Comment by kuhewa | Hacker News Reader

Comment by kuhewa | Hacker News Reader

Parent Full thread

> My philosophy professor had a multi-volume (non-English-language) textbook called “Introduction to the science of philosophy”, so, uh, I dunno.

I too am not sure what you are implying there. The enterprise, assuming it had some merit, sounds like evidence of complementarity?

> it brings along the burden of pointing out some holes in the scientific enterprise, a general conception of how to fill them, and at least some practical success in doing so.

Well, yeah, all of those are a given. The rest of your comment goes more toward philosophers speculating on aspects of the material world (i.e. the realm of scientific inquiry), but there is a lot more to philosophy. The holes in the scientific enterprise have long been well delineated, which isn't a criticism, but for example we can't derive human values solely from science [1].

Also, philosophy underpins science. Whenever a hypothesis is tested, there is are philosophically-grounded assumptions being made. The epistemological implications for any given scientific finding depend on the underlying philosophical framework being assumed.

[1] https://en.wikipedia.org/wiki/Is%E2%80%93ought_problem

I'm not sure "underpins" is the right word here.

A theory of ants wouldn't "underpin" the behavior of ants. If the ants behave differently than the theory, then the theory is wrong and should be changed. The ant theory is something in the minds of outside observers, not the ants, and ant behavior doesn't rely on it. The dependency goes the other way: the observers modify their theory to better describe the ants. The ants would exist without the theory, but theory would be pointless if there were no ants.

Is a philosophy of science similar? Or does it have a practical effect on scientific work?

Unlike ants, scientists can learn a philosophy of science, and perhaps believe it. Does this affect their work?

One reason it might not affect their work in practice is that they didn't learn that particular philosophy. Also, perhaps different scientists might learn different philosophies without practical effect.

Underpins is the correct word. There is no comparison of observations of any behaviour to a theory or hypothesis about any behaviour outside of an epistemological framework that makes certain assumptions, even if it is unwittingly. Neither those assumptions nor the framework itself can be derived from science itself.

> Is a philosophy of science similar? Or does it have a practical effect on scientific work?

Absolutely, what a p value is depends on your philosophy of science. Whether your statistical analysis even involves p values also depends on it.

> Also, perhaps different scientists might learn different philosophies without practical effect.

Yes, if you stumble upon some simple causal relationship of such massive effect size it is undeniably present beyond a reasonable doubt, it may not matter if a Bayesian or frequentist practitioner came across it. However it certainly can matter what framework evidence gets analysed, considered and aggregated when the observable data themselves are essentially the same.

> There is no comparison of observations [...] to a theory or hypothesis [....] outside of an epistemological framework that makes certain assumptions, even if it is unwittingly.

Okay, let's assume animals (such as scientists) observe things, record them, and react to them without an explicit theory in mind, but there's an implicit epistemological framework that describes how they behave.

It seems like you still need to build your epistemology to match the animals' behaviors, or it's not the one they use? When scientists do math, you need to observe how they actually use math. How do they actually set up and run an experiment or write a paper? It might be different than you imagine?

This is what David Chapman calls the "ethnomethodological flip" [1].

Scientists also might use math differently from how they claim they use it in a formal paper, which doesn't include all the blind alleys and mistakes. A scientific paper is a cleaned-up just-so story.

A fun example of ethnomethodology is studying exactly how a scientist follows the formal procedure for doing a PCR test, including small mistakes that they don't explain and you might not even notice in the demonstration video unless you watch it very carefully, multiple times. [2]

It seems like a very cool thing to do that's rarely done. It might help for coming up with better philosophy?

[1] https://metarationality.com/ethnomethodological-flip [2] https://metarationality.com/rational-pcr

Science is not an observable thing in itself like ant behavior is. Science (especially the scientific method) is a framework/pattern of thought that can be used to state/observe things about the universe.

Why that pattern of thought is correct in stating anything is absolutely part of philosophy.

I'm under the impression that science is what scientists do, which does seem to be observable?

