...subject to our assumed framework and chosen prior. People always forget that Bayes theorem only marginalizes over the unknown parameters. It assumes the model is otherwise given (true). You can only marginalize over the known-unknowns, not the unknown-unknowns.

People like to slip assumptions into their models where the uncertainties are not accounted for. The reason why everyone trusts the LIGO result is that we have an extremely high degree of confidence in general relativity.

More generally, what's even the point of placing a confidence level on a non-falsifiable claim? What utility does it serve?

For context: This reply was written when the title on here was "With 95% confidence, we expect our Universe to last more than 10^58  years"

>> ...subject to our assumed framework and chosen prior.

This is usually implicit in such analyses.

>> More generally, what's even the point of placing a confidence level on a non-falsifiable claim? What utility does it serve?

It's not about falsifiability in the sense that someone should go out and measure the predicted lifetime of the universe experimentally.

The point of such lifetime calculations (and their associated confidence levels) is more as a test of the underlying theory - the Standard Model of particle physics. If this theory would make the prediction that the lifetime of the universe should be shorter than the observed lifetime of the universe, then clearly there is something wrong with the theory, or with the way the lifetime has been computed, since clearly we are still here to make that calcuation. Usually, this is seen as a sign that there should be some new physics - i.e. physics that is not described by the standard model in its present form - that should enter at an energy scale relevant to the calculation. Since this we do not know about this physics yet, the calculation is necessarily incomplete, which would explain finding a "too short" lifetime of the universe, while with a more complete theory one would (hopefully) find a result, that explains observed lifetime of the universe.

The fact that the paper finds a "long enough" lifetime of the universe, could be seen as an indication that the theory is able to make predictions at energy scales relevant to this calculation.

> The reason why everyone trusts the LIGO result is that we have an extremely high degree of confidence in general relativity.

No, that's not true. The reason everyone "trusts" the LIGO results is because we believe the people running the experiment are trustworthy. The results have nothing to do with GR except insofar as they were designed to test a heretofore untested prediction of GR (which prediction they have confirmed).

And the reason we want to test a prediction of GR is that we want to try to falsify it, because that is the way progress is made in science.

I think you may have meant to say something about how we cannot know for certain that GR is true, that that's right, we can't. But it's not like Einstein just pulled GR out of a hat. It's the only known solution to the problem that fits all of the known data. That doesn't prove it's true, but it is the best explanation we have so we might as well carry on as if it's true until someone proves otherwise or comes up with a better idea.

My impression was that they made advanced predictions what a gravitational wave observation would look like from a binary black hole merger, and then they measured pretty much exactly that. This is why we believe the machine worked.

The other reason we believe them is that they (together with VIRGO) found a neutron star merger, and that was confirmed by many other observations in many different spectra.

I don't think that LIGO was designed (only) to test GR. It was also set up to be an astronomical observatory. (I don't know which goal was considered more important...)

Why do you think those are different goals?

Because in the astronomy part, we're assuming the GR is correct, and using it to watch astronomical events in a way we couldn't before.

For example, the detected neutron star merger tested that gravitational waves travel at the speed of light, since we were able to also detect the fast gamma ray burst from it. That's testing GR. But it also showed that such fast gamma ray bursts come from the merger of two neutron stars. That's not testing GR.

>> More generally, what's even the point of placing a confidence level on a non-falsifiable claim?

One can answer questions like: Is it worthwhile for people to attempt to colonize the galaxy? When things like getting to other stars will take lifetimes, and getting across a galaxy 100M light-years across may take more time than has already passed, one needs to have some idea if there is even enough time left for such undertakings. Of course the lifetime of the galaxy will probably be less than 10^50 years, at least it puts the universe as a whole into better context. And really, in spite of any crazy ideas just putting ourselves into context seems to be important to a lot of people.

We might be able to generate hypotheses using it as a premise?

> Furthermore, since the lifetime is finite and the Universe infinite, there is likely a bubble of true vacuum already out there, far away. It is sobering to envision this bubble, with its wall of negative energy, barreling towards us at the speed of light. It seems the long-term future of our Universe is not going to be slow freezing due to cosmic acceleration but an abrupt collision with one of these bubble walls.

It seems like the more interesting calculation is how far away one of these bubbles is likely to be.

I think the most solid argument in favor of the quantum multiverse is along the lines of "The universe is virtually certain to fall into a lower-energy quantum vacuum state. That we have not observed it to be the case means either that we have been astonishingly lucky, or that there is a selection process going on. There is such a selection process; we can only witness futures in which the entire past lightcone of the universe has not collapsed; despite the near-certainty that such has occurred there are still far more possible universes than that." In short, the quantum multiverse argument can be run backwards in time, too; a given quantum system will observe a superposition of all possible past states of the universe as it evolves forward in time, and we can only see the ones in which we still survive.

Now, I still don't believe in the quantum multiverse. We have not proved to my satisfaction that lower-energy quantum vacuum states are inevitable, on the grounds that we don't know enough yet to know why it hasn't happened. (i.e., there may be other effects keeping it from happening, or perhaps there are reasons why it actually can't happen such as the lower state being non-viable for some other reason, etc.) But in my opinion, this is still the most solid argument in its favor; if it can be established that our current universe with near-certainty should evolve into something that we can't live in or observe with a bubble expanding at the speed of light, it's going to take something equally potent to counteract that.

