When can we say that a theory is failed?

[needimprovement]

I am interested to know your answer about the question above.

[swansont]

When it doesn't properly predict phenomena within the realm of its application. You either modify or discard it, depending on the nature of the problem.

[lemur]

I don't think theories ever get totally discarded in an absolute sense. Whatever it was that caused them to merit pursuit in the first place will tend to resurface in a new form. The more they failed, the more different they will look when they are resurrected. A lightly criticized theory will be modified and new avenues of research pursued. A "failed" theory will appear to be abandoned but aspects of it can be found in newer theories. A good genealogist of science could trace epistemological continuities across seemingly divergent theories, even when they are in very distinct disciplines.

[Mr Skeptic]

The first way an idea can fail to be a theory is if it is unfalsifiable (equivalent to making no real predictions). An idea that fails at that point would never have been considered a valid hypothesis nor a theory. "God did it" is an example -- it "makes sense" but does not make predictions. Once an idea has survived the initial rounds of attempt to falsify it, and has made the correct predictions, then it gains the status of a theory. At this point, the theory can be set in stone by limiting the range and accuracy to within certain limits, as we have done with Newtonian gravity, treating it as an approximation rather than "the truth". Most people however will consider the theory to implicitly attempt to describe the entire universe with impeccable accuracy. In this case, experiments can be done in newer, more extreme situations, or with greater accuracy, and might show that the theory fails in one of those regards.

 

A theory can also "fail" when a competing theory makes more precise predictions with the same accuracy, or less premises.

 

A failed theory might be either discarded entirely or modified so it correctly predicts within the new data range.

[Darwinsbulldog]

I think it depends.

Suppose I present a hypothesis like:

 

"All vertebrates require haemoglobin".

It sounds good, I can test it on many vertebrates, and it would ring true. But as with all theories, it is tentative. As we expand our knowledge, theories can become false, as in this case:-

 

http://www.eurekalert.org/multimedia/pub/3289.php

 

[Antarctic Ice Fish]

 

But was the theory really a false one? It depends on what assumptions I made. At STP, the theory definitely works in all cases. Vertebrates are too large in all three dimensions to work around the surface area to volume ratio. [small worms and flatworms do it by not getting so large that the cells most distal to the oxygen supply are still only a few cells away from the surface-but in any case, they are not vertebrates]. Vertebrates compensate for their small surface area to volume ratio by having gills and/or lungs. This increases their surface area because although lungs and gills are "inside" topologically they are still on the surface of the organism. They also have a circulatory system which further reduces oxygen gradients, thus preventing anoxia in the deep tissues.

So perhaps my new hypothesis should read:

 

"All vertebrates require haemoglobin in STP conditions. [sTP=Standard Temperature and Pressure].

 

Obviously, one can't put everything into a model, or it would be as complex as the system it is trying to emulate. So, almost by definition, every theory has to be "wrong" in that they are approximations. Of course, well thought out models have the least mistakes. Well tested models that are true under a wide range of conditions are more parsimonious than models that only work in more restricted conditions.

 

In Newton's day, they could only be accurate to one part in a thousand, so it is doubtful that he could have ever formulated Einsteinian relativity. Today, physics and sciences genrally have better measurement techniques and instrumentation that is accurate in some cases, to dozens of decimal places. this surely affects the quality of the models we can produce.

[CharonY]

That is why Swansont said:

When it doesn't properly predict phenomena within the realm of its application.

(my bold).

A theory is not thrown out without a context.

[redsaint182]

I am interested to know your answer about the question above.

 

I think a theory is failed, or falsified, when you look at the predictions it makes, or its attempts to describe reality, and then make an empirical observation which falsifies one of those predictions, then the theory, in current form, has failed. Now, this observation may call for one of two things. You can either revise the theory to fit the empirical observation, or you must scrap the theory altogether. But in either case, if that observation is made and is correct, the theory in whatever form it was in could be considered wrong (I think.) Also, it is good to keep in mind that a theory must first be a hypothesis, and I think a valid hypothesis must make testable predictions about reality/must be falsifiable, in order for it to be considered valid to even have a chance at progressing to the "theory" level.

[needimprovement]

Thank you for the excellent responses. Is there a specific timeframe for a theory to be proven (rationally) in order to be known as true?

[swansont]

Thank you for the excellent responses. Is there a specific timeframe for a theory to be proven (rationally) in order to be known as true?

 

Not really. There was some aspect of Brownian motion that Einstein predicted would never be confirmed, since he could not foresee the advances in technology, which were verified about a hundred years after has solution was published. Darwin lacked a mechanism to explain important details of evolution (DNA) which took almost that long to discover. Bose-Einstein condensates were predicted in the 1920's and only realized 70 years later. However, there were other tests which confirmed the validity of evolution, and the other examples are really subsets of theories, which likewise had other methods of verification.

 

A good test might be how long we entertain string theory in the absence of experiments, before it becomes little more than a novelty.

[ewmon]

Discrepancies typically exist with any scientific theory — or law, for that matter. I think most people will tentatively accept a theory (or law) until another comes along that better fits the data ... even when there's obviously something that doesn't fit with the current theory/law.

[Mr Skeptic]

Thank you for the excellent responses. Is there a specific timeframe for a theory to be proven (rationally) in order to be known as true?

 

However long it takes to test the theory within the range and accuracy you want to consider it to hold true within.

[Darwinsbulldog]

Discrepancies typically exist with any scientific theory — or law, for that matter. I think most people will tentatively accept a theory (or law) until another comes along that better fits the data ... even when there's obviously something that doesn't fit with the current theory/law.

 

Laws are different from theories. Theories are composed of model, prediction, and explanation. Laws are mathematical relationships which do not change.

[ewmon]

Yes, I appreciate the difference between theories and laws (and I have argued the same elsewhere), but historically, some laws have been improved upon, and I suspect that we have not seen the end of it.