Recently, I discussed with physicist Santeri Laurila (PhD) about the nature of science, especially experimental science. We brainstormed some issues about science and its history that seem, prima facie, important in understanding how to investigate the estimation of futures of science. In this post, I discuss some of these issues. All the good insights are to be credited to Laurila and all the defects in thinking are my fault.
In the previous post “The Hard Problem of Future of Science”, I mentioned an argument against the possibility of knowing the future of science. That argument was based on the idea that an existing conceptual system cannot be used to describe a novel conceptual system. This seems to be a fact (perhaps even a necessary fact) about our epistemological and conceptual conditions. However, there is also an “ontological” version of the problem. In this version, it is pointed out that there are facts about the universe and methods that limit was is possible for us to know. Of course, we can only know what exists in our university and what our methods are able to reveal – this is trivial. Rather, the point is that there might exist things and mechanisms in the universe that are such that we cannot know them due to the way these things and mechanisms are structured and distributed. For example, the properties of the Higgs boson are such that its discovery required certain kind of technology. Had that technology not been produced, scientists would not have discovered the boson. In a sense, the scientists (and the whole society) were lucky enough not to quit before building the Large Hadron Collider. It is interesting to notice that even though we might have theoretical reasons to believe that something will be discovered, we might still be unable to estimate whether that something will be discovered because we do not know whether the universe is such that it allows the right kind of causal connection between us and the thing to be discovered. Again, the problem is that in order to make a prediction about our future knowledge (e.g. about a discovery), we should already know that the universe is such that it allows us to gain that knowledge (to make the discovery). And if we knew this much, we would again have so much knowledge about the properties and causal structure of the universe that the discovery would be redundant. Here the problem is not about the relationship between two conceptual systems but about the contingent causal relationship between us and the universe.
This relates to another topic we discussed. It concerns “the pull of the nature” intuition. This intuition tells us that, in the end (whatever that means), there can be only one kind of scientific description of the universe because the universe has certain structure which shapes our description of it. Nature “pulls” our scientific theories towards itself. Sooner or later, the intuition goes, we would have discovered gravitational deflection; once we were investigating that possibility, someone would have made the same observations as Eddington did because light bends around the Sun (and elsewhere). This intuition, and the related topic of contingency vs. inevitability of science (see my 2018 paper), cannot be discussed in this post but we can already note that this simple intuition seems defective in the light (no pun intended) of the considerations above. The intuition tells us that reality has a causal influence on us and, if we are attentive enough, we will follow that influence “backwards” and learn how the reality is like. However, we just noticed that sometimes we push (for theoretical reasons) towards some part of the reality without any guarantee that we are lucky enough to have the right kind of causal relationship with it. Perhaps it would be better, then, to use the metaphor of “the bend of the nature”: We learn some things about the universe and this knowledge influences the direction of our epistemological effort. But just like an asteroid going towards the Earth (the Earth bending the path of the asteroid) and passing it without a collision, we might get closer and closer to some aspect of reality without getting there; and once we run out of resources, we focus on something else. For example, what we already knew about physics pushed us to search for the Higgs boson; and even though we found it in the end, it could have been the case that we were close to the discovery without achieving it. There is no guarantee that the universe reveals itself to us, and the straightforward causal relationship between us and the universe, implied by “the pull of Nature” intuition, does not exist.
One could perhaps defend the pull of Nature intuition by pointing out that, once we have theoretical reasons to expect the discovery, we know (i) what we can know in the future and (ii) what we cannot know in the future, and it is a contingent matter whether we achieve that knowledge. For example, once we had theoretical reasons to believe in the Higgs boson, we knew that there are only two possible futures: one where we discover it and gain knowledge and one where we do not discover it and lack knowledge. There is no way, the argument continues, that we could come to know that the boson does not exist. This argument, however, is defective for the reasons cited above: If our theoretical reasons for the beliefs were strong enough to exclude the possibility that we find out that the Higgs boson does not exist, then the discovery would not add anything to our knowledge. We would already know that universe is such that the Higgs boson exist. Surely, the discovery is important for the very reason that it tells us that the theoretical background is adequate and captures (at least some aspects of) the nature of the universe; and in the cases where we do not make the discovery we thought we would, we know that there is some problem in our theoretical background (or in the auxiliary assumptions).
What does this mean to our ability to estimate the future of science? One lesson that seems to repeat itself is the difficulty of using current scientific knowledge to estimate the future. From the perspective of science, estimating the future involves estimating what kind of research and findings could be significant for improving the existing theories, models, concepts etc. These estimations should not be seen as attempts to tell how things are going to be but rather as hypothesis to be tested. For example, the prediction of the discovery of the Higgs boson should be seen as a derivation of a consequence from a theory: “Given the Standard Model, the boson should exist”. In order to use this conditional as a basis of an estimation of how science will, in fact, develop (rather than as a tool for testing the Standard Model), one needs to assume that the Standard Model (or something close to it) is correct. There can be no other reasons to suppose that the model is correct than the scientific evidence for it. This means that our best science-based estimations of the futures of science coincide with attempts to test that very science. A sense of paradox remains.
Perhaps the sense of paradox is undermined once it is noticed that science is not an activity that is distinct from other human activities. Science requires decisions and reasoning on what to do on the basis of insufficient evidence. It is impossible to know what results an experiment will produce before the experiment is performed. That, of course, is the very reason why experiments are performed. There is no area of life where we can go beyond our current knowledge when estimating the consequences of our actions. In science, those actions are performed in order to produce knowledge. Whether the knowledge we use to ground our actions is correct can be seen in the consequences of our actions. This means that every time we act, we, in a sense, test our beliefs. In science this is more evident than in other areas of life because science is explicitly associated with production of knowledge. What seemed like a paradox turns out to be a general insight in the relationship between knowledge, action and future. We cannot act without testing our beliefs. Every time we estimate the future, we test our beliefs: if the future does not turn out the way we thought it would, there was something wrong in our beliefs. In other areas of life, we just do not think our actions as testing a set of hypothesis as we do in science.
The interconnection between knowledge, future estimation and action could also provide an insight on why social sciences, economics and even fields such as epidemiology are so difficult and messy. Estimating what will happen in the future (or even next month, in the case of COVID-19) is difficult because we need relevant data to estimate what will happen but getting that data depends on what decision we make. “There are no experiments in social sciences” is a cliché. The real problem is that we cannot make those experiments before we need the knowledge they can produce. Our decisions, outcomes and knowledge are intimately interdependent both in physics and in social life and sciences.