Bruce, a question about An Optimized Application

Quite right! Ultimately space is the place to be!
( What 'good' really means is that the 'dark port' of the interferometer has been held at an interference fringe minimum by the active use of the incredible control system of the whole array. One then takes the signals used in that system to perhaps deduce the flexing/distortion of spacetime induced by a passing gravitational wave propagating through the entire LIGO. The poorer alternative is to not do that control effort, but measure the light fluctuation at that port..... )
Anyhows, what makes the project not entirely hopeless here on Earth - given that we are looking for alterations in spacetime of the order of 10 to the power of minus 21 while using instruments, wavelengths, noises etc all far larger than that - is that most 'background' is incoherent. This means that many non-astronomical influences like the 'shot noise' of a laser, or thermal movements of the atoms in the mirrors do not have a preference with respect to the direction of the effect of scientific interest. A good analogy is the seaside where there is a lot of slosh, froth, wind, waves and spray etc.... but if you watch for long enough ( much, much longer than these 'obvious' disturbances ) the tide comes and goes. In other words averaging smears away such temporary upsets and allows trends to be observed. Another helpful method is to 'park' the noise in less critical parts of the observed data. So if you have some noise component that blurs a region of particular interest ( technically in the power vs. frequency spectrum ) then one can cleverly design widgets to interact physically with the detector parts to channel energy away into other 'modes'. So you ask the tenor to sing just as hard but in the baritone range! In addition, if you can't do that, then make the noise well defined, 'narrow', well studied, and understood - hence you can confidently subtract it away from the data. Another expectation is that widely separated detectors ought not suffer identical patterns of incidental noise - like a tree being felled in the Louisiana woods, or a brushfire in Washington State - so comparison between sites ( with simultaneously locked interferometers ) is quite valuable. Another trick is to get the chalk and pencil out and make a well educated guess as to what sort of signal ought to be heard for some given astronomical phenomenon, and then search for that in all the hubbub.
Cheers, Mike.