Jack Stilgoe suggests that "scientists and politicians need to get better at talking about what we don't know", perhaps alluding the emerging field of 'ignorance studies' that books like this will do something to establish. Which reminds me of the main story on the front of Metro, a couple of days ago One Sausage a Day Ups Cancer Risk, which was helpfully accompanied by a photo of some rather fine looking bangers. As Metro put it:
Eating just one sausage or three rashers of bacon a day can increase the risk of developing bowel cancer by a fifth, it is being claimed.
Leaving aside the somewhat surreal headline (assuming nobody actually eats one sausage a day, are we now expected to evaluate our nutritional intake using a calculus of sausages/day?), just look at how much uncertainty is contained beneath one screaming headline.
Firstly, 'it is being claimed' that one sausage a day ups the cancer risk. Well lots of things are 'being claimed', especially in the scientific community. The claimant in this instance happens to be a scientific advisor to the World Cancer Research Fund, which gives some cause for confidence. His name also happens to be 'Prof Wiseman', which may have been the legitimating factor which tipped this over into front page news. But overall, the article focuses entirely on the claim, without even any hint of there being any proof or where to look for it.
Secondly, and more strangely, the article involves two levels of risk. There is the risk of bowel cancer happening (one fifth for any folks operating on uno porcus per diem), then there is the risk of this risk occurring. Note that the article states that eating these meats 'can' increase the risk of developing cancer by a fifth, not that it does do so. So there is a meta-risk at work here. For your next trick, Prof Wiseman, kindly inform us of what the risk of this risk occurring is.
And finally - and here I am straying into foreign territories of philosophy of science - statistical probability is surely not a form of knowledge, but a way of representing and organising ignorance. Knowledge would identify a causal relationship - this sausage caused this bowel cancer. Knowledge would identify those people who could safely eat dozens of sausages per day and still not develop bowel cancer.
It is precisely because we lack the financial, organisational and methodological resources to gather and process this type of knowledge that we resort to representing things in terms of statistical probabilities instead. It is because we can't, practically speaking, find out whether eating sausages will give an individual bowel cancer that we tell them simply that it might. A probability is a peculiar type of political-scientific device, that satisfies our need for quantification without directly attacking ignorance.
What the article actually means is "one in five people should avoid eating sausages, but we are not about to go and find out who they are". The risk is actually a social one, that exists in the mind of a policy-maker, but deliberately mis-represented as an individual one in the hope of altering behaviour.