Chris Peterson wrote:Nereid wrote:Applying (Newton's) theory of gravity to observations of Uranus produced inconsistencies; not one but hundreds of observations 'contradicted' the Newtonian theory, ergo it was 'proven wrong'. You, dear reader, know the rest, don't you?
Applying (Newton's) theory of gravity to observations of Mercury produced inconsistencies; not one but hundreds of observations 'contradicted' the Newtonian theory, ergo it was 'proven wrong'. Ditto.
A great many, very good, observations of certain 'beta decays' 'contradicted' the 'theory' of conservation of energy, in the 1920s and 1930s; ergo it was 'proven wrong'.
I'm afraid I don't understand how these examples argue against the idea of falsifiable theories.
I'm pretty sure they don't ... nor were they intended to!
The context of my comments was a comment on bystander's "
With only one observation that contradicts a theory, that theory can be proven wrong".
That's one way to summarise a version of falsificationism (called 'naive falsificationism') that is, unfortunately I suspect, all too common. Note that Popper himself was far too good a thinker to have ever proposed this version (except, perhaps, as a foil).
Let's explore this further ...
The observations of Uranus did not in themselves falsify Newton's theory of gravitation. What they did was present a question: is Newton wrong, or is their another gravitational influence we are unaware of? Given the existing strength of evidence for Newton, the reasonable conclusion was the latter.
Indeed.
But how do you decide - ahead of time - what constitutes a "reasonable conclusion"? And how do you measure "strength" (of evidence)?
Clearly, to be a good (or at least a useful) theory of science, falsificationism needs to have some caveats appended, some bells and whistles added.
Of course, had a search continued for long enough without finding anything, this would have started shifting support away from Newton. But the observation alone was not (and could not) be sufficient to falsify Newton's law.
How long would "long enough" have been? 1850? 1900? 2050??
Another aspect: perhaps Uranus is (was) "an anomaly"? Everything else moved just as Newton ordered, both within the solar system and beyond. Should the textbooks have then said "Newton's universal law of gravitation (except for Uranus)"?
And then there's this: it wasn't just one or two observations (of the position of Uranus), it was already hundreds (by the second half of the 1840s), and very precise ones they were, and independent (different observers, different observatories, etc).
The observations of Mercury did, in fact, correctly falsify Newton (although that had largely already been done). Of course, they served to bolster the support for GR, which extends Newton under particularly conditions. That is, Newton wasn't actually falsified. Rather, the conditions under which the law is applicable were more accurately defined.
The history is a little murkier ... the observations showing inconsistency with Newton were both extensive and precise
several decades before Einstein was concluding his work on GR. And 'the Neptune solution' had been tried ("Vulcan"), and at the time it seemed to work, sorta.
A couple of follow-on questions, if I may:
* what do you mean by "
although that had largely already been done"? AFAIK (as far as I know) the anomalous advance of the perihelion of Mercury was the only well-established observational or experimental result that 'falsified' (to be anachronistic) Newton, before publication of GR
* in terms of
content, GR couldn't be more different than Newtonian gravity! The latter is an 'action at a distance', an (instantaneous) force; the former is geometry. Sure you can show that GR 'reduces to' Newton in the appropriate limit, and so that is good enough for that region of parameter space. But to say that "
Newton wasn't actually falsified" surely adds some rather sweeping restrictions to "falsify" doesn't it?
The observations of beta decay behavior presented a similar question to what I posed above: is conservation of energy wrong, or is there another explanation? Of course, it drove the development of additional theory and did not invalidate conservation of energy.
These examples don't argue that theories can't be invalidated, only that the conditions of falsifiability need to be carefully defined. Observations can be misinterpreted, and that's something that has to be guarded against. When an observation appears to contradict a theory (especially a very strong theory), it is important to make sure that the observation has been interpreted in the only possible way.
Indeed.
But doesn't that raise rather knotty questions? Such as "what constitutes misinterpretation of a theory"? and "to what extent can the 'conditions of falsifiability' be defined beforehand"?
Of course, it is easy, almost trivially so, to look backward and write a history of the development of one or other theory (in physics), using falsifiability as a guiding principle. However, how can - could - this concept be applied, in any useful way, when you're in the middle of things?
Take a (just barely historical) example: the Pioneer anomaly.
If we had a time machine, we might read in (future) textbooks of physics that this was pivotal in the discovery of a theory of gravity that replaced/extended GR. I.e. it 'falsified' GR.
Or we may read, as a mere footnote in just a fraction of those (future) textbooks that this 'so-called' anomaly turned out to 'nothing more' than a combination of poor modelling and noisy data. I.e. it did not 'falsify' GR.
Or ...
The point is, how can we tell - today - what the (future) status of 'the Pioneer anomaly' is?
Reading the history of 'the solar neutrino problem' is quite sobering (or uplifting, or ...), when considered with 'falsifiability' in mind.
If you examine well crafted modern theories, they include- from the beginning- cases that would falsify these theories. That is quite different from accidental observations that appear to invalidate some theory. Those need to be evaluated very carefully.
I'd like to explore this in more detail later; for now I'll merely note that whatever 'falsify' is, it has essentially no value when it comes to actually doing (modern) science ...