5
$\begingroup$

Discovery in Astronomy/Astrophysics (of astronomical objects) vs discovery of physical phenomena in experimental Physics (as such) - do they differ in required burden of evidence (with the notion, that astronomy/astrophysics deals with real physically existing astronomical objects, while the aim of experimental physics per se is mostly to verify rather somewhat abstract concepts of theoretical physics)?

Perhaps answering my question in some definite way would require some analysis with historical perspective and consideration of precedences ...

This generalized question of mine was motivated by more specific case featured in https://physics.stackexchange.com/q/236107/25575

$\endgroup$
  • $\begingroup$ Isn't astronomy a part of physics $\endgroup$ – user5402 Jul 15 '19 at 15:38
6
$\begingroup$

There is no central authority in science. There is no council that sets the standards. The criteria for a discovery are the same: You publish your findings, and your peers accept your results.

There is the 5 sigma rule in particle physics. Perhaps you were thinking of this. But that is not an official rule, instead it's a convention among particle physics. And it's not applicable to every field.

Things are never cut and dry. What does "Peers" mean. What if only most are convinced. Does it count? Discoveries are usually judged in retrospect. After the dust has settled and the debate s are over.

//edit

Why is the bar set so high in physics? When "discovering" a particle there is no way to "see" it, instead you observe a mass of data and see a statistical discrepancy. Cern produces masses and masses of data, and it is sifted for anything unusual. And with so much data, the chance of seeing something unusual is actually quite high. (imagine searching for repetitions of '9' in the digits of pi - if you search far enough you can be sure to find a string of 6 9s even though the chance of a random string having 6 9s is very low)

With a mass of data, and the opportunity to repeat experiments it makes sense to set the bar very high.

Compare with an observation, such as the "chirp" at Ligo. We can't repeat it at will, the data is there: something happened that is consistent with a black hole merger. No other theory has been proposed that can explain the observations. The observation is less dependent on a statistical finding after running multiple experiments, but a single direct observation.

In fields in which a result could be explained by chance, then a statistical analysis is done, and published. This contributes to the quality of a finding, and so the number of people you will convince. And convincing your peers is the only criteria that counts in the end.

$\endgroup$
  • $\begingroup$ Good point about the $5\sigma$ rule in physics. This actually distinguishes physics a bit from astronomy, as in the latter field, people tend to happily publish a $3\sigma$ result (although they may call it "tentative evidence"). $\endgroup$ – pela Mar 5 '16 at 14:29
  • $\begingroup$ @pela - thanks for the comment - this is amazing, since, IMHO, astronomy/astrophysics deals with real physically existing astronomical objects, while the aim of experimental physics is mostly to verify (somewhat abstract) concepts of theoretical physics ... $\endgroup$ – Alex Mar 5 '16 at 14:42
  • 1
    $\begingroup$ Would it be worth a comparison with psychology when p=0.05 is "significant" (equivalant to 2sigma) and p=0.1 is "Highly suggestive and tending towards significance") $\endgroup$ – James K Mar 5 '16 at 16:42
  • $\begingroup$ @JamesKilfiger - why not ... I wonder if there already exists some research/publication, covering this issue across variety of science domains. $\endgroup$ – Alex Mar 5 '16 at 16:49
  • $\begingroup$ @Alex: The number of sigmas needed to take a result seriously does depends on the type of observation. But for instance, if one detects something which would be unusual — like a galaxy breaking the distance record — then a $3\sigma$ would make most people happy. After all, this still means that "we're 99% sure it's there". Then better follow-up observations may confirm or reject the observation. And as James Kilfiger mention, we're still doing better than most social sciences :) $\endgroup$ – pela Mar 5 '16 at 17:21

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

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