tag:blogger.com,1999:blog-17232051.post5658875089496397154..comments2024-03-28T03:16:14.104-04:00Comments on Noahpinion: Should theories be testable?Noah Smithhttp://www.blogger.com/profile/09093917601641588575noreply@blogger.comBlogger50125tag:blogger.com,1999:blog-17232051.post-88954069786078360842014-12-28T13:48:53.945-05:002014-12-28T13:48:53.945-05:00Noah, this won't take long and is quite straig...Noah, this won't take long and is quite straight forward;<br />http://weapedagogy.wordpress.com/2014/12/26/is-scientific-methodology-axiomatic/Thornton Hallhttps://www.blogger.com/profile/11402495641975262697noreply@blogger.comtag:blogger.com,1999:blog-17232051.post-48646891944329848242014-12-25T00:12:49.709-05:002014-12-25T00:12:49.709-05:00Noah, are you going to eventually take this post d...Noah, are you going to eventually take this post down like the one about the Spanish Empire?<br /><br />What kind of psychopath imagines that university professors come cheap? How many more taxpayer dollars could we spend educating and employing people not in the military? <br /><br />This equivocation of different uses of the word theory is also one of the flaws that characterize creationist apologetics. Good company, eh?<br /><br />The ignorance among economists as to how science works truly boggles the mind. Thornton Hallhttps://www.blogger.com/profile/11402495641975262697noreply@blogger.comtag:blogger.com,1999:blog-17232051.post-53552957130745035782014-12-24T17:20:36.418-05:002014-12-24T17:20:36.418-05:00"A good example is the Stable Matching Theory..."A good example is the Stable Matching Theory developed by Al Roth and Lloyd Shapley. When this theory was developed, it didn't describe anything that existed in the world."<br /><br />It may not have existed on the your home planet, but it more certainly did exist on Earth - ie any matching that is self-enforcing, such as certain types of job search & recruitment. IIRC, Shapely/Roth use the example of college admissions in their paper. Not only did the problem exist, they were formalizing a solution that existed for at least a decade before they did it ie National Resident Matching Program.<br />A "good example" indeed.Anonymousnoreply@blogger.comtag:blogger.com,1999:blog-17232051.post-49413493057016175942014-12-23T05:29:51.590-05:002014-12-23T05:29:51.590-05:00As DCosmology writes in a blog the author here con...As DCosmology writes in a blog the author here confuses theorems and theories. He should know the difference. <br /><br />On Falsifiability http://shar.es/13Ms6J via @DCosmology<br /><br />Anonymousnoreply@blogger.comtag:blogger.com,1999:blog-17232051.post-61744208881566867482014-12-22T23:31:08.826-05:002014-12-22T23:31:08.826-05:00"When it was made, there existed no phenomena..."When it was made, there existed no phenomena that matched the theory."<br /><br />You are confusing two meanings of the word "theory". A theory in physics makes a claim about physical reality. A theory in math is a body of knowledge about a class of abstract entities: number theory, game theory. These "theories" are very different beasts from, say, "string theory", and for mathematicians, it's an occasional nicety when something in one of their bodies of knowledge happens to map onto something in the real world. Theories in physics are claims about how the world works.<br /><br />So a claim about the way the real world works that makes no testable or usable predictions about what happens when you do things in said real world is very much "not even wrong". Mathematicians don't have that problem: they're playing a formal game with its own rules. String theory does have that problem.<br /><br />David J. Littleboyhttp://www.pbase.com/davidjlnoreply@blogger.comtag:blogger.com,1999:blog-17232051.post-63570040498371430312014-12-22T19:07:58.697-05:002014-12-22T19:07:58.697-05:00TofGovaerts,
In real economy, V = (R - NIC)/Q fo...TofGovaerts,<br />In real economy, V = (R - NIC)/Q for all time periods in NIPA. This can be derived from the basic definition of P Q= GDP. In QTM, V = GDP/Ms at unknown "equilibrium" time period.<br /><br />Thus, we can falsify QTM theory for the time periods which QTM's V value is not equal to (R - NIC)/Q. So far, we have not yet found these two values are equal in any single year from NIPA. Technically speaking, this QTM "equilibrium" time has not came yet. But please don't stay tune!. It may never come in the future.pliu412https://www.blogger.com/profile/06404437429487666914noreply@blogger.comtag:blogger.com,1999:blog-17232051.post-43142917540711062592014-12-22T16:58:35.066-05:002014-12-22T16:58:35.066-05:00Your link to "Stable Matching Theory" li...Your link to "Stable Matching Theory" links to the "Stable Matching Problem". Stable Matching is proposed as a mathematical problem (with motivation drawn from the real world). It is then worked on as a mathematical problem until theorems can be proved; in this case showing that there is an algorithm for solving the problem. This theorem is then testable. You can create input sets for the algorithm, run the algorithm, and see whether or not the resulting pairing is stable. Just because a proof exists showing that the result will always result in a stable pairing doesn't mean you can run the test. In this particular case, it's reasonable to test the implementation of the algorithm for typos.