[kitchi]

As another dude with a doctorate in Physics, I have to disagree with you (at least somewhat).

> From this point of view as our ability to connect experiment with outcome has increased our ability to actually say what it is we are even talking about outside of the purely instrumental has decreased since the 19th Century. Back then we though we knew that there were atoms or electrons or whatever. Light waves or photons. Now, I would argue very strenuously, we genuinely have no understanding at all of what those things are outside of a set of purely instrumental definitions which leave a lot to be desired.

I disagree with this entirely. The existence of QFT, and our knowledge of the inconsistency between say GR and the quantum realm does not negate the idea of photons and electrons as real, measurable quantities. The fact that we have GR does not negate the fact that we still use Newtonian gravity in regimes where it is sufficiently accurate.

All the new knowledge we have learned still is (and absolutely must be) consistent with our old knowledge that has been proven correct in the regimes that they were proven correct.

This is effectively what Asimov is saying (as I understand anyway) - the knowledge that the Earth is a sphere does not invalidate the assumption that the Earth is flat approximately and locally.

I would also argue that the only things we can "know" are what you call the instrumental definitions. We only know what we measure. The rest is interpretation, and self-consistent understanding.

String theory can tell me that we have several dimensions etc but until we have a way to measure and check it remains a conceptual framework to make predictions, rather than a description of how things really are.

GR is much closer to a description. It told us about the precession of mercury, it told us to account for time dilation so we can use GPS satellites. It also predicted black holes, which were conceptually consistent but it's only been in the last ~ 5 years that we have the closest thing yet to experimental verification with the Event Horizon Telescope and gravitational wave measurements. If another theory comes along and explains all of GR with a different explanation for black holes, we will need still more accurate measurements to discriminate between the two theories. Knowledge is only as accurate as we can measure.

[glenstein]

I agree much more with your approach.

The way I've heard it best described is these notions of electrons and photons etc will still be retained as a special case of whatever theory supersedes them, which is critical, because that's at the heart of the "relativity of wrong" argument.

Some take the prospect of a future revision of theories to mean our present state of knowledge is no different than any prior failed theory, which I think is an urgently, catastrophically wrong, catastrophically confused way to regard the history of scientific knowledge.

[tycho-newman]

QM is so platonic though. Reality consists, ultimately, of forms that can only be described mathematically. It just happens that the math returns (what we see) as a probability distribution/wave function.

I’ve never quite understood what a quantum theory of gravity would be though. QM involves the observer but gravity engages spacetime - the place where you are observing things. A quantum field theory of gravity seems like a contradiction in terms to me. Unless quantum gravity is really about the Big Bang?

[nathan_compton]

The observer/observed thing is present in every part of physics and isn't really about observers but about where we choose to draw the line between one physical system and another (with the other usually containing ourselves).

Think of it this way: classically or quantum mechanically, when we pick out a physical system to talk about we are isolating the terms for that system in the universal lagrangian and assuming that in the time of interest that the terms for our physical system couple weakly to the rest of the universe (which happens to contain us).

In principal nothing really weird is going on here and in classical mechanics the idea is totally trivial as far as it goes. On short time scales with appropriately sized actions the deviation from the isolated system and the real system (which is weakly coupled to the world) can be demonstrated to be small as long as the coupling is small.

In quantum mechanics two things complicate this situation. The first is that quantum mechanical systems sort of defy separation into distinct subsets except in special situations. Classically there is a strong sense in which we can point to two different parts of a system and call them separate, but quantum mechanically we really only know how to time evolve _the whole system_ and from a mathematical point of view its the actual object of interest. This is what we are getting at when we talk about entanglement: the two spin 1/2 particles flying away from one another in Bell style experiments are not separate things in the QM description: there is just one wave function.

But in practice I don't think there is any real reason we can't quantize gravity. I'm not an expert but loop quantum gravity seems like a reasonable approach and its very straightforward and its base: just find an acceptable description of geometry and then apply the ordinary quantum mechanical tricks we use to quantize it.

Also, you don't observe things in spacetime. Observations are always purely local. You just infer the existence of spacetime from local observations which are conveniently organized by putting them on a curved 4d Lorentzian signature manifold.

[nathan_compton]

I don't think you get my point because I don't think of anything you are saying as having anything to do with what I was saying.

If you have a purely instrumentalist view of reality, where, as I said, your so-called knowledge is actually just a model of an unknown thing which you employ to predict the measurements you read out against a ruler or on a meter or something, then yes, we've made progress exactly of the kind you describe.

But I was trying to make a point about epistemology and ontology. Physics has actually been pretty catastrophic for ontology. I don't think its wrong to say that from the point of view of physics we simply do not know what anything actually is.

> I would also argue that the only things we can "know" are what you call the instrumental definitions. We only know what we measure. The rest is interpretation, and self-consistent understanding.

Yes. But this is a fairly radical position historically and philosophically. Most people would say that there is more to existence than measurement and I while I share your instrumentalist sympathies, like most physicists, I don't see the philosophical case that we can have a consistent worldview if we denounce all knowledge not related to measurement as a total non-starter.

Think about what instrumentalism really means. When you utter the sentence the earth is an oblate spheroid, you are actually making an incredibly complicated set of statements about the outcomes of experiments. If we take the instrumentalist view the measurement doesn't actually tell us the earth is an oblate spheroid - it just tells us that if we make a series of measurements then they come out in such a way as to be concordant with a model of the earth as an oblate spheroid. Are you really prepared to give up the idea that the earth is a thing you can know about?