Or, in other words, under this theory, for every split second of your entire life, vastly, vastly more of your worldlines are getting lightspeed-annihilated than not. %99.999... "keep writing 9s as long as you like" of them. But the destruction is so total and instantaneous that you can never observe this fact, so you go on happily living in the 0.00...01% of the remainder, happy as a clam. Again, this isn't my own personal belief, but it's a theory.

> The universe is virtually certain to fall into a lower-energy quantum vacuum state. That we have not observed it to be the case means either that we have been astonishingly lucky, or that there is a selection process going on.

If that is the most solid argument in favor of the quantum multiverse, this paper is a disaster. See Eq. 6.28: the probability that we should have seen a bubble by now is < 10^-516.

When I quoted that argument, I was not saying it was 100% accurate; after all, I don't believe it myself. If further work comes along to prove that the collapse into the lower state is not highly probable, then the argument collapses. Again, not accepting the multiverse anyhow, that the best argument in its favor collapses would not particularly bother me.

Of course, it would not constitute evidence against, merely a lack of evidence for. Now, some people with a insufficiently deep understanding of probabilities and such will say those two things are completely different, but they aren't; lack of evidence for is indeed a form of evidence against, especially if one is actively searching for the evidence. It just isn't very strong, and it certainly isn't a proof.

'infinity' is more of a logical construction than a physical one.

Going with the paper's result: with 95% confidence, more than 10^58  light years. :)

I think that was about the entire Universe, not our local part of it.

The result reported by this paper is decay rate per unit volume (see Eq. 6.26). That tells you how long it takes for a bubble to nucleate. After that, it expands at the speed of light. How long does it take for a bubble expanding at the speed of light to destroy an infinite universe?

So it would be multiple bubbles spread throughout that would wipe out an infinite universe.

... which is expanding. So now you have a percolation problem: what grows faster, the bubbles or the universe? The latter is also thought to be expanding at an accelerating rate, which leads to funny scenarios like this: https://arxiv.org/abs/1503.08130

So "get out before the bubble pops" applies all the way up to cosmic scales?

It’s worth pointing out that we don’t know if we live in an infinite or finite universe, so that’s an assumption made by the authors.

I think your question actually raises a key point ... "one of these bubbles". How accurate is their prediction that these bubbles aren't common? And is there somewhere nearby that might be hiding one?

On their time-scale, Musk's goal of becoming an interplanetary species seems like a good near-term hedge. Being able to traverse our own galaxy would be a good second step but exponentially harder. People have discussed the almost impossible task of moving between galaxies (without something like a worm-hole).

So, assuming the multi-verse theory is correct, I guess we humans need to look into how to get to another universe before this big cataclysm. (It seems pretty easy on "Stranger Things")

Physicist here: I hardly understand this paper so don't worry if neither do you.

I was about to say, this headline sounds like one of those fake papers generated by a machine learning algorithm, like SCIgen (https://pdos.csail.mit.edu/archive/scigen/).

I don't understand what is the point of your comment, I am a physicist and I understand this paper but this does not mean that others should worry if they don't.

I think it's public commentary on the ease of consumption of these papers on HN on a wider scale than just physicists.

In your opinion, what does it say about HN that people have the need to say "this is very complicated, don't worry if you don't get it"?

It says that people here expect to be able to understand things. "I'm not smart enough to understand physics" is not a normal self-image on HN.

Bingo.

Nice. Can you ELI25 it?

I only skimmed it. I don't understand it now, but I guess that in a week I could get a good enough understanding if I skip the most difficult parts. So my possible wrong ELI25:

The vacuum can transform spontaneously in another type of vacuum (like liquid water transform into ice at low temperature). Lucky this is very improbable because otherwise we would be all dead, and the transition is due to quantum tunneling.

To estimate the probability of the transition it is necessary to calculate the contribution of the virtual particles in the loops in the Feynman diagrams. A naïve calculation has too many infinite values to be useful.

In this article they calculate the contribution of multiple virtual particles, in particular when the Feynman diagrams has many loops. They have a nice pattern and somehow they cancel (most?) of the infinite values.

With this new calculation they can estimate how long we should wait until this tunneling effect between the two vacuum destroy the universe and luckily it's a long time. (Perhaps other thing can kill all of us before :).)

For this calculation it is particular important the mass of the heavy particles.

(Linkbait title: "Scientists say that God's particle may open portal to universe doom at any moment". Sorry. I couldn't resist.)

As I said earlier, this short version may be wrong. Any corrections or missing important points are welcome.

Talk about fine-tuning! How is it we are in the metastable area?

The anthropic principle, had we been in an unstable region, we wouldn't be able to talk about it now.

A completely stable universe might also have some issues that are not immediately clear.

That's actually a very good question. Moreover, if inflation really happened in the early universe, its almost impossible that the Higgs would have remained in the metastable vacuum during that period. Thus only fine-tuning wouldn't be enough to explain our existence.

More on this: https://arxiv.org/abs/1505.04825

Some systems are self-tuning, perhaps that's what's happening.

https://en.m.wikipedia.org/wiki/Self-organized_criticality