<br /><br />If you can take a real world problem, whether markets or organs, and cast it into the form in which the Stable Matching Problem is defined, then you can use Stable Matching Algorithms to come up with stable matches for your real-world problem.<br /><br />The important thing here is that a theory must make a prediction in order to be useful. If a theory is not falsafiable, then it does not make a prediction. Now, we already have a perfectly good non-falsafiable theory that explains absolutely everything: God. <br /><br />Cesiumhttps://www.blogger.com/profile/02025636403503365433noreply@blogger.comtag:blogger.com,1999:blog-17232051.post-15004091447528430392014-12-22T16:39:11.388-05:002014-12-22T16:39:11.388-05:00"most of the things people are doing in math ..."most of the things people are doing in math departments aren't testable"<br />wft? You should have taken more math classes in college. Test the pythagorean theorem all you want; we've proven you won't find any counter-examples. The Riemann hypothesis is more interesting: we haven't proven you can't find a counter-example, although a large number of tests have been performed.Cesiumhttps://www.blogger.com/profile/02025636403503365433noreply@blogger.comtag:blogger.com,1999:blog-17232051.post-25075674117267803922014-12-22T16:25:12.914-05:002014-12-22T16:25:12.914-05:00Your playing with words, especially in the ambigui...Your playing with words, especially in the ambiguity between theorem and theory. In math departments, you prove theorems. You do not work with theories. A theory is the application of math to reality, and thus must be testable.<br /><br />It's perfectly fine to do math in physics or econ and prove theorems and logical implications. It's a completely different thing to hypothesize that a beautiful mathematical structure has something to do with the real world.Cesiumhttps://www.blogger.com/profile/02025636403503365433noreply@blogger.comtag:blogger.com,1999:blog-17232051.post-87627082793258122712014-12-22T15:57:45.661-05:002014-12-22T15:57:45.661-05:00Leave the math to the mathers. Leave the math to the mathers. J. Edgar Mihelichttps://www.blogger.com/profile/08735224229199089531noreply@blogger.comtag:blogger.com,1999:blog-17232051.post-69697383205748990222014-12-22T13:54:52.001-05:002014-12-22T13:54:52.001-05:00First, it is science that insists on experimental ...First, it is science that insists on experimental test-ability, not math. Math is not science, rather it is a logical relationship exercise that has proven valuable to use in science and everyday life.<br /><br />So, if science is not testable, then how is its validity checked? That is, what would then differential science from philosophy and religion? It is the quality of experimental test-ability that makes science the unique expositor of physical truth.<br /><br />The inverse is also true, that if something is not testable by experimental methods then it is not science.David W. Vielhttps://www.blogger.com/profile/16559249540276639324noreply@blogger.comtag:blogger.com,1999:blog-17232051.post-50674561588461380852014-12-22T13:01:23.385-05:002014-12-22T13:01:23.385-05:00You don't define "testable" here. Is...You don't define "testable" here. Is the standard for "testable" just "consistent with data" -- in which case there's no claim that "testable" implies the ability to differentiate between the quality of different theories many of which are likely to be consistent with the same data? Or is the standard for "testable" that one can actually test the causality implied by the theory?<br /><br />This to me seems the more relevant problem.Nonymnoreply@blogger.comtag:blogger.com,1999:blog-17232051.post-83878584624748025792014-12-22T05:10:20.280-05:002014-12-22T05:10:20.280-05:00On Falsifiability http://shar.es/13Ms6J via @DCosm...On Falsifiability http://shar.es/13Ms6J via @DCosmologyAnonymousnoreply@blogger.comtag:blogger.com,1999:blog-17232051.post-41285607259558068632014-12-22T04:56:49.294-05:002014-12-22T04:56:49.294-05:00Thanks!Thanks!Noah Smithhttps://www.blogger.com/profile/09093917601641588575noreply@blogger.comtag:blogger.com,1999:blog-17232051.post-17103059061147211662014-12-22T02:53:08.749-05:002014-12-22T02:53:08.749-05:00I liked your collection of links, your "philo...I liked your collection of links, your "philosophy of science" posts (especially related to economics) are my favorite. Keep it up!Anonymousnoreply@blogger.comtag:blogger.com,1999:blog-17232051.post-77537970502980901862014-12-22T01:45:49.874-05:002014-12-22T01:45:49.874-05:00Ben,
