[lisper]

It's true that science is not complete, and cannot be. We know, for example, that we cannot solve the halting problem, we cannot predict chaotic systems, etc. But that's the wrong thing to focus on. There are a lot of things we can predict, and the scientific method produces better predictive theories than any other known method.

> disinclined and demotivated to look outside the box of its current models

OK, but that is not the scientific method's fault. If you want to look outside the box, nothing in the scientific method says you can't. An indeed the biggest breakthroughs have come when people have thought outside the box.

The problem is that it's often hard to distinguish between brilliance and crackpottery. But again, that's not a shortcoming of the scientific method, it's just the Way The World Is. Some problems are hard.

[dekhn]

The halting problem isn't scientific, though. It's entirely mathematical (mathematics and science are typically treated as distinct domains, see https://www.maths.ed.ac.uk/~v1ranick/papers/wigner.pdf for some discussion). We do not know if there is a scientific way to make a machine which exceeds the ability of a Turing machine, see for example this paragraph from the wikipedia page on the halting problem:

"It is an open question whether there can be actual deterministic physical processes that, in the long run, elude simulation by a Turing machine, and in particular whether any such hypothetical process could usefully be harnessed in the form of a calculating machine (a hypercomputer) that could solve the halting problem for a Turing machine amongst other things. It is also an open question whether any such unknown physical processes are involved in the working of the human brain, and whether humans can solve the halting problem"

One of the most important things to recognize about science is that we rarely, if ever, work with absolutely well-determined systems with analytically solvable equiations. INstead, we work almost untirely with underdetermined systems with only approximate methods, and while somewhat unsatisfying, those methods are almost always a more efficient way to make falsifiable hypotheses and run experiments. I don't think anybody ever truly makes a falsifiable hypothesis- in the sense of Descartes' great deceiver, we can't truly know for certain what the underlying state of the system was.

[lisper]

> The halting problem isn't scientific, though. It's entirely mathematical

No, it isn't. The halting problem arises out of a mathematical model of a physical system. We don't know for certain that it's impossible to build an oracle for the halting problem, just as we don't know for certain that it's impossible to do an end-run around the Second Law. But the evidence in both cases is (IMHO) equally compelling.

[dekhn]

Everything written about halting machines presumes a mathematical system, not a physical system. It makes statements about model systems, some of which have physical counterparts. THe whole point of a turing machine is that it's abstract, not made of tape or transistors or anything else.

The halting problem is not really widely discussed in physics journals. They care much more about the physical limits of actual computing. If you have good examples of physics people talking about the halting problem in a non-theoretical and distant way, I'm happy to see it. But from waht I can tell, physicists are not concerned with uncomputable functions.

[lisper]

> The halting problem is not really widely discussed in physics journals.

So? Golf clubs aren't widely discussed in physics journals either but they are physical systems nonetheless. You can't draw valid conclusions about what is not a physical system based on what is absent from physics journals.

[strogonoff]

> the scientific method produces better predictive theories than any other known method

Better theories in what sense? (If possible, in terms that are genuinely extrinsic to scientific method itself.)

> If you want to look outside the box, nothing in the scientific method says you can't

Nothing in the scientific method says you can’t, indeed. However, some people tend to misinterpret it (possibly due to a suppressed religious impulse finding its way) as revealing objective truth through its models, as opposed to what it does: offer predictions as to what we would observe if we do X. For many of those (very smart) people, imagining radically new models that focus on different aspects (e.g., that conspicuous we above, or something else) and sideline other aspects (e.g., the various entities described by current models) is taboo.

[lisper]

> Better theories in what sense?

In their ability to make accurate predictions.

> However, some people tend to misinterpret it (possibly due to a suppressed religious impulse finding its way) as revealing objective truth through its models

But science does reveal objective truth, in the sense that it reveals truths (or at least very good approximations to truths) that are independent of what anyone's opinions are. What it does not do is reveal metaphysical truth, but that's not the same thing. But even then, it does put constraints on what metaphysical truth could be. For example, unless quantum mechanics is wrong (which is extremely unlikely) then it is not possible for metaphysical truth to be classical.