I actually rather think physics strongly encourages us to adopt the instrumentalist view, primarily because it seems so clear that physics has a local character. In GR there simply is no state of affairs whatever about what is happening "right now" except at the point in spacetime where you make a measurement. Really think about what that means. If we are standing at the north pole and make a measurement of some kind, how can it pertain to the earth as a distributed object in space when we know GR says there is no state of affairs pertaining to that object at the moment of the measurement?

GR tells us all about what the outcome of various measurements will be, but it also calls into question what precisely those measurements might mean. The instrumentalist is committed to the idea that the only thing we can talk about is the results of measurements. What the measurements operate on is just not something we can know. I think that's weird. Physicists often conflate their mathematical models with reality and that lets them think an instrumentalist view is sufficient: the measurements coming out such and such a way is taken as evidence that the universe is filled with objects consistent with the model. But that association is non-trivial in modern physics.

[glenstein]

It must be the case that our measurements are interfacing with something ontologically real, so I can't agree with the minimizing tone that treats respect for them them as implicitly necessitating a total exclusion of ontology. Far from breaking the link between measurement and reality, examples like GR’s localism or QM’s treatment of particles show that measurement is what pushes us into new and unfamiliar ontological commitments.

To then suggest that Asimov's essay depends on such a radical version of instrumentalism feels unnecessary, given that his argument attaches just as well, I would say better, to a view that makes space for ontology but still underwrites the history of progress he was describing. His essay seems less a failure to follow deeper philosophical detours than a decision to keep his argument tied to that central point.

You also seem to be implying a kind of ontological "regression" in the history of knowledge which I think comes from projecting specific ontological attitudes retroactively onto the past when those distinctions weren’t even part of the vocabulary.

A big problem with Steven Jay Gould’s self-posturing as a savior of Darwinian theory was that he manufactured a crisis of his own invention about "gradualism" and then claimed to resolve it. Darwin never assumed a fixed speed of evolution, so the supposed crisis dissolves once you stop reading one into the history. A similar attitude helps in reading the history of science: early measurements and models need not be seen as knowledge of ontology that we somehow "lost" but as data accompanied by overconfident declarations that can be separated out.

If you correct for ontology in this way, treating the instruments as interfacing with whatever is ontologically real, then you see steady progress toward more accurate knowledge. Far from being subtly wrong this reinforces the core of Asimov’s point that successive models are increasingly less wrong because they are tethered to reality (e.g flat to globe to spheroid to GR’s spacetime and beyond). It's that same arc of progress carrying us through those examples into more exotic ontologies. And the sense of crisis comes not from instrumentalism's commitments but from identifying progress with 'intuitively familiar' as if outgrowing our conceptual inheritance were failure rather than discovery.

[glenstein]

Long reply already but one more thing. I would go so far as to say that all-out instrumentalism is a way of reading what Kuhn advocated (and perhaps actually what Kuhn believed depending on who you ask), and it was key to him to get away from notions of progress that rely on ontological commitments to external reality. And so a lot of the pushback was to rescue "progress" by insisting on ontological commitments that make the notion of progress make sense. In that light I can't see a way of reading Asimov advocating progress that would commit him to that kind of instrumentalism.

[the_af]

To be honest, I think you're straying from science into philosophy.

"The nature of being", "ontology", "it calls into question what precisely those measurements might mean".

Science is not about the search for meaning, it's about describing and understanding the physical and natural world. Meaning is up to you, and me, and everyone else.

Asimov was not talking about meaning.

It's easy to see that he wasn't, because nobody would make the categorical statement that there's unequivocal progress in the search for meaning, because meaning is such a cultural (and personal) thing. Meaning is not scientific, and is highly subjective. I wouldn't even know how to measure meaning. Self-reporting? Chemically measured happiness? What.

Had the English professor complained "we think we know so much, yet we're so ignorant about what it all means! People are so unfulfilled, live such empty lives! Wake up, sheeple!" I bet Asimov would have... -- well, he probably would have had humorous words for that too, but his argument would have been different.

[nathan_compton]

I do not think of science as distinct from philosophy. But I'm not talking about the sort of meaning you are talking about. I'm talking about the fundamental basic question of what the physical world consists of and how we relate to it.

When I say that most people think there is more to the world than measurement, I don't mean art or love or whatever. I mean that most people think that the world is made up of things that exist whether we measure them or not, things which have their own nature which we somehow can understand. The purely instrumental thinker says we can't understand those things, we only have measurements and mathematical expressions that relate measurements.

[kitchi]

I'm not a native speaker so I had to look up what "ontology" means specifically, and what I got is : "The branch of metaphysics that deals with the nature of being. " (among other things)

I agree in that sense that physics does not do well here. I actually think physics _cannot_ do well here. Physics (and the "hard" sciences more generally) are good at describing what something is, and how it interacts with things around it. We "know" by assembling individual pieces of information that form a consistent model, then declare that model to be true. When a piece of information outside of that model arises, we then have to call that model in question.

I do not "know" the Earth is spherical from direct experience. I do "know" it from other indirect means - reading, measurements, images from space and so on. I do not also think that Asimov is saying anything about how we know what we know - he just talks about how "science is wrong" is not a true statement. Science is always approximately correct, but how approximate is the question.

So perhaps I'm missing the point entirely here, but I don't understand your distinction of "instrumentalist knowledge" vs other kinds of knowledge. If you say physics cannot explain my knowledge that I enjoy watching the sunrise - then absolutely yes. That is not it's realm. In the same way that physics cannot explain the history of medieval China to me. A common issue among physicists is to assume that this is the only way to view the world, and I disagree with that. There are many systems of knowledge, and each is good at certain things. Rejecting one as the "global" system misses the richness of other kinds of knowledge building.