whatsoever that this change could be reversed by human action.

There are innumerable examples of humans being able to disrupt or break complex systems that they then have zero ability to remedy or control - there are many examples in medicine. For example it is easy to poison somebody, using a vast range of agents and doses, after which they cannot be cured.

If humans really have caused global warming, and overcome the natural homeostatic resistance; then it would be most likely that the change would be impossible to stop and would go a long way.

This is, in fact, what the 'father' of global warming theory James Lovelock always used to say (he also used to say that methane production from mass agriculture - cow farts etc - was *far* more important than CO2).

Jonathan C said...

I guess I didn't phrase claim 2 very well, but my intention was to separate out the question of whether humans can affect global temperatures (claim 1) from the question of whether higher temperatures are good or bad (claim 2), regardless of the cause. From your response, I still cannot tell whether you are even faintly aware of the possibility that global warming (regardless of the cause) might be a GOOD thing.

Although we cannot know with any reliability whether global warming (regardless of cause) would be a good thing or a bad thing, I think what evidence we do have suggests it is at least slightly more likely to be a good thing than a bad thing.

I do not understand why this is never discussed in global warming debates. It seems to me that everybody implicitly buys into the idea that global warming would be a bad thing, and every discussion of global warming (even by skeptics) proceeds as if the badness of global warming were Newton's Fourth Law, when in fact it strikes me as a radical and unlikely claim.

Bruce Charlton said...

@JC - Oh, I am completely familiar with the argument over what would be the optimal temperature for the earth - it crops up quite a lot. Warmer climate would benefit the UK standard of living almost certainly - looking at history.

But the idea has no traction in public discourse because 1. it gets complicated and the answers aren't really known, and 2. the AGW activists simply focus on some poor place somewhere where warming would probably make things worse, and blame it on the West.

However, I prefer not to talk about this because (at a psychological level) it fuels the wishful-thinking, zero-evidence fantasy that humans can control the climate to suit themselves , plus the world government transhumanist fantasy that IF we did have such climate control it would be used to benefit 'humanity'.

It also assumes that the main constraint on life on earth is climate - or standard of living; that the purpose of life on earth is comfort and convenience; in other words, it is worldly. Environmentalism is supposed to be unselfish, in practice utterly focused on this world and spiritually me-focused.

It therefore went from being focused on nature descriptions and self-sufficient low tech farming to international conferences and flashy mass media campaigns in a single generation.

But the fact that pretty much all the pioneer environmentalists went along with Al Gore environmentalism (indeed avidly embraced the opportunities) shows their careerism, hedonism (e.g. sexual) and superficiality - shows that environmentalist movement not just is, but always was, a fake; and its good qualities were merely temporarily residual.

The sensible and realistic attitude is to regard climate changes as something we cannot predict but just have to cope with - as best we can.