Before I get to the subject matter, a little bit about my background. Since my PhD, I have been studying the use of quantitative evidence in the context of complexity, uncertainty and pluralism. My work has been greatly influenced by Post Normal Science, by the fossil joke and the realisation that numbers can be ambiguous, vague and uncertain. They can be interpreted in different ways, I learned that 1+1 is only occasionally equal to 2, depending on what one is counting. These considerations take numbers out of the abstract world of mathematics, and into the applications and uses of numbers in the life world. My interest is the use of numbers at the science policy interface.
My background is in development economics, which can be seen as a dialogue between economics and social concerns, and later on I was trained in ecological economics, which is a dialogue between economics and environmental concerns. The combination of these perspectives make me a sort of sustainability scholar, and I tend to pay attention to how these different social-economic-environmental perspectives are combined, and even more to the contradictions, inconsistencies and discrepancies between them. I am naturally drawn to complexity and have come to think of these three perspectives as non-equivalent and of sustainability as an issue of irreducible pluralism.
I am specifically interested in the science-policy interface, and I use the word interface to indicate the dialogue between science and policy. Numbers are one way of establishing dialogue and in order to understand how numbers are different from other ways of dialogue, I would like to reflect about some of the issues that are at stake in the relationship between science and policy.
(1.) First, facts and values. If one assumes that facts and values can be separated, one could think of the science as being in charge of facts and policy as being in charge of values. But we know from post-normal science, and other fields, that this separation cannot be done, facts are value-laden. In that case, the chore of dealing with values may be buried in numbers, and when policy receives evidence, value-choices have already been made and decisions can seem rational. The boundary between science and policy becomes more blurred and there are no separate responsibilities – rather, science becomes valuable to policy because it buries the value judgments and makes the difficult choices of values easier to handle. This is the case in welfare economics: wellbeing is studied at the aggregate level, and from the point of view of the state (the aggregate), one only needs to worry about improving welfare. Values abut how welfare is distributed and who has access to welfare pertain to a lower level of analysis, that of individuals. Decisions taken at the level of the aggregate can be seemingly value free: this is why the argument that the tide lifts all boats is appealing, by working for the common good, value choices are transferred to individuals. Numbers conceal values and are therefore a very useful means for dialogue between science and policy, which does not need to resolve the question of who takes responsibility for value choices.
(2.) Second, there is the question of knowledge and uncertainty. The contract between science and policy is based on the presupposition that science can provide knowledge about what should be done. Uncertainty questions this contract: if there is scientific uncertainty, can science know what should be done? Numbers play a specific role in this contract, by being the language of mathematics, numbers are associated with precision. As Saltelli says, one can be precise about irrelevant issues, he speaks of spurious precision. I think that this observation needs to be taken seriously: is the use of numbers driving policy-makers and scientists towards the irrelevant? If the interface between science and policy requires certainty, does this lead to a lamp posting governance of specific issues for which one can have some certainties?
I think this is the case of the energy transition debate: the technical knowledge about how to produce solar panels is getting better and better, uncertainty is being reduced. But knowledge about solar panels does not provide any knowledge about when there will be sun, and does not solve the uncertainties related to weather variability, and therefore does not make solar panels a substitute for energy sources such as coal, gas, or hydropower, which can be produced when needed and in the amounts that are needed. Numbers are then mobilised by collecting information about consumption patterns, through smart meters, but again, information does not solve the challenge that peaks in demand are independent from peaks in supply. Yet, the unquantifiable uncertainty is unspoken and when so-called transitions reach a plateau or are too slow to mitigate climate change, the problem is diagnosed as some sort of some oil-companies conspiracy, lack of political will, need for binding agreements. As a response, more is invested in renewable energies, in a lamp posting exercise. [Just to be clear, I am not arguing against transitions, I am saying that improving the technology alone won’t do and that a transition would require changes in consumption patterns to adapt to intermittent supply!]
(3.) Third, there is a difference between lifeworld problems and laboratory problems. Lifeworld problems are messy, wicked, hard to define, and the knowledge needed to define these problems is different from the knowledge needed to solve these problems. Like in the case of uncertainty, lifeworld problems may lead to a break down of the relationship between science (as the advisor) and policy. Laboratory problems, or technical problems, are puzzles for which there is only one possible solution, and if the solution is not known, this is a temporary problem that can be solved with more time, more research, more tests. Laboratory problems can be controlled, and the knowledge gained from them can be transferred to policy makers to control the economy, society and the environment. Numbers frame problems as laboratory problems, and carry with them the promise of control, planning, monitoring, setting goals and measuring progress.
