In Defence of Rumsfeldian Epistemology

December 7th, 2007

Views: 899

Posted by ChrisG at 3:28 pm

Well, tangentially, perhaps. Following Jack Straw’s recent use of the phrase, Rummy’s ‘unknown unknowns’ received some more attention recently. A lexicographer, Tony Thorne, responds rather testily to the use of the phrase:

“It’s typical of politicians, they either use the short, brutal, unanswerable soundbite or the convoluted like a cantation, a magical mantra to bamboozle people. First you have to compute it. If you actually try to decipher it, it takes five minutes to work out.”
But with phrases such as “bottom line” moving from the boardroom to the street, will we all soon be discussing “unknown unknowns”?
Only sarcastically, says Mr Thorne, because it does not fill what linguists call a “lexicon gap” in the English vocabulary.

Presumably, what it actually does not fill is a lexical gap. However, pedantry aside, what the phrase does do is open up a different kind of gap, which elevates the concept above the level of a target for derision. This has to do with risk, and the way taking risks is judged as involving culpability for harmful consequences.

In Rumsfeldian terminology, risk involves, first of all, ‘known knowns’. Take this example from Derek Parfit: if I break a bottle whilst camping and leave the glass in the undergrowth next to the site, should a child happen upon it at any point in the future and be injured, then that’s my fault.

Similarly, if I choose to ignore safety guidelines for storing uranium tailings near my uranium mine, and as a direct result, people in a nearby village have their water supply contaminated a couple of years later, then that’s down to me too. In both cases, my knowledge about the world at the time I committed a given act made me capable of a degree of foresight, i.e. knowledge that there was a significant probability my actions would result in harm, and I did nothing in the light of this knowledge. This is the standard moral and legal view of what constitutes culpability.

But risk also involves Rumsfeld’s ‘known unknowns’. These are probabilities related to items of knowledge we do not possess, but which are closely connected to knowledge we do have. This kind of ignorance is familiar from scientific research. By isolating certain naturally occurring processes under special conditions with fixed parameters, it is possible to determine, over time, the regularities that define these processes. As knowledge of these regularities advances, the researcher gains a vantage point, somewhere to stand so as to survey what is not known. What is still to be discovered might not be positively known, but the region of what remains outside knowledge begins, as time passes and more research is done, to take on a more defined negative shape. In other words, we know with increasing assurance what we still need to find out.

Science, in other words, advances by mopping up ignorance.

In this case, it is still possible to act on the basis of known unknowns in such a way as to make oneself culpable for harm. Aristotle’s concept of culpable ignorance describes the kind of ignorance behind an action of this kind. You knew there was more to be found out about the effects of substance X, yet you introduced it to the market without making the effort to do the research. In other words, you had the vantage point, but chose not to pay any attention to what the view from there told you about the location of what you still needed to explore.

Now we come to the unknown unknowns, and here metaphors based on vision (foresight, vantage points) fail us. Both the technological application of science and some areas of purer research (e.g. in synthetic chemistry or the nano-sciences) land us in an area where science no longer simply fills in the spaces. Instead, it actually produces ignorance. This non-knowledge (see Dieter Schummer’s paper here in Hyle) derives from the fact that the products of technology are often genuinely novel. Novelty is not the same as innovation: it’s perfectly possible to innovate whilst not adding anything genuinely novel to the world. Technological change may reduce the possibilities in the world, and make existing structures more rigid (a point made independently by Andrew Barry and Georgina Born). Genuine novelty, however, opens up new areas of ignorance – as if we reached a vantage point from which the landscape around us was so unfamiliar in form we couldn’t even interpret the structures we were faced with.