Not exactly. Scientists are labelled scientists based on the methodologies they use and the fields they study in (which we call science).

When a person practices science, they are called a scientist. When a scientist exhibits a behavior, that behavior is not necessarily called science.

That's a good clarification. However, exactly which methodologies a scientist uses doesn't seem like it makes a difference as to whether it counts as science, so long as it's still in the general spirit of the thing? This is decided culturally.

Also, it seems like there's more to understanding methodology than deciding what counts as science?

To go back to the ant analogy, people had fuzzy ideas about what an ant looks like. Some people might have called other bugs ants even though today we don't. This later led to more precise definitions under scientific taxonomy, where some species are scientifically classified as ants. But there's a lot more to understanding ant behavior than deciding what counts as an ant.

(Also, the definition of what an ant is co-evolved with scientific understanding of ants. Taxonomy existed before the theory of evolution and taxonomies were refined with genetic testing.)

The heart of the science profession is the scientific method (just like the heart of the firefighting profession is fighting fires), but there are many other activities that scientists and firefighters perform that are not science or firefighting, such as writing grant proposals or doing maintenance on firetrucks.

Ants (today) are classified differently from scientists and firefighters. They are not defined based on a specific thing they do, like "anting"; rather, they are defined based on what they are, and their behavior is irrelevant to their classification.

Historically, animal categories were defined much more like professions. A fish was something that primarily swims; a bird was something that primarily flies; a worm was something that primarily slithers; a beast (the category that ants fell into) was something that primarily crawls. Even concepts like "animal" and "plant" were defined this way: animals are animated, while plants are planted in place. There was a lot of debate on how to categorize lifeforms that exhibited less-than-crytal-clear modes of locomotion, just as there is debate today on whether a given person is actually a scientist or not (do they do real science or pseudoscience? do they do a lot of science or is too little science for it to count? etc.).

This, of course, is radically different from the way we classify biological lifeforms now, although there are a few odd historical holdovers (like "fish", which is a catch-all term for aquatic vertebrates without terrestrial ancestors, even though some of them are more closely related to land animals than they are to other fish).

>Also, philosophy underpins science. Whenever a hypothesis is tested, there is are philosophically-grounded assumptions being made. The epistemological implications for any given scientific finding depend on the underlying philosophical framework being assumed.

P(A|B) = P(B|A)P(A)/P(B). To the extent that philosophy underpins science, it does so because scientists are bad at math.

Your comment appears to imply Bayes theorem is all you need for a scientific framework and outside of that is just inadequate mathematical know-how to deploy it. I find it amazing really, a quintessential straight-from-the-summit [1] HN comment for a couple reasons:

First, it appears to imply that such a probabilistic inductive approach to science would be free of any philosophical baggage or assumptions, when deploying Bayesianism requires an interpretation about what a probability itself is. Don't take it from me though, perhaps from Andrew Gelman, the guy who wrote the book on Bayesian data analysis [2,3].

Then, wrt the charge that those who do not use such an inductive approach (or outright rejected it, e.g. in favour of falsificationism) are bad at math. Which would include the statisticians who developed the null hypothesis significance testing framework that is still pretty dominant in science today: Jerzy Neyman, Egon Pearson, Ronald Fisher (who literally coined the term 'Bayesian') etc. There's a lot of criticism worth making about Fisher, but I'm not sure if anyone has called the guy that developed linear discriminant analysis bad at math before.

[1] https://www.smbc-comics.com/comic/2011-12-28

[2] Bayesian Data Analysis http://www.stat.columbia.edu/~gelman/book/

[3] Philosophy and the practice of Bayesian statistics http://www.stat.columbia.edu/~gelman/research/published/phil...