[strogonoff]

> In their ability to make accurate predictions.

That is a bit too close to defining them as better within the framework of scientific method.

For example, though I suspect you won’t like this line of question, are we by chance able to make increasingly accurate predictions about something increasingly irrelevant or not beneficial to ourselves?

Edit: I would be the first to say that the answer to that question is probably negative, but that is just to illustrate, maybe this would push you to define “better” better.

> But science does reveal objective truth, in the sense that it reveals truths (or at least very good approximations to truths) that are independent of what anyone's opinions are

Models are metaphors to aid our minds in coming up with more predictions to test. If a model was able to predict N outcomes that does not make it correct, unless you can guarantee that there will not be a future outcome that makes that model incorrect, which you cannot as that notion would presume you have come up with a provably correct and complete model in finite time.

> But even then, it does put constraints on what metaphysical truth could be. For example, unless quantum mechanics is wrong (which is extremely unlikely) then it is not possible for metaphysical truth to be classical.

I cannot object to that, except the part where you claim that quantum mechanics being wrong is extremely unlikely. I will stand by my initial assumptions and claim that it is not just extremely likely but a near certainty that quantum mechanics is wrong—just because it is foolish to assume that any of today’s models is finally correct and true. It may be useful in meantime, though.

[lisper]

> That is a bit too close to defining them as better within the framework of scientific method.

No, that is simply pointing out the reason that science is a thing at all.

> are we by chance able to make increasingly accurate predictions about something increasingly irrelevant or not beneficial to ourselves?

Probably, though putting effort into this would obviously not be the wisest choice.

> except the part where you claim that quantum mechanics being wrong is extremely unlikely

You need to read "The Relativity of Wrong" by Isaac Asimov.

[strogonoff]

I will read it some time.

I still do not know what makes a theory “better” if it will never be provably correct, is most likely drastically wrong, and comparatively brings us little value. A theory that would instead focus on ourselves, whatever that might look like, seems like potentially a much “better” option, but because the path is poorly trodden the scientific community would defend its own dignity by laughing at those who venture there. Anyway, I’m not awake enough to argue well at this point.

[lisper]

> brings us little value

You don't see value in the ability to make accurate predictions about the future?

[strogonoff]

> We know, for example, that we cannot solve the halting problem, we cannot predict chaotic systems, etc.

This violates the second axiom I assume: in your examples, we know where the incompleteness lies. If you disagree that not knowing that would necessarily hold for modeling the system that we are part of, then perhaps we won’t be on the same page to argue productively.

[lisper]

Sorry, that didn't parse. What do you mean by "the second axiom"? What axioms are you talking about?

[strogonoff]

> My pet theory is that success of predictions of modern science helped us achieve a local maximum which, while useful, also makes it difficult to evolve towards a higher (globally) maximum.

> The axioms are:

> 1. Some models are useful, but all models are necessarily incorrect, incomplete, and/or unfalsifiable. (This can be restated in an unsettling, for a natural science enthusiast, manner as “there is always the unexplained”.)

> 2. We cannot know where the incorrectness/incompleteness lies.

(It is an axiom because I strongly suspect I will maintain this position but am not willing to spend time defending it. Maybe I should call it “assumption”.)

[lisper]

Ah. Sorry, I'm responding to a dozen different threads and I'm having trouble keeping all the context in my head.

I think your axiom #2 is almost certainly correct. But just because we can't know where all of the uncertainty lies doesn't mean we can't know where some of it does.

[strogonoff]

I think of models as maps—sure, we are able to map Earth with high accuracy (though still have numerous maps all useful for different things), but only because we are now able to be outside of it; this will never be true for the system that we are modeling using scientific method and that we ourselves are part of.

We will forever have many incomplete maps. Scientific method offers one such map, a product of a particular way of attending to the world, and while it is useful for some purposes I am not sure it is inherently better than other maps. It does not help that it tries (maybe not by design, but at least that’s how it seems to be playing out so far) to sidestep the fact that our minds are both map-creators and part of the territory.

[lisper]

> ignoring the fact that our minds are both map-creators and part of the territory seems foolish

That's a straw man. Of course it's foolish. No one is ignoring it.