6 ms·

> 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. Sci

by lisper 2y ago

> 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.

ninetyninenine 2y ago
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 2y ago
  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 2y ago
    I explain here: https://news.ycombinator.com/item?id=41882469 https://news.ycombinator.com/item?id=41882469
- lisper 2y ago
  > 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-introduction-to.html https://blog.rongarret.info/2024/03/a-clean-sheet-introducti...
  - ninetyninenine 2y ago
    False. At the very technical level, a hypothesis is defined by statistics. https://www.sciencedirect.com/topics/mathematics/statistical-hypothesis#:~:text=Statistical%20hypothesis%3A%20A%20statement%20about,alternative%20to%20the%20null%20hypothesis 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 2y ago
      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.
- Scarblac 2y ago
  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 2y ago
    > 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.
raincom 2y ago
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 2y ago
  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.

by lisper 2y ago

ninetyninenine 2y ago

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 2y ago
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 2y ago
  I explain here: https://news.ycombinator.com/item?id=41882469 https://news.ycombinator.com/item?id=41882469
lisper 2y ago
> 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-introduction-to.html https://blog.rongarret.info/2024/03/a-clean-sheet-introducti...
- ninetyninenine 2y ago
  False. At the very technical level, a hypothesis is defined by statistics. https://www.sciencedirect.com/topics/mathematics/statistical-hypothesis#:~:text=Statistical%20hypothesis%3A%20A%20statement%20about,alternative%20to%20the%20null%20hypothesis 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 2y ago
    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.
Scarblac 2y ago
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 2y ago
  > 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.

raincom 2y ago

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 2y ago
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.