Thanks for some sanity here. There is a lot...Ben,<br /><br />Thanks for some sanity here. There is a lot of reading by the author and his reporting on the reading is good; comprehension of what was read is, well, I better not grade! It might be still not too late for the author to go back to Physics. The issue is very simple: string theory is not a theory in the classical sense (per your link). However, it does extend our thinking to the smallest energy level of well proven physics (Planck energy), and how this energy may manifest as a string that may be closed, open, twisted stretched etc. This formulation using math predicts how the particles and stuff can come about from such construct. Testing physics of small requires bigger and bigger machines and it is likely that such machines may not be feasible for a long time, except that cosmos itself may provide the data...<br /><br />Someday, Noah will realize that there nothing scientific about economics - it is more like Alice in Wonderland.KVnoreply@blogger.comtag:blogger.com,1999:blog-17232051.post-21668989749507999002014-12-21T22:03:13.597-05:002014-12-21T22:03:13.597-05:00Re: "most of the things people are doing in m...Re: "most of the things people are doing in math departments aren't testable"<br /><br />That's not true at all. They are testable for internal consistency and for consistency with the rest of mathematics in whole or in part. Even something as problematic as the axiom of choice gets its day in court as the usefulness, necessity and weaknesses of its various strong and weak forms are tested against what we know, don't know and can't know.<br /><br />Mathematical concepts are also testable by experiment, but for mathematicians that is by calculation. For example, the planar four color theorem was proven by exhaustively exploring a set of graphs and their properties. The experiment has since been repeated and optimized, just as various physicists and biologists have repeated classic experiments over the years.<br /><br />No, they aren't testing the Pythagorean theorem by drawing and measuring triangles, but mathematics is built on testability.<br /><br />P.S. Giancarlo Rota in 'Indiscreet Thoughts' pointed out that one common method of testing a theory is to consider some of its theorems to be axioms and see if one can prove its axioms as theorems. Mathematics is about building logical linkages.<br /><br />P.P.S. Fermat's last theorem was proven by demonstrating a logical linkage between modular forms and elliptic curves. Mathematicians are always testing things. It is one of the few ways they know to move forward.Kaleberghttps://www.blogger.com/profile/05283840743310507878noreply@blogger.comtag:blogger.com,1999:blog-17232051.post-4883833389724006282014-12-21T18:48:52.083-05:002014-12-21T18:48:52.083-05:00It's clear from the discussion that nobody got...It's clear from the discussion that nobody got my point.<br /><br />There are postulations that aren't testable because too many moving parts are involved to fashion a controlled experiment. This is the situation for social sciences.<br /><br />And there are theories that aren't testable because they don't relate to the observable universe. This is the situation for metaphysics.<br /><br />These are two totally different issues. Trying to combine them just generates confusion. The Nature editorial is written against proponents of certain variants of string theory who have left the realm of scientific physics and ventured into conjectures about metaphysics, and yet still want their conjectures about metaphysics to be considered part of scientific physics. Nobody’s saying their conjectures about metaphysics have no value. They’re just not science.<br /><br /><br />Tomnoreply@blogger.comtag:blogger.com,1999:blog-17232051.post-44055519626531155642014-12-21T17:19:21.021-05:002014-12-21T17:19:21.021-05:00There's nothing wrong with people doing math. ...There's nothing wrong with people doing math. But to the extend that they're just donig math, they're not doing science hence physics, which is a science.<br /><br />We don't really know how cheap it is, since we don't really know what the opportunity costs involved in nominal physicists and scientists not actually doing any science.<br /><br />And it might keep their math skills sharp, but again, that doesn't mean they're doing science and that they're scientific skills are necessarily being sharpened.