Also, a minor point but there are only two LIGO detectors online at the moment, with a third sister detector in Italy (named Virgo), and a fourth coming online soon in Japan (named KAGRA). There does in fact exist a third LIGO instrument, but it's currently mothballed, awaiting construction in India.
Prior to our first detection, the overwhelming prime directive of our collaboration was _not to make a false detection_ and we went to insane lengths to avoid one; e.g., we had a small team of people secretly injecting false signals -- "blind injections" -- into our data, so that we all expected to be regularly seeing them and wouldn't be tempted to gin the analysis to find a detection where there wasn't one. (Amusingly, because of this, it took weeks for many of us to believe the first detection was real, even after the blind injection team swore it wasn't one of theirs. In the end, we charged an independent team to do a forensic analysis of everything from the security cameras and seismometers in the observatories, to every computer and disk the data passed through, to convince ourselves this couldn't possibly have been maliciously injected by hand by a rogue scientist. It was a wild few weeks!)
Now that it's clear gravitational waves exist, however, we focus on optimizing the false alarm rate for astronomers (who want it well above zero so they don't miss anything) rather than optimizing for ~zero false detections.
We believed that if we had announced a discovery that turned out to be wrong, it would probably have meant the end of our experiment, and in practice the end of the field, at least for a long time. It was fortuitous that the first detection turned out to be gold plated and unambiguous, otherwise we would have probably published a bunch of "we saw something interesting but can't claim it as a GW discovery" papers on the next few weaker events before we would have felt comfortable making a confident claim.
This is now my favorite AMA
But with time dilation as black holes near each other, shouldn't the frequency change be exponential? Or does it cancel out - the frequency goes up as they get closer, and then time dilation lowers it back as it dilates into infinity?
I know nothing about the actual hardware. But I can reasonably speculate that LIGO can afford to have local atomic clocks. So they can timestamp their observations to nanosecond precision.
Given the finite speed of light, the candidate astrophysical events are necessarily detected many milliseconds apart. There can only be two ways these potential events would correlate:
1) there are so many many possible false positives constantly happening that there is some reasonable probability of this correlation occurring by chance.
or 2) a highly sophisticated "goof" or fraud. Someone could presumably set up some local source near each detector and spoof a signal. E.g. carefully "wiggle" large masses, each only a few km away from each detector.
So, if it's not astrophysical, can Sabine tell us which it is? False positive? Goof?
They're ALSO not independent observations, though. The LIGO observation was made, and then the directive was "go look for correlating events". That's a dependent observation.
To be truly independent, they would have to inject synthetic data (say 3x) into the observation reports and make sure that the multi-messenger results don't correlate with the synthetic data, much like GP says they did "at the beginning of LIGO" with internal (non-multimodal) signals.
This would have the side benefit of making the analysis pretty simple to do (just a chi-squared analysis) and you don't have to have a PhD in signals analysis with a specialization in the filters used by LIGO to believe with a high confidence that 1) GWs exist and 2) LIGO is actually measuring GWs.