>First, it appears to imply that such a probabilistic inductive approach to science would be free of any philosophical baggage or assumptions

I'm arguing that philosophical baggage is irrelevant to the current practice of science, because the overwhelming majority of published papers have serious and obvious methodological deficiencies that we have collectively agreed to ignore. Science as practised today is a desperate struggle to demonstrate something (p ≤ 0.05) by any means necessary. Established statistical methods have become a means to conceal rather than illuminate. This isn't the fault of individual working scientists, but the fault of the basic information architecture of science and a ritualistic, cargo-cult approach to understanding data. Bayes theorem probably isn't all we need, but it is all we need to spark a scientific renaissance if only we would use the damned thing.

Something to keep in mind is that science is not only practiced in cutthroat academic institutions. Arguably most of the science being practiced today is happening in R&D departments across the world, where there is a strong financial incentive to beware of false positives due to the expense of policy changes.

I worked in private agricultural research for many years, and the issues you describe here did not apply to us. A single instance of statistical significance was insufficient to make any assumptions or adjust any policies. Of course we would present those results if we got them, but nobody was getting excited about it until we got the same results consistently several more times, in several different locations, with several different research teams. Being the first to get a positive result was no more meaningful that being the 10th to get a positive result. Promotions were based on your ability to design and conduct solid research, regardless of the outcome.

I hold a master in philosophy and it was my first big love.

Over the years I came however to a conclusion opposite to yours.

> Also, philosophy underpins science. Whenever a hypothesis is tested, there is are philosophically-grounded assumptions being made. The epistemological implications for any given scientific finding depend on the underlying philosophical framework being assumed.

I think that is not true. Science does not get its merit from philosophical underpinnings, but from working in practice.

Methods that work to generate and test knowledge. Math is the precise language needed to speak about these methods and the knowledge. Also because it works. Look at the achievements of science. That is how you get convinced it gives us a grip on reality. Hic Rhodos, hic salta.

My working assumption at this time is that philosophy has no such methods. We are not further than when Kant said that no fighter ever won and could stand his ground in metaphysics on any topic.

The reason might be that philosophy is actually practically mostly irrelevant. I have not seen one undisputed statement of philosophy. And so it can neither test its statements, nor let others see their validity.

I concede it is practically relevant in another sense: world views have taken grip of groups and still do, and influence history one way or the other. But that seems at least to me more a social-anthropological phenomenon like support and resistance in trading.

Here’s one way it’s relevant: universities are structured according to a philosophical opinion about the nature of the universe (or at least the nature of knowledge about the universe). That set of decisions in turn steers, at a very fundamental level, the path and velocity of scientific inquiry.

I fail to see what you mean exactly. Could you explain?

Specifically why the opinion is philosophical and not just some historic-pragmatic pattern matching and grouping?

Did you get much exposure to the philosophy of science during your masters? I imagine the answer would be yes, but I am surprised that concept doesn't ring a bell as it sounds quite similar to what Kuhn describes in The Structure of Scientific Revolutions, although not necessarily with universities as the institutions upholding scientific paradigms

> I think that is not true. Science does not get its merit from philosophical underpinnings, but from working in practice.

In the day-to-day practice of science i.e. when empirical inquiry 'works', there certainly are underpinning philosophical assumptions, whether or not they are reconsidered or appreciated with every experiment. The implicit in the act of hypothesis testing, some variant of which most scientists in day-to-day practice, are assumptions about the nature of probabilities, inference etc. The NHST framework that is typically used came about after extensive battles over philosophical considerations that apply to significance/hypothesis testing between Neyman and Pearson vs Fisher. The fact that I never write that a hypothesis is (as a research biologist) is 'proven', but some variant that of it having withstood an attempt at falsification, is loaded with Popperian critical rationalism.

> Math is the precise language needed to speak about these methods and the knowledge.

Except math can't map directly onto reality, or data generating processes that are studied, unless you are presupposing some kind of logical positivism (and I doubt you are). We need probabilities, statistics, and frameworks to map all of this uncertainty, and they must be underpinned by some sort of philosophical assumptions that can't be derived from science itself.

> Look at the achievements of science. That is how you get convinced it gives us a grip on reality.

