What I think you cannot read elsewhere are the free thoughts I had as I listened to the announcement by the VIRGO collaboration. So maybe this may be a different kind of contribution, and of some interest to you.
I arrived a bit late to "Aula Rostagni", the main auditorium of the Physics department in Padova. The large room was entirely full of professors, researchers, students, and administrative and technical personnel, and one of the first slides of the speaker was already up.
I was expecting a longish introduction, but there was none. So I was rather surprised when the speaker showed a slide with the raw data from the two antennae - two parallel wiggling noise signals, interrupted at some point by two matching larger and coherent wiggles. As I stared at the "evidence", presented in pristine form (without an overlaid fit, or any manipulation), I could not help thinking that this was indeed something. As somebody had circulated on twitter a few days ago, "the signal is strong enough that it is clearly visible by eye".
I was then also startled to hear that they collected that signal _before_ they started officially the run of data taking. They got it little after they turned the things on! This got me thinking that they might have had to tweak their pre-defined procedures for declaring data good for analysis... A second thought I then had was "Okay, you've explained that the pre-upgrade VIRGO was three times less sensitive, so it could not have discovered a signal like this one; but then, why didn't VIRGO produce several less-than-discovery-level signals ?"
But then I had another thought. If this black-hole merger was collected so quickly, how frequent is something like that ? They said they analyzed 16 days of data taking so far. So was there anything smaller, but still worth reporting, in the rest of the data ? I would have liked to see more information on the whole data, not just the very clear and clean signal that justified the discovery claim. For instance, showing the "next strongest" signal, showing that it was spurious, and qualitatively different from the big one, would have been appreciated.
As the speaker discussed the three phases of black-hole merging and showed a corresponding feature in the collected signal, I thought it was awesome that we were sitting there, commenting on an event happened 1.3 billion years ago, which sent a tiny ripple around, and we were able to detect it. That is kind of the same feeling I get when I aim my Dobson telescope at a faint galaxy, and think at the hundreds of millions of light years that those photons had to travel in order to make it to the meeting with my retina.
Back to the gravitational wave signal, the magnitude of the released energy also shocked me: three solar masses converted into gravitational wave energy in the matter of a tenth of a second ? That's something!
While other speakers alternated to discuss ancillary information on the collaboration and the experiments, I kept thinking at the details of the data analysis. They claim their signal is a 5.1 standard deviation effect, but they did not give much information on how they computed that. Two black holes orbiting and merging have many degrees of freedom: their masses (2), their spin vectors (3+3), and the global angular momentum of the system (3). And I might be forgetting something (such as the speed of recession of the system from us). If you change the parameters, you get different shapes of the gravitational wave signal. So we are talking about fitting something with 11 free parameters... I am reminded that with 11 parameters you can fit *anything*.
But still, the signals are so perfectly similar in the two detectors, that there can be little doubt that the effect is systematic.
In general, when one quotes a significance, one is considering "how likely it is that such a feature is observed, or one even more striking, if the null hypothesis is correct", the null hypothesis being that there was no signal to detect. But here they must be doing some multiple hypothesis testing, as their signal has many features to be considered. This again confuses me a bit.
And then there is a different chapter: systematic uncertainties. As one of the most striking features of the signal is the complete coincidence of the two detector outputs, one is led to believe that something was detected indeed. But can it be something different from gravitational waves ? For instance, a seismic wave propagating from somewhere in the US in-between the two detectors could give a spurious signal that might in principle have the same features in both. I heard nothing on the kind of potential spurious signals that can give a coincident jolt to the two detectors in the press release. I would have liked to.
The bottomline of the train of thoughts I had yesterday is that indeed the signal looks genuine, but the collaborations have offered insufficient information on the details of how they extracted it. I believe this is a real detection of gravitational waves, but I will spend the weekend trying to figure out if the articles published yesterday provide more clues of the details I wanted to know about.