The headline as it appeared in the Times:
The headline as Popper would’ve written it:
Gravitational Waves Detected, Not Falsifying Einstein’s Theory
The headline as it appeared in the Times:
The headline as Popper would’ve written it:
Gravitational Waves Detected, Not Falsifying Einstein’s Theory
FYI, I was reading up on it a bit and saw this:
“We use the Stan and emcee Markov-Chain Monte Carlo samplers (Foreman-Mackey et al. 2013; Stan Development Team 2015b,a)”
https://dcc.ligo.org/P1500217/public
He would have said ‘corroborated’, but the truth is that the recent use of gravitational wave astronomy wasn’t testing GTR, but learning about colliding black holes. My take (as an outsider) is that they used (long time) severely tested aspects of GTR in order to explore this relativistic phenomenon via gravity waves, and what makes it a big success story is that for the first time they have a reliable instrument for learning about such domains. Since all viable relativistic theories predict gravity waves with certain properties, it’s legitimate to make the assumptions needed to use gravity waves as a tool for probing. It’s redolent of Ian Hacking’s (entity realism): we have evidence of the reality of an entity when we use it to learn about other, unknown phenomena.
Suppose if the chirp hadn’t been heard: Did two huge black holes not collide or was Einstein wrong? My guess is most would continue to assume that GTR is correct and that the black hole collision never happened.
Did two huge black holes not collide or was Einstein wrong?
I may be wrong, but a third option could be that LIGO isn’t sensitive enough. If I’m not mistaken, there have been multiple previous gravitational wave predictions that weren’t detected, and, in at least some cases, the instrumentation was (thought to be) at fault.
Yes, but there’s a joke to be made! So what if the joke relies on a straw man version of Popper!
Popper_0, as Lakatos would say.
Well played, sir.
Gravitational wave astronomy absolutely tests GR — in particular it is sensitive to strong-field GR which is not particularly well constrained using current measurements. The biggest test in on the overall rate of gravitational wave observations, but this will have to come later once enough candidates have been found (in particular, had no gravitational waves been observed by Advanced LIGO in the next few years something would be seriously wrong). But these observations can also be used to test other aspects of GR immediately, for example GW150914 has already been used to put a new constraint on the mass of the graviton.
The coalescing binary waveforms used in the modeled transient searches are in no way “severely tested” prior to these events beyond the 0th order term, which was probed by results from timing pulsars in compact binaries. Also testing GR and learning about black holes are inextricably linked, so if you learn about the latter you will be able to test the former. And any deviation from GR will have consequences for what the instruments’ see from a binary black hole system. The two supplementary parameter estimation papers drive home this point in multiple investigations.
It’s great that now we know the machines as is are sensitive enough to get a certain class of signal with beautiful clarity and they will only get better with time. But that doesn’t mean we can ignore the host of things learned about this particular source and the theory of gravity that describes its dance of death.
In “logic of the scientific discovery” Popper actually suggested for this situation: the phrase “Gravitational Waves Detected, Corroborating Einstein’s Theory” rather than “Gravitational Waves Detected, Confirming Einstein’s Theory” ;)
IEHO Popper is not what he is cracked up to be in junior high school science. There is a long history of experiments that failed to detect gravitational waves, including LIGO runs. Did they falsify general relativity. No, because GR provided a wonderful coherent model which explained a large number of observations and was accepted by relevant experts. It is a coherent, consilient theory accepted by a consensus.
Kuhn’s frame of normal science is a much better model for science and scientific progress
Eli:
I’ve written in many places on Kuhn, Popper, Lakatos, and how they relate to statistics and science. You can start by searching this blog for those three names, or by reading my paper with Shalizi. Short answer is that I think Popperism has valuable lessons, and I take the Lakatos tack of treating Popper’s writings as a resource rather than a bible. I’ll also say that I found Kuhn’s writings very irritating when I first read them, but now I think of Kuhn’s perspective as part of the big picture too.
Mr.Gelman you should get Rafe Champion’s guides to popper as he does a great part of sythesizing his corpus/affiliates and also takes into account his postscript to the logic of scientific discovery, which was cut from 1k pages to what was published.
http://www.amazon.com/s/ref=nb_sb_ss_i_1_12?url=search-alias%3Dstripbooks&field-keywords=rafe+champion&sprefix=rafe+champion%2Cstripbooks%2C225&rh=n%3A283155%2Ck%3Arafe+champion
Eli I think you should purchase/read it before insulting Popper. It contains a rebuttal of many myths of Popper.
I should add he is going to double the price soon so the time is ripe.