But that in and of itself is a philosophy of science, one of instrumentalism. However, it only extends to whether science can be useful, but not whether it is accurately describing reality or is true.

Thank you for your thoughtful comment.

You have many good points.

Let me just say that Popper has been philosophically criticized to the point that some say it is a dead horse. Why are we still using this mixture of Fisherian and Neymar-Pearson hypothesis testing (that is if we don't use bayesian methods)? Because it practically works well, not because Popper was right or found a deep philosophical truth.

These methods just generate more often than not knowledge, as we can judge from the consequences.

I argue that nobody cares if the assumptions we put into the frameworks are philosophical true - they are possibilities, and we try some out. So far we seem to be doing pretty well, no matter what philosophers say about the truth of these assumptions.

I also think bending philosophy to apply to practical advice like "use the tool that works" will not leave much to the notion of philosophy.

But it's not that I have a fixed metaphysical position here. I really only use the tool that was most promising in the past for the task at hand. Never needed philosophy.

> I also think bending philosophy to apply to practical advice like "use the tool that works" will not leave much to the notion of philosophy.

I reckon maybe there's something to this. One thing that comes to mind though is: Granting the fact the NHST is now broadly used by practitioners without knowledge of its background simply because it works, I am not sure that necessarily indicates the background isn't important, as my ignorance of my monitor's inner workings does not mean that electricity is not important.

> The reason might be that philosophy is actually practically mostly irrelevant. I have not seen one undisputed statement of philosophy. And so it can neither test its statements, nor let others see their validity.

This is the core confusion I think. I find philosophy very relevant for the way I reason and solve problems and evaluate arguments, and in this sense philosophy is powerfully practical. But it’s true that for any given claim, there is always the possibility of taking the opposite position. This lack of final, case-closed consensus doesn’t mean that philosophers individually haven’t converged on true beliefs or haven’t made progress. It’s just that unlike mathematical truths, we can write out the proofs of our arguments, but there’s always someone who disagrees about one of the starting assumptions. So then civilians who haven’t heard of people like Parfit think to themselves, wow 2k years and you can’t tell me anything about ethics or logic or epistemology — it’s like bro just read the literature

I can't say "been there, done that", because I know little about how you came to the position you hold. And I've had my share of burying hypotheses I've held for long times, so chances are I'm wrong.

But I held similar views. What moved me away from these views was the experience that in science you have methods that will let you see with high probability when your thesis is wrong.

Philosophy does not have such methods. You cannot only take the opposite claim for almost anything, but in my experience that claim has actually been taken by another philosopher for almost any topic.

The current consensus is in my experience lead by the people with the loudest megaphone. It's not the best theory given the things that really happen.

No philosophy (in the sense of actual writings of a philosopher) was causally involved in bringing the first astronaut to the moon; in building the first pacemaker - I would argue in none of anything where you could say: if you can do this with it, its probably on to something. Its methods seem to work.

As said, I cannot judge your thinking in any way, but this led me to question that philosophy is not practically relevant to me. How can I even judge if it works? And if I can't, is this not blind trust? Like an imaginary screwdriver for imaginary screws.

My education and work history are in the biological sciences (although I have admittedly recently made a career change).

To my mind, science (formerly known known as natural philosophy) is a subset of philosophy. However, it is a subset that has grown to dwarf the other subsets and is given separate, special attention. You do not learn much, if any, science in a philosophy degree because philosophy degrees now focus on philosophies that have not been spun off into separate degrees/fields.

Within the sciences, you do learn philosophy (or at least it factored highly in my undergraduate degree), but it's not about Aristotle or Nietzsche. It's about the assumptions and logic underpinning the scientific method, statistical analysis, etc.

My introductory classes, at least, covered a number of arguments and assumptions that (to my ear) are very much questions of philosophy. For example, scientific inquiry is dependent on the assumption that the laws of the universe are consistent across time, meaning that experiments performed now can nonetheless offer insight into past and future phenomena.

Reply on news.ycombinator.com