Talk:Natural science

From Wikipedia, the free encyclopedia

	This article has been identified by the Version 1.0 Editorial Team as a Core Topic, one of the 150 most important articles for any encyclopedia to have. Please help improve this article as we push to 1.0. If you'd like help with this article, you may nominate it for the core topics collaboration.
Start	Natural science has been rated as Start-Class on the assessment scale.

Contents 1 Early talk 2 Why "nonhuman"? 3 Nice URL 4 Computer science 4.1 Why Computer Science IS a natural science 4.2 Why computer science is NOT a natural science 4.2.1 Objections 5 Scope of natural science 6 Mapping the sciences: scientific adjectives/name of the science(s) 7 This Article is Forever Consigned to be a Quagmire. 8 Expansion of this page

[edit] Early talk

I must dispute this definition. I've always seen "natural science" used to differentiate between "science" as a discipline following the scientific method, and "science" as a field of knowledge generally, e.g. computer science or even "the science of theology". -- April

This distinction is sometimes made, but the distinction made here is also used, and I put it in because the article "Physical science" alludes to it.

• I understand, and will put in the alternate definition as, well, an alternate definition. -- April

Physical sciences are part of natural sciences, so I eliminated the first paragraph, that "contrasted" the two terms. AstroNomer 22:05 Jul 30, 2002 (PDT)

The sentence beginning "The term natural science is also used to differentiate" leaves the reader no wiser as to how it does this. If I say, "the term sensible person is used to differentiate between people with left-wing beliefs and those with right-wing beliefs," which belief system am I espousing? Okay, what I really came here to ask about was, "Is computer science a natural science?" Jdforrester claims so in a recent edit to the Main Page, but I would have thought that a natural science should have something to do with nature... -- Oliver P. 23:51 1 Jul 2003 (UTC)

Hmm. No, I don't think CS is a natural science, much as I don't think that Mathematics is one. The heading says 'Philosophy, Mathematics, Natural Science' and that's what I would say CS comes under. Certainly, it fits much better there than as an applied science, where it was previously. -- James F. 00:05 2 Jul 2003 (UTC)

CS is not a natural science. It's a mathematical science. It belongs with Mathematics and Statistics. Poor Yorick 05:10 2 Jul 2003 (UTC)

there is always going to be a conflict between historical and descriptive names. The historical classification of natural sciences is much broader, and includes loads of stuff including maths and computing which today we would not call natural. But the prevalence of old institutions in education mean we have to accept definitions which don't match everyday use of the words. Don't forget in older UK universities a degree in science makes you a "master of arts" and a doctrate in science makes you a "doctor of philosophy". I have tried to explain the text a bit this way but you are welcome to improve my attempt --(talk to)BozMo 10:11, 21 May 2004 (UTC)

I have always seen "natural science" as synonymous with "biological science," and in opposition to "physical science". Astronomy would be a physical science, but not a biological one. but with the definition listed here, it works as a natural science. perhaps we should clarify that there are two different ways to define the term natural science, and have the version that excludes the physical sciences link to a biological sciences page?. --zandperl 17:38, 23 Nov 2003 (UTC)

Well it's worth noting that the 'Natural Sciences Tripos' is still going strong at Cambridge, UK and that under it I studied Physics, Chemistry and Crystalline State in the first year (the rule was three 'experimental sciences'), ... and maths :-). Linuxlad

Hi all. Would you please consider commenting on my POV that CS is a natural science! It's the science that studies computations, a natural phenomenon (as opposed to supernatural...) Regard, Powo

The only natural numbers are those used for counting sheep. All else mathematical is product of (fevered?) imagination.Linuxlad

[edit] Why "nonhuman"?

What about human biology? —siroχo 04:20, Sep 7, 2004 (UTC)

Or human anatomy and physiology, which includes a healthy dose of biochemistry? What happened to this discussion? Let's reopen it, this is an important topic... Exmachina 12:33, 28 Jul 2005 (GMT-5)

[edit] Nice URL

Short, simple, to the point. FuelWagon 19:13, 23 August 2005 (UTC)

[edit] Computer science

Hi all. THe article says that Computer science is not a natural science, but provides may tools for other sciences.

[edit] Why Computer Science IS a natural science

• Computer science studies Information, just like physics studies Energy.
• Computer science studies computations, the evolution over time of a physical system (a computer).

