2
$\begingroup$

An observation could validate the predictions made by a previous theory. If something unpredicted is observed, then a new theory which is compatible with the observation should make predictions which later are confirmed by new observations.

But what if no such new predictions are possible to make? What is the scientific status of a theory which after the fact is only compatible with all data, but which cannot make any new predictions? If it would be considered weak, does this mean that discoveries through observation can actually destroy science?

Is this a valid argument for delaying the construction of better observatories until theories have matured given all existing data? Kind of comparable with the forward planetary protection argument which says that human spaceflight to Mars should wait until it is certain that there's no life there which could go extinct when it meets Earthly life.

$\endgroup$
6
  • $\begingroup$ I think even though science is renowned for creating theories that make bold predictions of events before they are ever observed, it is still fundamentally an experimental/observational based field. Just think all the way back to Newton, he didn't predict gravity first, he observed it first when the apple fell on his head. Also I agree with the Mars argument but it would be impossible to ever rule out life on Mars because we haven't even found all life on Earth yet. $\endgroup$ – Dean Feb 21 '16 at 13:29
  • $\begingroup$ Data trumps theory every time. No need to stop our ears down until theory catches up. $\endgroup$ – Wayfaring Stranger Feb 21 '16 at 15:47
  • 1
    $\begingroup$ Agreed. Astronomy does not follow the "pure" scientific method. It is very often the case that data leads theory. $\endgroup$ – ProfRob Feb 21 '16 at 19:45
  • $\begingroup$ @WayfaringStranger Well, Relativity is the most perfect example of the opposite.If General relativity was formulated for the first time today when all its predictions have already been observed (let say so here), would it be a less scientific theory? $\endgroup$ – LocalFluff Feb 21 '16 at 22:10
  • 4
    $\begingroup$ I'm voting to close this question as off-topic because it is only tangentially related to astronomy, and seems to be primarily opinion based. $\endgroup$ – James K Feb 21 '16 at 22:22
2
$\begingroup$

The question leaves me confused on how to interpret it. As some of the comments have noted, observations have routinely predated a theory that (accurately) predicts them. In astronomy and cosmology this has tended to be the main driving force behind seeking new theories. Yet I sense a more subtle question concerning how scientists would view a theory that predicts everything so far, but nothing new.

Ultimately I think there is a confusion concerning the typical mantra for adopting a new theory as the main theory for physics/science/whatever: if a new theory doesn't make any new testable predictions, and so can only reproduce what existing theories already predict, then preference is given to simpler and existing theories. This has nothing to say about a single theory. It's strictly about multiple theories being compared to each other. It does not imply that a theory, viewed entirely on its own, that makes no new predictions is undesirable. It simply means if we're going to replace theory X with theory Y, then theory Y better predict more stuff than X does. It doesn't care if X does or doesn't predict testable things that have yet to be proven/disproven.

$\endgroup$
2
  • $\begingroup$ Can't one make up (very) complicated theories ad hoc which fit existing, otherwise unexplained, observations but which make no new predictions? $\endgroup$ – LocalFluff Feb 21 '16 at 22:21
  • $\begingroup$ @LocalFluff That's a lot harder than you think, but yes. Existing unexplained data would support the new theory that fits it. Are you trying to envisage a theory that basically just bootstrap unexplained observations onto existing theories? Where you just sort of declare by fiat that it gives the unexplained observations without an actual mechanism? "Predicting the observations" means it should fall out as a consequence of the theory in some fashion, rather than by a fiat. $\endgroup$ – zibadawa timmy Feb 21 '16 at 23:04

Not the answer you're looking for? Browse other questions tagged or ask your own question.