In my work in informal settlements, I encountered a case in which an NGO had installed solar panels on the roof tops of some of the shacks, and could not understand why people who owned a solar panel did not change their use of paraffin and gas. We found out that solar panels provided very little electricity, which was use for street lighting – a very important function, which made the settlement safer at night. But there was not enough electricity for cooking, heating, boiling water, and the consumption of paraffin remained unchanged, along with the number of shacks catching fire, the health problems derived from breathing paraffin fumes. This is a problem of complexity: when everything else is not equal, intervention leads to adaptation, the system changes its behaviour. It doesn’t mean that interventions are ineffective, but the effects are not those that were planned.
In the event organised for Jerry Ravetz 90th birthday, Jerry talked of teddy bear numbers. That is, numbers are comforting. At the interface between science and policy, comfort comes from relief from values, certainty and control. An alternative to teddy bear numbers may be a more mature use of numbers, one that takes into account the critiques and limitations. I have called this alternative the non-positivist use of numbers. Positivism is using numbers at the science-policy interface because they describe reality, and provide a value-free, certain and necessarily reductionist account of reality. The non-positivist use of numbers is a re-conceptualisation of numbers as heuristic tools. I will come back to the idea of heuristics, but first I want to give two examples of non-positivist use of numbers.
One is the NUSAP system, in which numbers are used to convey uncertainty. Jerry Ravetz suggests a notation scheme for the numeral, the N of NUSAP, that adjusts the use of significant digits to different orders of magnitude depending on the level of uncertainty (CARBOM – close to, around, roughly, ball-park of, order of magnitude). Silvio Funtowicz and Jerry Ravetz explained that there is a trade-off between spread and assessment, and that trade-off requires an understanding of statistics before statistical tools can be applied, it may be a remedy against p-hacking. Jeroen van der Sluijs has done a lot of work in developing the Pedigree part, and giving voice to tacit knowledge about the quality of data, models, theories and scientific consensus. NUSAP makes the vagueness and ambiguity of numbers explicit and communicates it. This is very different from numbers as conveying certainty.
The second example is quantitative story-telling, which uses complexity to provide multiple non-equivalent quantitative descriptions of the same problem. Numbers, just like words, can be combined in different ways and used to tell different stories. Different quantifications, however, cannot always be aggregated: one cannot sum kilowatts to temperature. This approach highlights the irreducible contradictions within science, and systematically applies the wave-particle duality problem to quantification. The juxtaposition of multiple perspectives through quantification brings attention to problem framing, to the analytical steps that precede quantification, where the values are buried and concealed.
These two approaches create literacy about issues of uncertainty and complexity, without dismissing numbers. But literacy is not enough, the use of numbers requires also craftmanship. Numbers become something else, they become heuristics. Heuristics are understood as opposed to blueprint procedures for the use of quantitative information. The example is that of a baseball player, who does not need to compute the trajectory of the ball based on velocity and angle, but adjusts his position while observing the trajectory of the ball, and knows from experience and not from calculation how to adjust. Heuristics should not be confused with improvisation but rather with the ability to adapt to the irreducible singularities of the object of study. The heuristic use of numbers also means that one needs not fall into dichotomies between certainty and uncertainty, that if Truth cannot be known than everything is relative. One cannot predict the exact height of a human being, but we can know with certainty that it is biologically impossible for a human being to be 4 meters tall (see Taleb’s Black Swan for more examples). One can work with orders of magnitude. The challenge is that there is an asymmetry, as Popper pointed out, between what can be know with certainty, which is very little, and the much easier task of saying what cannot be known.
My last point is that if numbers have the function of Linus’ blanket, What happens if teddy bear numbers are taken away? What happens if one talks about uncertainty and complexity, if one says that there is no optimal choice, that all options have trade-offs and decisions come down to choosing winners and losers? It’s either increasing agricultural productivity or sustainable agriculture. It’s either 100% intermittent renewable energies or a high-tech smart economy that is run by servers and supercomputers that are powered 24/7.
Then the dialogue between science and policy may break down. We tried delivering some of these messages in the MAGIC project, and the messages were not appreciated. I think that moving away from teddy bear numbers requires not only more responsible quantification, but also attention to people. The dialogue is not between science and policy in the abstract, it’s between scientists and policy-makers. I go back here to the concept of heuristics: heuristics are created through experience. So what is important is not just the type of knowledge that is used – uncertainty and complexity instead of reductionism and positivism – but also how knowledge is acquired. One thing is to acquire knowledge in the classroom, and a different thing is the tacit knowledge, the craftmanship that is acquired through practice. One thing is to produce lists of principles and methods for quality assurance, and a different thing is to engage in a continued dialogue with policy-makers, to pay attention to the lifeworld of policymakers and to work towards a different use of numbers anchored in engagement and not from an outside perspective.
0 Comments