Arguably, computer science is broader than either of those two parametrizations. Information science can be thought of as a branch of physics which studies physical information. Likewise the study of computations is entirely mathematical and not an empirical subject, per se. What's more, computer science encompasses the study of systems, hardware, software, etc. that are not technically part of either of those two studies. Computer science, in many ways, has parts that are natural science and parts that aren't. It shouldn't be categorized as a natural science any more than, say, engineering. --ScienceApologist 12:25, 24 April 2006 (UTC)

[edit] Why computer science is NOT a natural science

• Because it is more like mathematics
• Becasue it is more like engineering

[edit] Objections

• "More like mathematics": CS has sometimes a very mathematical approach. And so? Physics too!
• More like engineering: CS is strongly linked with engineering, it creates and studies objects (software/hardware) which were man made. And so? Chemistry too!

I dont think CS can be listed as not being a natural science without explaining why it is not. Perhaps, in fact, should CS be listed as being a natural science, but with short explanation as why this classification would be accepted or challenged.

Regards, Powo

Natural science is to do with manifestations of the Natural (aka real) world - and in Dr Johnson's famous definition real=kickable (ie something you can do experiments on). A number line or Argand diagram are _not_ manifestations of this real world, they are constructs, like Shakespeare's Macbeth; neither are such different human _constructs_ as C++ or x86 microcode (the microelectronics of the electrons in the Si substrate of an x86 _is_ however a fair subject for experiment, part of a materials science topic). Natural Science used to be studied by 18th century clerics - their question I suggest, is 'am I looking at a piece of God's handiwork or an artefact of Man's' Linuxlad 15:57, 23 April 2006 (UTC)

Hi. Arguably, C++ and x86 microcode are not really part of computer science, or at least, are part of the engineering side of computer science (also called computer engineering). I am not saying that CS is not related to computer engineering. Iam saying that CS studies computations and information, and that these are physical phenomena, part of the real world, and that they are natural (as opposed to super-natural) phenomena. I do not understand why you talk of a number line and an Argand diagram. What is the relation with computer science? I understand that mathematics are not a natural science. I understand that computer science is a formal science, but I have not yet seen a good reason why it is not also a natural science (and probably also an engineering discipline. Perhaps the stunning beauty of CS is that it transcands disciplines, and make the usuall classifications somehow unadequate. I would like this article to aknowledge the fact that CS is, amongst other things, a natural science (or to be convinced that it is not). Do you consider computations or information to be non natural? Do you consider that studying them is not science? I agree that mathematical approaches are often used to study these concepts, and that experimental or empirical approaches are not the rule, but this is also true in theoretical physics (e.g. chord theory). Is theoretical aspects of physics not a natutral science? My points: compuations and information are natural phenomea, CS studies them, therefore CS is a natural science. Powo

It is because "experimental or empirical approaches are not the rule" in computer science that CS is not a natural science. Even theoretical physics is judged in terms of its ultimate agreement with empirical observations (although this may be delayed until those observations can be made).

I agree that computer science can usefully be viewed as an engineering discipline, as can statistics. This does not make them natural sciences.

Whether computations can only be conducted by something "natural", through "physical phenomena", is (I think) a matter of faith, and not one we can hope to resolve on this talk page. -- Avenue 12:23, 24 April 2006 (UTC)

Hello. You say it is a matter of faith wether computations are a "physical phenomena" that can only be conducted by something "natural"? But what do you mean? I am only aware of computations as a mathematical object and as the evolution over time of a computer. The second being, I understand, a natural and physical phenomena. What faith got to do with that? Regards, --Powo 12:51, 24 April 2006 (UTC)

I agree that the "evolution over time of a computer" is a physical phenomenon, although its goal is to perform computations and these are usually better considered at a more abstract level. Computations could also be considered as a process within the mind, and this is where I saw a possible supernatural interpretation. But given that computer science usually focuses on computations in silico, a physical process, perhaps this is irrelevant to the main question here (of whether CS is a natural science). In hindsight it's probably the weakest of my three points above. -- Avenue 13:55, 24 April 2006 (UTC)

ok... So next, about the fact that you say that even theoretical physics is ultimately judged by its agreement with empirical observation. How is this different from CS??? I have never heard that (even theoretical) computer science produces theories wich are contradicted by empirical observation. When I say that empirical observation is not the rule, it doese not mean it can not be performed. Actually, although it is not the rule, empirical observation often motivates analytical introspection which is then experimentally validated (but that is not the rule either!) Typically, the "physical system" (the computer/network/etc...) is modelized formally. The model is studied mathematically. The conclusions are validated experimentally. Regards, --Powo 14:24, 24 April 2006 (UTC)

