A blog post arguing that we shouldn't give bonus points to off-beat and bold theories just because they are off-beat and bold.
https://petterhol.me/2023/08/09/the-twilight-of-fantastical-science/
@pholme very cool. I agree with being very careful to not evoke many levels in explanation, unless one needs to. But I do think that a complex system is one where you need more than one level to explain a phenomenon---which becomes pragmatically, epistemologically irreducible to a single mode of explanation (Pattee's control hierarchies). Wave-particle duality is an early example. But the pandemic is a great recent example, whereby a biochemical event (zoonotic transfer) led to very long multi-level control/causal chains from immune response and organ failure, to human transport networks, social and political change.
@lmrocha No worries! This is not about the bulk of complexity or systems theories. It is not about studies mapping out causal structures across hierarchical levels. (See footnote 4.)
@lmrocha I added a paragraph to clarify this.
Note that there is nothing wrong with mapping out complicated networks of causal effects. [...]
@pholme I generally agree, but the best models of COVID spread (e.g. to predict casualties per location) do include multiple levels in what one may call a computational theory of actionable conked systems models (cf Vespignani). In such multiscale models (or theory) you have simultaneously variables to model molecular transmission, human contact, transportation networks and psychological behavior like mask adherence according to political preference data (per US state, for instance.) True, the tongue-in-cheek link to Godel is not there, but certainly approximations to deal with computational complexity are very much part of such computational models and implicit (complexity) theory :)
@lmrocha I think you agree with me completely :) Vespignani style models are at most three levels microbes-people-society (usually just people-society), and the couplings are separated, obvious, and mechanistically validated.
They go: mutation -> more contagious -> increased spread -> governmental intervention, not mutation -> larger state-space of governmental decision process cellular automata -> intervention (I'm just hallucinating, but to give you an idea of off-beat and bold)
@pholme yes, we are mostly or entirely in agreement, I think :) Though I never spoke against mechanistic validation. I'm just saying that a pandemic, per the best models we have of them, are complex systems in the way we have discussed: they include several irreducible levels as constituents. A reduction to molecular or to social does not do as good a predictive job as modeling several levels. (BTW, I disagree there are only 3 levels in those models, with the transportation or technological layers involved, but that is a minor disagreement :) to be clear, I think complex systems are and should be amenable to (multi-layered) mechanistic modeling---emergence in this sense (necessity of different models for different levels of experience) is not magic. As a parallel, consider that complementarity in wave-particle duality did not preclude the mechanistic prediction that either explanation affords; indeed, that was the whole point.