The weirdness of this Lovecraftian prospect is currently most associated with nanotechnology, a field in which what we don’t know we don’t know is of crucial importance. Nanoparticles, and materials composed of them, possess properties which larger particles of the same substances do not have. Which creates a problem that even the Economist recognises is a real one:

Research on animals suggests that nanoparticles can even evade some of the body’s natural defence systems and accumulate in the brain, cells, blood and nerves. Studies show there is the potential for such materials to cause pulmonary inflammation; to move from the lungs to other organs; to have surprising biological toxicity; to move from within the skin to the lymphatic system; and possibly to move across cell membranes. Moreover, these effects vary when particles are engineered into different shapes. There is currently no way of knowing how each shape will behave, except by experiment.
Britain’s Royal Society was concerned enough about all this to recommend in 2004 that nanoparticles be treated as entirely new substances. The European Commission concluded that each new material should be assessed on a “case by case basis”.

To acknowledge that new technologies essentially turn societies (and ecologies) into laboratories, and with this, to admit a certain humility in the face of the unintended consequences of creativity should be a necessary response. But what about culpability? A complete, and indeed, generally prevalent lack of knowledge about the possible consequences of one’s actions would seem like a valid defence against being held liable for them. In Parfit’s example, leaving the glass behind is itself a violation of the rights of others, given one’s knowledge of what hidden broken glass can do to bare feet. With substance X, about which we still have fairly well defined questions, we can still point to a choice on the part of the person we hold culpable for its il-effects. In the end, she decided that the opportunity costs of marketing substance X were too great for her to bear and, so she imposed risk upon others without offering them the opportunity to fully consent to take them on. But what happens when, at the time of acting, we do not even know what the possible consequences of acting could be?

Ian Hacking, in an excellent essay¹ published over 20 years ago, writes that the core of the problem of non-knowledge, and of its ethical implications, is the phenomenon of interference effects:

Interference effects occur when we combine two pieces of technology, perhaps in the same overall plant, and obtain a radically new resultant that is not merely the averaging out of the ‘pull’ from different laws of nature.

p. 148

Such effects do not have to involve technologies which imply radical interventions into the structures of living or non-living matter. Hacking cites the use of impure river water as a coolant for nuclear power stations, and the low-tech use of baffles inside cooling-tower chimneys for preventing the emission of ash. In both cases, unintended effects of the technological solutions arose despite well-grounded knowledge of the natural laws affecting individual components of the systems involved. What caused the problem was their interaction, which could not have been foreseen in laboratory modelling. In such procedures, the emphasis is always on recreating phenomena under conditions that are known. The whole point of experimentation, as Hacking points out, is to remove as much interference as possible from the phenomenon of interest. That is the power of the experimental method, and its weakness.

Interference effects, we might add, can arise not just where two or more technological components are brought together in a system where all parts have to function according to a delicate pre-planned arrangement. Here, as Hacking argues, unplanned interference effects are by definition disastrous. Interference effects can also arise wherever technological components are introduced into already-established complex ecological systems. For this to be the case, the introduction of such components has to happen in a way which threatens the larger system’s stability, a consequence which, given the complex relationships involved, will inevitably fall on the side of non-knowledge. Crucially, the threat of this happening is increased wherever the relationship between the technology and its effects is assumed to be linear, and intrinsically free of the possibility of generating interference, and also where the technology involved operates by intervening within the fundamental structures of living and/or non-living matter.

Given that industrial technology is now increasingly premised on ever more radical interventions into the structure of matter, such attitudes to technological solutions become an urgent problem. The leap from this level of intervention to the mass application of the products that result from it is one that promises us more non-knowledge that we’ve ever had.

This leaves us with at least two questions: does this therefore suggest that culpable ignorance has to be understood differently in this new context? Given that, in cases of true unknown unknowns, there are no calculable probabilities of given harmful effects occurring, how do we hold responsible those who do the economically rational thing, and try to realise the benefits of new technologies faster than they can be regulated?

¹ “Culpable Ignorance of Interference Effects”, in Douglas MacLean (ed.) Values at Risk, Totowa NJ. Rowman and Allanheld, 1986, pp. 136-154.