<br /><br />Feynman on mathematicians vs physicists:<br /><br />https://www.youtube.com/watch?v=obCjODeoLVwAnonymousnoreply@blogger.comtag:blogger.com,1999:blog-17232051.post-23993984741009622672014-12-21T16:04:06.153-05:002014-12-21T16:04:06.153-05:00To be a bit picky, when Gale and Shapley published...To be a bit picky, when Gale and Shapley published their paper on the deferred acceptance algorithm, it actually did describe something in the world, namely the method used already at that time to match new graduates of medical schools with residencies. They didn't know that, which is perhaps the important point.<br /><br />For those who don't know, the boy-propose version of deferred acceptance begins with each boy sending a proposal to his favorite girl. Each girl who receives more than one proposal rejects all but her favorite, each rejected boy sends a proposal to his second favorite, each girl now holding more than one proposal rejects all but the best, the rejects send proposals to their favorites among those who have not rejected them, and so forth. Eventually each boy has been rejected by every girl he prefers to the one holding his proposal (if there is one) and outstanding proposals are accepted. This pairing is stable because whenever a girl prefers some other guy, that guy is paired with someone he prefers to her.Andy McLennanhttp://cupid.economics.uq.edu.au/mclennan/noreply@blogger.comtag:blogger.com,1999:blog-17232051.post-7744715747698586482014-12-21T13:54:21.622-05:002014-12-21T13:54:21.622-05:00List a single economic theory that is testable in ...List a single economic theory that is testable in real life. Anybody wants to tackle oil price?<br /><br />KVnoreply@blogger.comtag:blogger.com,1999:blog-17232051.post-75242259876759933272014-12-21T12:51:42.474-05:002014-12-21T12:51:42.474-05:00It's all pretty much data dependent now isn...It's all pretty much data dependent now isn't it. How frequently are National data post hoc revised ? As for the fisher equation, this holds for either P (lot's of controversy) and Q. When it comes to V, we have a very volatile variable here as well. I left this so-called equilibrium approach a long time ago, i am on financial markets for 14 years now. TofGovaertsnoreply@blogger.comtag:blogger.com,1999:blog-17232051.post-81522419501760453432014-12-21T12:33:53.405-05:002014-12-21T12:33:53.405-05:00Term “testable” is defined as either verifiable (m...Term “testable” is defined as either verifiable (making it true) or falsifiable (making it false) in logic and computing. <b>Non-fiction</b> scientific theories require testability. Only <b> fiction</b> scientific theories do not need.<br /><br />A scientific hypothesis can be either ALL or EXIST logic statement. Examples,<br />Does GOD exist? Is General Relativity theory correct?<br /><br />Due to our limitations in enumerating all instances in real world, <br />for an ALL statement, it is easier to be falsifiable, but harder to be verifiable.<br />for an EXIST statement, it is easier to be verifiable, but harder to be falsifiable.<br /><br />However, scientists keep open minds before finding any instance to test a hypothesis if it is false or true. They modify the theories if finding a new instance against it. I think this is the way how sciences can make progress in discovering real world knowledge in a consistent way.<br /><br />A valid scientific theory (fiction or non-fiction) means two things (one coin of two sides):<br />(a) So far, we have not found any real world instance against it.<br />(b) All instances we find so far are consistent with the theory.<br /><br />The more instances consistent with a theory, the more valuable the theory is. Unfortunately, many economic theories miss this scientific principle and thus have less value (see my previous comment on economic theories in this topic)<br />pliu412https://www.blogger.com/profile/06404437429487666914noreply@blogger.comtag:blogger.com,1999:blog-17232051.post-74476526712129179862014-12-21T12:08:53.680-05:002014-12-21T12:08:53.680-05:00Oops, that's wisegeek.orgOops, that's wisegeek.orgAlan Reynoldshttps://twitter.com/AlanReynoldsEcnnoreply@blogger.comtag:blogger.com,1999:blog-17232051.post-89339200688480491152014-12-21T12:07:26.310-05:002014-12-21T12:07:26.310-05:00Courtesy of wisgeek.com - A hypothesis attempts to...Courtesy of wisgeek.com - A hypothesis attempts to answer questions by putting forth a plausible explanation that has yet to be rigorously tested. A theory, on the other hand, has already undergone extensive testing by various scientists and is generally accepted as being an accurate explanation of an observation. This doesn’t mean the theory is correct; only that current testing has not yet been able to disprove it, and the evidence as it is understood, appears to support it.Alan Reynoldshttps://twitter.com/AlanReynoldsEcnnoreply@blogger.com