My impression of at least some theoretical CS is that agreement with empirical results is not seen as an important criterion. But let's leave that aside, as you seem to be arguing more from explicitly empirical CS studies. I think here the issue is whether the system being studied is a "natural" system. Clearly for computers, networks, etc it is a physical system. But "physical" does not necessarily mean "natural", and the systems studied in CS are probably better described as artificial than natural. No doubt there are areas at the boundaries where this is less clear. There are also similarly fuzzy boundaries for some social sciences, e.g. psychology. But I believe that most of computer science does not study natural systems. -- Avenue 15:51, 24 April 2006 (UTC)

Hi again. I disagree with your opposition between natural and artificial. Artificial constructions (like a computer) are natural, as opposed to super-natural. Is this not the usual acceptation of the word "natural" in philosophy of science? What is nature? Some thought on this can be found on the nature page, under the section "natural and artificial". If you want to distinguish between natural and artificial instead of distinguishing between natural and super-natural, I am afraid large part of chemistry and physics will not be natural. What about OGM's? Powo

I thought I'd add my thoughts here, since the many points made here seem somewhat oblique to my understanding of the position of Computer Science. The course I took at university was once called computation, has since been renamed computer science. I'm of the understanding that the following list is what usually comprises the study of computer science, and many an undergraduate will have touched many of them:

• Complexity
• Algorithms, program transformation and correctness
• Formal languages, Set theory, Mathematical logic
• Numerical Analysis
• Discete Maths, Algebra
• Category theory
• Lambda calculus
• Concurrency

Since these are all deductive disciplines, none of which actually require a computer for their formalisms, and what I view to be the mainstay of computer science, I'm hard-pressed to see how computer science is a natural science (which, in my lexicon, is pretty synonymous with the harder, empirical sciences that are branches of physics, chemistry, and biology). I'd say that programming is essentially a (soft) engineering discipline, and the workings of modern computers is probably in the the realm of one of electronics, materials, and physics. For examples of what I'd consider areas and results in computer science, I'd point to the Halting problem, the Curry-Howard isomorphism, Hindley-Milner type theory, lowest bounds on eg matrix multiplication. Tez 17:13, 24 April 2006 (UTC)

Hi! I agree with you: CS is very often deductive. That is also true (less) often for physics. Probably for other disciplines too, even less often. So what is the criterion to be a "natural science"? Deductive reasoning is forbidden? Up to what percentage? I get your point I think, but I dont like the arbitrarness of it. I think it is pretty ovious that my point of view will not get approbation by the community. It seems CS is different from other sciences. Obviously, it has a very strong enginneering component. It has a very strong mathematical flavour: it is a formal/deductive science. However, unlike mathematics, the results in CS describe what can be expected by a very concrete physical system: a computer/computing system. Unlike other disciplines, the physical system it studies are exclusively artificial (unless, maybe, the universe itself is strongly deterministic and follows computational rules?). Ok, I reacon there will not be an allusion on this page to the fact computations and information are physical (natural) phenomena. Computer science...a new kind of science??? Regards, Powo

Up to what percentage? Up to any! If the final arbiter of something's correctness and accuracy is experimentation and testing, it is an empirical science, and not maths. I'd be hard-pushed to see it any other way, but I'd be dissuaded if you could find reference to some empirical experiment designed to falsify some hypothesis of copmuter science (I will note here the circularity, that I'm looking to define CS by finding refs to a CS experiment...). I believe this will be very hard, since most theorems' results (and they are properly theorems) are expressed in some formalism, whether it be UTMs or a typed lambda-calculus, say, and not to 12 decimal places, but exactly.

It just so happens that the theories are very well-modelled by computing hardware and processes, but no theorem in CS has any empirical content. There are no operational definitions or experiments. In fact, I'd say that natural sciences are the ones that can most easily be seen conforming to the scientific method, which opposes it to purely deductive disciplines, and the more 'social' sciences.

For example, if someone comes up with a new computing paradigm (a quantum computer, say), that will still not change the fact that complexity of quicksort, say, is O(n) = n log n. Similarly, Riemannian geometry is maths even though it happens to well-model the spacetime of general relativity, but the study of Riemannian geometry is surely mathematics.

To make it clear, I believe CS is an entirely deductive discipline, and thus a purely mathematical pursuit. Tez 09:00, 25 April 2006 (UTC)

CS is most certainly not an entirely deductive science. You run simulations, and observe results. That is empirical. You propose an algorithm, conjecture that it is efficient at doing something, that it runs quickly, etc..., and run simulations to show that it succeeds in achieving whatever youre claim was about. That is experimental, without an ounce of mathematical deduction. (Of course, an analytical proof is prefered to an experimental verification when there is one). You say Riemannian geometry is math, it models well spacetime relativity??? I dont see it this way. Relativity is mathematically formulated, and it models quite well the real world (i.e., physics is often being deductive). It is easy to think of a way of falsifying erroneous CS claims through experiment. E.g., if you come with a proof that merge-sort runs in O(n), I can simply run merge-sort on many instances, measure its run time, and see its growth is not O(n) but O(nlogn). I will then say that there is something wrong in your proof because it implies something which is contradicated by the real world. Also, if I was not able to find a proof of the time complexity of merge sort, I could run experiment and find empirically that it has an O(nlogn) worst case complexity. I think you would be surprised of the number of CS publications which finish with a paragraph titled something like "Experimental validation"... Regards, --Powo 13:06, 25 April 2006 (UTC)

