I confess the intractability of "systems" never seemed especially mysterious to me. Maybe this is an advantage of being a bit stupid. I was never smart enough that I should expect to understand things, so things not being understandable never seemed especially mysterious.
From this idiot's view: the obvious thing missing in ethics, and also in most of social science, is that last bit about "being able to try a lot of different options and evaluate their worthiness". But this isn't anything intrinsically mysterious about the problems, it's just that the problems are Too Big to practically experiment and iterate on.
If you could grow hundreds of human economies in a controlled laboratory environment over the course of a few weeks, we might understand economics as well as we do fruit flies--which is to say, highly imperfectly, but better than we do actual economics. If you could do the same thing with ethical principles--maybe engineer some virtue-knock-out spiritual leaders or something--then maybe we could find out with some certainty which ethical principles are most conducive to human thriving.
But since we are inside the system, living at the timescale of the system, we can only ever iterate on and optimize small subproblems. Try to do anything else, and you are necessarily generalizing far beyond your tiny slice of space-time data. Of course it doesn't work!
I think that sort of would work for Econ, but not for Ethics. The trouble with Ethics is that, in a way, everything is up for grabs. I don't see how experiments would tells us anything about the goals to be "optimized", or about allowable actions and tools.
For ethics it depends on people agreeing on a particular measure of, ah, divinity? Godliness? The Good-itude? Which is certainly not in the cards for us! But I do think there is some degree of consensus that a successful ethical theory should lead to the "thriving" of societies that "adopt" it.
Of course "thriving" is just as subjective as "ethics", technically speaking; but there really is general consensus on what it means for a baby to "thrive", at least in a "we know it when we see it" way. If we were in a position to grow millions of societies in a petri dish we might manage something similar.
I agree that in the biological and biomedical sciences things are messier, but maybe not only because we lack good models. Even if we had perfect knowledge of the underlying biochemistry/biophysics at the molecular level, there might be fundamental limits of the systems approach itself. Anderson's "More is Different" paper from 1972 comes to mind. While reductionism (breaking things down to fundamental laws) is powerful, the reverse is not trivial. Knowing the parts and their interactions doesn't automatically yield understanding of the whole, because at each hierarchical level of organization, new laws, concepts, and phenomena emerge that are irreducible in practice.
YET, at the same time, I find it really interesting that in biology evolution-driven optimality principles are deeply reconciled with systems-level thinking, emerging bottom-up from complex, feedback-rich processes involving constraints, historical contingency, and rugged fitness landscapes rather than imposed top-down.
Some of us think that this makes biology a domain where the two perspectives complement each other: systems views illuminate the dynamics shaping optimization, while optimality provides strong predictive power under intense selection.
I think that last bit of “can try a lot of different options and evaluate their worthiness” is hard in biomedical science. Sure we have all the various model organisms but it’s not always clear what we learned experimentally by some simple perturbations generalize to humans (otherwise many more drugs should come out of phase2/3 clinical trials). If we can’t really be sure the experiments are even valid, we are not going to be confident with our conceptual models - for example, for many diseases of the immune and nervous system we just don’t fully know how things go wrong over time to fix it. If we are not confident with conceptual models then I suppose it’s hard to decide the relative worthiness of the outcomes of various experiments to advance or modify such models. I can’t quite pin it down, but it’s just messier than some problems in physics where a better model should fit data with less residual or in material science where you try to synthesize materials with some better measurable quality, or in maths and computer sciences where you can push the bound on some conjecture. In contrast, the solution space for many problems in biology are not even well defined (there are cases where you can try, say to optimize antibody properties or discover new antibiotics, but these are really engineering problems in the first place with some ad hoc science after the optimizations are already done)
I confess the intractability of "systems" never seemed especially mysterious to me. Maybe this is an advantage of being a bit stupid. I was never smart enough that I should expect to understand things, so things not being understandable never seemed especially mysterious.
From this idiot's view: the obvious thing missing in ethics, and also in most of social science, is that last bit about "being able to try a lot of different options and evaluate their worthiness". But this isn't anything intrinsically mysterious about the problems, it's just that the problems are Too Big to practically experiment and iterate on.
If you could grow hundreds of human economies in a controlled laboratory environment over the course of a few weeks, we might understand economics as well as we do fruit flies--which is to say, highly imperfectly, but better than we do actual economics. If you could do the same thing with ethical principles--maybe engineer some virtue-knock-out spiritual leaders or something--then maybe we could find out with some certainty which ethical principles are most conducive to human thriving.
But since we are inside the system, living at the timescale of the system, we can only ever iterate on and optimize small subproblems. Try to do anything else, and you are necessarily generalizing far beyond your tiny slice of space-time data. Of course it doesn't work!
I think that sort of would work for Econ, but not for Ethics. The trouble with Ethics is that, in a way, everything is up for grabs. I don't see how experiments would tells us anything about the goals to be "optimized", or about allowable actions and tools.
For ethics it depends on people agreeing on a particular measure of, ah, divinity? Godliness? The Good-itude? Which is certainly not in the cards for us! But I do think there is some degree of consensus that a successful ethical theory should lead to the "thriving" of societies that "adopt" it.
Of course "thriving" is just as subjective as "ethics", technically speaking; but there really is general consensus on what it means for a baby to "thrive", at least in a "we know it when we see it" way. If we were in a position to grow millions of societies in a petri dish we might manage something similar.
By the way, the same comment would apply, maybe to a lesser degree, for the biomedical examples mentioned by others here.
I agree that in the biological and biomedical sciences things are messier, but maybe not only because we lack good models. Even if we had perfect knowledge of the underlying biochemistry/biophysics at the molecular level, there might be fundamental limits of the systems approach itself. Anderson's "More is Different" paper from 1972 comes to mind. While reductionism (breaking things down to fundamental laws) is powerful, the reverse is not trivial. Knowing the parts and their interactions doesn't automatically yield understanding of the whole, because at each hierarchical level of organization, new laws, concepts, and phenomena emerge that are irreducible in practice.
YET, at the same time, I find it really interesting that in biology evolution-driven optimality principles are deeply reconciled with systems-level thinking, emerging bottom-up from complex, feedback-rich processes involving constraints, historical contingency, and rugged fitness landscapes rather than imposed top-down.
Some of us think that this makes biology a domain where the two perspectives complement each other: systems views illuminate the dynamics shaping optimization, while optimality provides strong predictive power under intense selection.
I think that last bit of “can try a lot of different options and evaluate their worthiness” is hard in biomedical science. Sure we have all the various model organisms but it’s not always clear what we learned experimentally by some simple perturbations generalize to humans (otherwise many more drugs should come out of phase2/3 clinical trials). If we can’t really be sure the experiments are even valid, we are not going to be confident with our conceptual models - for example, for many diseases of the immune and nervous system we just don’t fully know how things go wrong over time to fix it. If we are not confident with conceptual models then I suppose it’s hard to decide the relative worthiness of the outcomes of various experiments to advance or modify such models. I can’t quite pin it down, but it’s just messier than some problems in physics where a better model should fit data with less residual or in material science where you try to synthesize materials with some better measurable quality, or in maths and computer sciences where you can push the bound on some conjecture. In contrast, the solution space for many problems in biology are not even well defined (there are cases where you can try, say to optimize antibody properties or discover new antibiotics, but these are really engineering problems in the first place with some ad hoc science after the optimizations are already done)