BBC Radio 4 has a programme called Material World - basically science news. It recently featured a discussion with David King (previously Chief Scientific Advisor to the UK government) in which he was pretty scathing about homeopathy. Some listeners emailed to complain, and one used an argument I found interesting. He said that promotion of homeopathy is not science-based but it is evidence-based, by which I assume he meant... no, I won't put words into his mouth. This distinction I hadn't heard before. It sounds like sophistry. Could it have any merit at all?
I'm sure that Steve will jump in here, as it's one of his pet topics, but the difference between SBM end EBM is that SBM takes into account what we might call reality, whereas EBM looks at studies, without any cause to look at the world as we know it. So Homeopathy can pull up studies which show that it's marginally better than doing nothing, but you would have to throw out, or throw into doubt, large chunks of what we have learned over the past couple of centuries in the fields of physics, chemistry, biology, medicine and so on. The real fact is that homeopathy is not even any good in an 'evidence based medicine' frame - the few studies that support it are, to put it nicely, sub-par. Remember, there are studies that 'suggest' virtually any possible hypothesis - this is what fills second rate newspapers the world over ("Study suggests that preferring Simpsons over Futurama makes you a better lover!").
Perhaps we should extend the concepts beyond medicine to, say, Evidence-Based Knowledge (EBK) and Science-Based Knowledge (SBK). The practice of EBK without SBK is just the collection of data and the search for correlations. It is like the first stage of epidemiology: observing (for example) that the people who use a particular water pump are far more likely to get cholera than those who use another one, or that the people who work in the brewery and drink the free beer get it almost not at all. The SBK stage was to postulate a mechanism: infected sewage leaking into the water to the pump, or boiling of water in brewing sterilising the water .. and then testing it. Without that stage, all you have is correlation, which as we know may or may not be associated with causation.
Two prime, non-medical, examples of EBK to the exclusion of SBK are
Rupert Sheldrake, who needs no introduction, and
Piers Corbyn, weather eccentric, amateur climate change denier (in the sense that I think he really believes it and, as far as I know, receives no sponsorship from the usual climate denial industries or foundations), and believer that all climate and weather is caused almost exclusively by solar and lunar influences. He even claimed (based solely on alleged correlation) that the magnetic field cycle of the sun was (partly?) responsible for earthquakes and volcanic eruptions.
I've seen both of these speak on their chosen subjects. In the case of Sheldrake, twice six years apart, where the most notable thing was that the evidence he presented in 2011 was indistinguishable from that he presented in 2005, suggesting he had done nothing new in the intervening years and was content to plug books and talks on the same old stuff instead. In his latest book he seems content just to bash the scientific establishment without presenting anything much new of his own.
When either of these were asked to propose a mechanism, they said they had none, and more or less implied they didn't much care what the mechanism was. There would be occasional waffle about quantum entanglement or other forms of unseen connectedness of all things (reminiscent of
Dirk Gently), but nothing on which you could base a testable hypothesis. For the most part they simply said that there must be something going on because they could see all these correlations in the data, but it didn't really matter much how it worked. If only the blind, dogmatic scientists who rejected their work would accept their findings then presumably the whole wonderful edifice would be revealed.
To misquote
Rutherford, what these people do is not science but stamp collecting: a blind accumulation of data, carefully collated into groups which appear to correlate with each other (whether by chance or otherwise), with neither the ability nor the desire to try to explain why the data should be the way it is. It is tempting to think that they are afraid that if they analysed their data from this perspective then their cherished correlations would vanish in a puff of statistical noise.