Ok, firstly regarding the GR/riemannian geometry example; riemannian geometry on its own is a branch of mathematics, agreed? In GR, spacetime, light, massive entities are modelled by this geometry. That we need to experiment to determine whether it models spacetime accurately is exactly what makes GR part of physics. If, suddenly, we measured the bending of light that was at odds with the predictions of GR, that would in no way invalidate riemannian geometry, only (the current formulation of) GR.

So, what I've tried to describe is that there is a formalism, and a model. You are saying that the results in CS require, at least a little bit, experiments that determine the soundness and consistency between that formalism and enities in the "real world". I am saying that CS is purely the study of the formalism. Using computers can indeed help, just like mathematicians use Matlab or Mathematica, but is in no way required. I mean, I have never seen eg. type-theory formulated in terms of any real-life computer implementation, nor the upper-bound of merge-sort. Sure, you can test it, but these results are proved because they are always stated in a formalism, without any reference to non-abstract entities. I mean, you'd only need to state one result that is dependent on experiment for me to move on this point. I can't think of anywhere where the validity of computer science rests on the performance or accuracy of a computer. We can note that Turing, Church, Rosser all came up with their formalisms, and the Halting problem before anything like a physical computer became widespread.

Regarding simulations and the like; what are you simulating? Are you simulating finite-element models, aerodynamics? That is not CS, whether it is a natural or deductive science. You are merely using a computer as a tool for some other domain. If you are testing the performance, efficiency, or accuracy of a computer, I'd label that engineering (or whatever relevant flavour).

Actually, the test to determine CS's empirical nature would simply be to name a result that is dependent on experiment. Granted, if you do so, there might be additional hurdles classifying that result as within the realms of CS (and this is the sort of circular and/or equivocal reasoning I'm willing to admit I may have committed), but that would be easier to determine by references and citations from academic papers. Again, the results, theories and areas of study I have alluded to are all purely formal and deductive, and you would find all of them being at the core of any decent (equivocation again? Maybe) undergrad university course in CS. Tez 14:09, 25 April 2006 (UTC)

Hi again. You point out differences between physics and CS. E.g. you point out that the first CS results were anterior to the actual existence of real world computers (at least digital computers...). I agree with you. This only shows that the historical development of CS and physics are very different, and so what? I dont know what to say about your claim that CS is purely the study of the formalism. It just doese not seem reasonable at all: of course, CS is REALLY concerned with the applications to real world, and theoretists are REALLY concerned with modeling efficiently the physical world: compuations as they happen in physical systems!

You point out that simulating a physical system (e.g. aerodynamics) on a computer is not computer science. 100% agree with you. I speak of simulating computational systems on a computer, e.g. you can simulate distributed algorithms on a single powerfull computer. (I actually do this a lot myself...)

Finally, you ask for a result which is dependant on experiment. Mind you, I do agree that such results are NOT the rule. However, they are not rare at all either. A recent, fairly popular and (in my opinion) quite elegant empirical study in CS is the observation of phase transitions in NP-hard problems such as SAT. Here is a link: http://www.inf.ed.ac.uk/teaching/courses/propm/papers/400133A0.pdf google scholar with "phase transitions in computer science" will give you hundred more...) Powo

The distinction between deductive and empirical methods seems beside the point to me. This is an article on the natural sciences, not the empirical sciences. The important issue is what is being studied, not how it is being studied. Although mathematical biology uses deductive or computational methods, it is still part of biology, a natural science. Meanwhile similar methods are used in social simulation, not part of a natural science. -- Avenue 00:02, 26 April 2006 (UTC)

Avenue: Yes, I agree insofar as I think natural science is a subset of (empirical) science. I was trying to argue that CS isn't even an empirical science, but if it were, I would certainly deem it outside natural science.

Powo: Yes, good example. My only counter would be to say that that paper makes CS empirical as much as work around the Hilbert-Polya conjecture makes Number Theory empirical. Any thoughts? Tez 00:30, 26 April 2006 (UTC)

Hi again. I think an important difference between CS and math is that CS has a much stronger connection to the natural world, it mainly studies computers, computations and information, which are physical systems or physical phenomena. Also, the engineering aspects of CS and the non engineering (or scientific) aspects of CS are very closely related.Another difference is that although, as you pointed out, mathematics may be empirical or experimental from time to time, computer science is experimental all the time: many scientific CS publications have no analytical evidence of the asserted claims, but only experimental evidence. I understand to some extent the objections to seing CS as a natural science. However, the other extreme: flately declaring wiothout any further comments that CS is not a natural science shades some non trivial science philosophy classification questions. After all, science does not obey the classification rules and the philosophical definition of the scientific method. Science exists, and philosophers model this human activity and categorize scientific practise. Computer science is the latest born of sciences. Not surprisingly, it does not fit quite well in the model. The model is not robust! --Powo 15:58, 28 April 2006 (UTC)

[edit] Scope of natural science

I changed the intro to say "physical things we know about", because we also know about non-physical things (such as mathematics and logic).

Also, I would hope that the article will make a clear and easily-accepted distinction between the physical sciences and other areas of scientific scrutiny. I'm not sure "political science" is anything more than a fancy name for the academic study of politics, but I'm pretty sure that even psychology can be studied with a scientific attitude. --Uncle Ed 20:33, 2 May 2006 (UTC)

While I agree this is an improvement, "physical" is still not the ideal adjective. I believe most people wouldn't think of the following physical things as being studied within the natural sciences: medicinal drugs, weapons, transportation networks, and computers (perhaps - see discussion of computer science above). All of these things involve people in their creation and use. Also, "physical things" might seem less inclusive of biological topics like ecology or evolution. -- Avenue 22:22, 2 May 2006 (UTC)

Wouldn't anything invented by human being fall into the category of technology (or engineering)? Fluid dynamics ought to be a physical science or Natural science but "plumbing" is a trade.

I'm not making any arguments here, just wondering aloud how we should organize some of this knowledge. A buddy of mine has a degree in Library Sciences, maybe he can help. --Uncle Ed 23:56, 2 May 2006 (UTC)

Just a minor comment to say, once again, that I dont think natural should be opposed to artificial, it should rather be opposed (IMHO) to super-natural. Things invented by humans, from this point of view, are not automatically discalified from falling under the scope of natural sciences: it depends if it applies the scientific method or not (which is not the case of "plumbing"!!! ;)) --Powo 09:36, 4 May 2006 (UTC)

Except it seems to me that natural and artificial more accurately reflect the distinction made by the proponents of ID if we accept the premise outlined by Naturalism (philosophy) that the supernatural cannot be studied by the same methods as the natural. Hackwrench 00:19, 2 November 2006 (UTC)

Okay then if "natural" means other than supernatural (like seeing ghosts or Joan of Arc hearing voices), then that gives us one boundary at least. But what about sciences such as psychology? Only behavioral psychology takes a purely materialistic approach. Can science study phenomena which aren't directly observable with electrodes and dials and such? Or does scholarship consist only of "natural science" and "other organized bodies of research"? Sociology, economics, anthropology, can these be "sciences" by any stretch of the meaning? (Surely not natural sciences - but what kind of science then?) --Uncle Ed 12:30, 4 May 2006 (UTC)

Hi. On the nature page, I find this: [...]One approach is to exclude mind from the realm of the natural; another is to exclude not only mind, but also humans and their influence.[...] I think this acception of the wordnature is quite adequate in the context (natural science) we are intersted in. What do you think? --Powo 20:36, 4 May 2006 (UTC)

[edit] Mapping the sciences: scientific adjectives/name of the science(s)

Scientific adjectives is a sub-project of the WikiProject Conceptual Jungle, aiming at making an overview in a table of scientific adjectives and the various branches of (the) science(s) and qualify them by discussing them, improving the Wikipedia articles and make clear the interlinkages. Please feel free to add your contributions to the table. Best regards, Brz7 12:45, 3 January 2007 (UTC)

[edit] This Article is Forever Consigned to be a Quagmire.

Even though I deeply believe there will never be even a remote consensus of what the subject of the article is, one way to try to go forward, would be to acknowledge upfront in the article that various people through the ages have meant, and still do mean any number of things when they say "natural science". I know the very suggestion will make some people very upset, but I can't see you do anything but engage in an eternal tug-of-war on this article otherwise. Just my two cents. -- Cimon Avaro; on a pogostick. 07:36, 17 January 2007 (UTC)

[edit] Expansion of this page

I think this page could be readily developed along the same lines as the Social sciences page. That is, a section of history and sub-sections on the primary disciplines. — RJH (talk) 15:43, 19 January 2007 (UTC)