Science as Levy flights
We know from data, models and eco-evolutionary arguments that animals that moves in complex environments looking for resources (i.e. spider monkeys looking for food) do it in what is called a Lévy flight, which is a random walk type in which the step-lengths have a Lévy distribution (just the technical name), a probability distribution that is heavy-tailed (it present infrequent but extreme events). This means that the animal (it most likely be true for any kind of agent looking for resources) perform a lot of small steps in a local kind of search and then, from time to time, they travel a great distance (order of magnitude of the typical step size) to explore a whole new region of the environment.
So this kind of exploring pattern that emerge in nature, could be interpreted as a good balance between the local search in what we know and big jumps to the unknown (criticality).
It sounds like this should be how we scientist should explore the space of knowledge and ideas. We need to have a good balance between a local exploration of what we already know (we could call this the standard science) in which we get more details, increase precision and so on; with realy out of the box ideas (revolutionary science). The tricky part is to understand what does this balance means and how to implement it.
Take for example the case discussed in Nature and ask yourself, is this good, bad or an hoax? Keep in mind that the story in the Nature piece is intended to show that not only you and me, but even real experts in this field could have a hard time deciding if the claims are correct, not or just an hoax.
Take a deep breath, and give the following sentence a go. “We demonstrate that the lorentzian signature of the space-time metric (+ + + −) is not fixed at the Planck scale and shows ‘quantum fluctuation’ between the lorentzian and euclidean (+ + + ±) forms until the 0 scale where it becomes euclidean (+ + + +).” Confused? Don’t worry, you’re in good company. Physicists around the world have been unable to agree on whether the PhD thesis this line comes from is good, bad or a hoax.
The Bogdanof story point to a problem with some fields of modern physics that might have become too much speculative, some sort of “no data Physics”, where the work seems to be entirely imaginative reaching the extreme of thinking that no data contrasting is needed (see https://researchers.one/articles/19.08.00002). If we only make levy flights (big jumps in the form of scientific speculation) and no local search, we will never generated enough rigorous and detailed knowledge to for example build technology; or we could end up with a zoo of plausible but not tested incompatible hypothesis about how the universe work, not good. We most remember that Science is not only a set of ideas, or knowledge:
Science is on the one hand a philosophical approach to life, which allows us to put some order into the world we experience. That allows us to recognize what is likely to be known systematically, verifiable with the observations of the world, from what is not. That endeavor that seeks to identify what is an invariant objective of what is not. In this way this ting Science is a sensor, an instrument that allows us to explore the objective part of our reality https://lopezoliverx.medium.com/is-not-the-same-ting-49d2498574b4)
Of course if we close ourselves to new radical ideas, we might pass over for example important developments as Einstein relativity that took decades to be proved with rigorous data (specifically gravitational waves took 100 years). Then we may end up knowing all about a very narrow region of the knowledge space.
Be aware that most likely, for every Einstein who has proposed a theory that turned out to be true and which data verification came many years later, there must have been hundreds of proposals that never amounted to anything. We most then avoid to fall into get “Lost in Math” (Hossenfelder) as some stringists claim, and maintain the Newtonian tradition of contrast ideas with the data (https://researchers.one/articles/19.08.00002) as the key step of scientific knowledge construction, there is a reason why its call natural philosophy instead of just philosophy.
For me it is clear that a good balance is required. In that sense, it seems healthy that there would be different types of articles in scientific journals: hypotheses, preliminary results and final research reports.
For a hypothesis (not a mere speculation) I believe that as long as they have a solid conceptual basis (“scientifically sound”), they do not break any first principles, mathematics is rigorous and they show clear mechanisms of contrast with the data, even though those data still not available, its publication is valuable.
Of course, readers should take these articles for what they are, a hypothesis. They should not be used as evidence or justify, for example, public policies. In other words, its “hypothesis” label should not be ignored, confused or changed without having passed to one of the following two stages: preliminary results or final reports. And each state must have a very clear label that makes it possible to understand the level of rigor and verification that it implies. I think that this is not at all clear, at least for the non-expert reader, in the current publication system.
As with is being used for food labeling, papers could have labels that denote very clearly the state in which what is presented is. Perhaps more than three labels are required and that should be part of the work of the editors and reviewers. Just accepting the paper is not enough, proper labeling could be crucial. Additionally, as it is done for example in the citizen science system of Naturlista (https://www.naturalista.mx/) where there is a labeling system weighted on experience (all the way up until reaching the expert curators); even peer-reviewed articles could have a second labeling based on this kind of opinion from the readers.
In this transformation Libre Science could be key, see: https://lopezoliverx.medium.com/the-future-is-now-for-libre-science-95bb40dd85c
And finally funding should reflect this philosophical approach about science making. Consider a simple stocks and flows model as in the figure below. Three funding scenarios: (I) Risky but potential huge payoffs; (III) “safe play” scenario and (II) intermediate risk. For I and II -> funding is assigned by a lottery policy; for III all gets funding (1/n strategy, see video below). A simple thumb rule would be that I gets 20% of total funding stock, II 60% and III 20% (simple barbell type strategy, % could be calculated more sophisticated)
I think that this idea of science making as Levy flight would lead to an Antifragile squeme, as discussed by Taleb here:
