[lisper]

> If I say all swans are white as my hypothesis

That fails as a scientific hypothesis on purely structural grounds, before you have made any observations. Scientific hypotheses have to explain something. It's not enough to say that all swans are white, you have to say why. "All swans are white" is an observation, not a (scientific) hypothesis.

An example of a legitimate scientific hypothesis is that all swans are white because being white provides swans with some benefit in terms of reproductive fitness (and then you have to go on to say what that benefit is). You can then go on to predict that there might be non-white swans, but that these are expected to be rare because evolutionary pressure would drive non-whiteness out of the gene pool. Or something like that. But "all swans are white" by itself is a non-starter as a scientific hypothesis.

[raincom]

Exactly this. A hypothesis is postulated to explain phenomena (observations). An adhoc hypothesis is one that explains already-selected facts/observations/phenomena. That's why it is very important for a hypothesis to explain novel facts, solve novel problems. That's why Late Philosopher of Science Larry Lauren made a distinction between confirming instances and positive instances.

[ninetyninenine]

Disagree. The scientific method is an attempt at ascertaining the truth. It makes sense to extract a technical definition of what this process is devoid of human factors like “explanation”.

Similar to formalization where we determine the grammatical rules of a language in linguistics or formalization of a physical system to develop models for by using the statistical definition of hypothesis I am doing the same technical extraction here.

Sure informal definitions of hypothesis are useful but formal models are also useful and thus a formal model of science as defined by statistics is useful as well.

I am not denying alternative definition of hypothesis.

I am just saying that for the rigorous categorization of the term “theoretical computer science” we find that it doesn’t fit into the rigorous definition of the scientific method. There is a clear delineation here. If you want to use other definitions where “usefulness” and “explanations” are a factor then what literally stops knitting from being a science? Knitting can be useful and technical in terms composing fabric together…. why isn’t it a science?

You see the issue here and why technicality or formalization is needed? Intuitively we know knitting isn’t a science but we aren’t sure technically why… The technical definition of science elucidates the reason why knitting is not a science… because knitting doesn’t involve hypothesis testing. Plain and simple. We discard fuzzy terms like “usefulness” and it ultimately becomes clear why knitting isn’t a science.

The other reason why the statistical definition of hypothesis is useful is because it helps us delineate categories for terms where our intuition fails and becomes ambiguous. We aren’t sure where “theoretical computer science” fits into terms of science or mathematics. Using technical definitions we find that theoretical computer science is actually a math and not a science.

If we use informal definitions then nobody is sure what computer science is. The mind is confused thinking it’s both a science and a math at the same time and the op starts to use analogies in attempt to justify certain categorizations. Any time you use analogies as proof you hit a sort of failure because analogies don’t prove anything. You need to use definitions as proof.

Hence the need for formal definition's. Your philosophers definition of hypothesis is unfortunately too informal.

[raincom]

You should better read up research done in the domain of philosophy of sciences. I was not providing a definition of hypothesis. Instead, treat as a hypothesis about hypothesis. Definitions don't settle any dispute, whatsoever, because they just replace one symbol (definiendum) with another set of symbols (definiens).

[ninetyninenine]

You gotta make sure two people are disputing about the same symbols man. Otherwise you’re disputing different things.

If you aren’t talking about definitions then one person has one set of symbols and the other has none. So they are still discussing separate things.

[ninetyninenine]

I'm talking at a very technical level ignoring all the cultural stuff around the scientific method like "peer review" or explaining "why"

Basically a hypothesis is a statement that can be true or false. That's it.

The reason I refer to science in this very technical way is because the we are tackling the problem of classification. We are asking the question what is computer science? So to answer the question we need to use very technical definitions where the boundaries of categorization are extremely clear.

Again, at a very technical level a hypothesis is simply a statement that is true or false.

[abeppu]

I think it's a mistake to frame "why" as "cultural stuff around the scientific method" but not actually part of science.

> Basically a hypothesis is a statement that can be true or false. That's it.

No, a statement which can be true or false is just a proposition. The reason that we care about "why" is that a hypothesis has bearing on many falsifiable propositions. It's the difference between "the specific rock I dropped accelerated at 9.8 m/s^2" and Newton's law of universal gravitation.

[ninetyninenine]

I explain here:

https://news.ycombinator.com/item?id=41882469

[lisper]

> a hypothesis is simply a statement that is true or false.

No, that is a proposition, not a hypothesis.

And the requirement that hypotheses be explanatory has nothing to do with culture, it is the distinguishing feature of the scientific method. See: https://blog.rongarret.info/2024/03/a-clean-sheet-introducti...

[ninetyninenine]

False. At the very technical level, a hypothesis is defined by statistics.

https://www.sciencedirect.com/topics/mathematics/statistical....

When you cut through all the cultural and human stuff we place around the scientific method, in the end it is a statistics problem at the most technical level. Everything else makes it fuzzy and hard to define.

[abeppu]

You started by saying that in science you cannot prove things, only provide evidence, but that in math you can prove things.

Then you made an assertion that a hypothesis "is a statement that can be true or false. That's it."

Now, you're asserting that a hypothesis "is defined by statistics". Never mind that humans did a bunch of science before statistics were developed, this is different than your prior statement ("true or false. That's it."), and you seem extremely confident in it.

You're acting as if you have certainty in a statement, but you did not arrive at it by a proof. You've claimed that only the falsifiability of a statement matter, not its explanatory structure, but your own statements about hypotheses are structural and definitional ("defined by statistics", "at the most technical level", "ignoring all the cultural stuff"). You've asserted that a hypothesis is falsifiable, and intrinsically statistical, but you bring no quantitative data in support of this.

I don't know what activity you're engaged in, but it doesn't seem like a rigorous and principled search for truth. And you seem to be more willing to preach about the importance of falsifiability than to apply that concept critically.

[ninetyninenine]

>Then you made an assertion that a hypothesis "is a statement that can be true or false. That's it."

Yeah so? You can make a statement and the actuality of that statement is either true or false. But how you prove that statement to be true is NOT possible. That is my claim. I am also saying it is POSSIBLE to falsify the statement aka disprove it... All statements have this property.

>Now, you're asserting that a hypothesis "is defined by statistics". Never mind that humans did a bunch of science before statistics were developed, this is different than your prior statement ("true or false. That's it."), and you seem extremely confident in it.

Well you're using the "what came first argument" to say that the original scientific definition came before the statistical definition so it's more valid. I disagree.

Language emerges from concepts in attempt to explain things we only have vague understanding of. Concepts like Life and intelligence are ill defined and used for communication on topics we don't completely understand. It's useful to communicate this way but it's useful only because either we don't understand something completely or because we use it as a shortcut. Usually a new word starts off in this fuzzy state and as we understand things better the word takes on a more rigorous definition. Science started out without us understanding statistics, so that's why you have a lot of older definitions attached to it. The technical definition of hypothesis is part of mathematics. That is ultimately the most correct definition but it's not the definition with the most utility. If push comes to shove and we want to categorize a technical concept like computer science, then the technical definition is what matters more.

>You're acting as if you have certainty in a statement, but you did not arrive at it by a proof.

What. I never said this. You are entirely misunderstanding. I said statements can either be true or false. The proof of whether it's actually true or false is a separate issue. My original claim is that proving something true is impossible.

>You've claimed that only the falsifiability of a statement matter, not its explanatory structure, but your own statements about hypotheses are structural and definitional ("defined by statistics", "at the most technical level", "ignoring all the cultural stuff").

I didn't say only the falsifiability of a statement matters. I never said this. I said falsifying a statement is the only thing we logically have the ability to do. It still matters to do correlations and other types of things related to science but at a technical level we aren't proving anything. At a technical level things can only be falsified. This is not to say OTHER things don't matter. They do matter.

>You've asserted that a hypothesis is falsifiable, and intrinsically statistical, but you bring no quantitative data in support of this.

Did I not post a link to a resource stating this? This is Data supporting my point. If you want quantitative data for the English definition a word, I'm sorry but that's just not physically possible. English definitions cannot be quantified into numbers for any meaningful numerical analysis.

I know your question is just sort of rhetorical. Basically you think I'm being too pedantic and you're trying to illustrate a contradiction in my own logic. I don't deny it. In the end I'm using my own personal opinion here. But I share my opinion here because I believe if that the MAJORITY of people completely UNDERSTAND what I'm saying they will AGREE with me. That's all. But again opinion. And either way nothing can really be proven can it? Especially for english definitions.

>I don't know what activity you're engaged in, but it doesn't seem like a rigorous and principled search for truth. And you seem to be more willing to preach about the importance of falsifiability than to apply that concept critically.

What are you trying to say here? This is false. And it approaches the point of accusatory and a lie. I don't preach the importance of falsifiability. I am just saying that is the only possible technical thing we can do in terms of determining if something is true or false. I get technical because we ARE CATEGORIZING technical concepts and definitions so it's appropriate to DO THIS. If we are casually conversing or trying to understand concepts then of course we can revert to the more laid back way of communicating and fuzzy way of defining things.

But what's actually going on here is that we are determining whether or not a technical concept: "Theoretical computer science" is math or not? Such detailed and rigorous categorization REQUIRES the use of detailed and rigorous definitions.

[Koshkin]

> You can make a statement and the actuality of that statement is either true or false.

It can also be "not even wrong."

[lisper]

Now you're moving the goal posts. First, it was "a hypothesis is a statement that can be true or false. That's it." Now it's "a hypothesis is defined by statistics". Get back to me when you've decided what your position actually is. I'm not going to waste time debunking every half-baked idea you can come up with.

[ninetyninenine]

The goal posts were never moved. What moved is your understanding of what I'm trying to say.

A hypothesis is a statement that is TRUE or FALSE. And THIS is HOW it's defined by statistics.

>I'm not going to waste time debunking every half-baked idea you can come up with.

I'm sorry but this attitude is offensive and against the rules here. I can't participate in any further discussion with you because of this. Thank you for your time and good day.

[Koshkin]

I don't think that physics (for example) uses the word "hypothesis" in the same sense, even though there are, indeed, parallels between the two sciences.

[lisper]

> The goal posts were never moved.

Yes, they were. You started out by saying, "a hypothesis is a statement that can be true or false. That's it." But that isn't "it" as you yourself found it necessary to explain.

> I can't participate in any further discussion

That's fine, I'm writing this response for the benefit of others who might be reading this: not only are both of your claims wrong, they are transparently wrong. "Santa Claus lives at the North Pole" is a statement that can be either true or false (it happens to be false) but it doesn't have the same standing as a scientific hypothesis as, say, "the laws of physics are invariant under Galilean relativity." That happens to be false too, but it is not nearly as false as "Santa Claus lives at the North Pole." And neither of these has anything to do with statistics.

You clearly don't understand the first thing about how the scientific method actually works, starting with the fact that you cannot strip away "the cultural stuff." Science is a human endeavor. It is ultimately about accounting for our subjective experiences because that is all any of us have direct access to. This includes, but is not limited to, the fact that there seems to be this thing people do called "science", which has a bunch of interesting properties and turns out to be tremendously useful. Statistics are only a small part of that endeavor. An important part to be sure, but a small part nonetheless.

[Scarblac]

Yes, it's a statement thats true or false.

But it must also be useful. We don't do science just to enumerate trivial true statements, after all.

To be useful it needs to predict things.

And when it's falsified (say your hypothesis explained why swans are white, but you found a black one), it doesn't get discarded immediately. It's still useful until someone comes up with a better hypothesis that fits with white swans and the occasional black swan.

[ninetyninenine]

> But it must also be useful. We don't do science just to enumerate trivial true statements, after all.

Sure it can be useful. Think of it like a mathematical theorem. What’s the point of the theorem unless it’s useful? Why would a book define a theorem if it wasn’t useful?

So theorems in math need to be useful. But such a quality is human and fuzzy in nature. What does it mean to be useful? And everyone has a different definition of useful. That’s why the definition of a theorem doesn’t include the term useful Even though generally speaking it’s a bit of a requirement if an author were to define a theorem in a book.

The definition for hypothesis that I use follows the exact same process. It is a rigorous technical definition that we are using for rigorous and detailed categorization of another term: “Theoretical computer science”.

Thus in the face of such a task I use the most rigorous definition of hypothesis available. I discard fuzzy terms like usefulness or expositions into “why” to determine categorization.

The statistical hypothesis which defines the term hypothesis in a very technical way. In fact, in statistics, hypothesis testing is basically the technical definition of the scientific method. Following this definition we can clearly see the boundaries of things more clearly.

Theoretical computer science does not involve hypothesis testing. It is mathematics because it involves axioms and theorems.