My intuition is that this is unlikely, but I'd love to see someone do the math. Given the scale of interstellar distances, any locations on our planet (and even in our solar system) are going to effectively function as a single point. Given arbitrarily-accurate measurement, it could work, but I'd bet physical limitations will prevent that from being a possibility.

To my mind, it'd be roughly like trying to triangulate an earthquake in France with three sensors in a 1mm^3 cube in NYC (scale is probably way off, I definitely didn't do the math).

Yes.

There is a third one (VIRGO) near Pisa, Tuscany - a French/Italian collaboration.

Unfortunately it was not online for this event.

There is also one being built in Tokyo, and another being planned in India.

They'd have to build 4, actually. Imagine 3 intersecting spheres, there would be 2 possible positions remaining (just as with 2d triangulation, having 2 circles would leave 2 points remaining and we'd need a 3rd to find out which was correct

Even with just two LIGO locations, they were able to give a very rough estimate of the source location during the livestream this morning. (It was somewhere kinda in the direction of the Magellanic clouds. Probably.) Once advanced VIRGO comes online (later this year?) we'll have three, which will significantly improve directional accuracy for much of the sky. And when KAGRA turns on in Japan sometime around 2018, we'll have four. During the livestream this morning, someone said that would lead to source locations accurate to within 5-10 degrees on the sky.

One would think so. I'm not an expert in this area, but from the descriptions it seems like they would have at least two options for finding the source direction:

One is to compare the arrival time at each of the detectors and infer direction from the speed-of-light delay.

Another option is to measure the difference in relative strength between each of the detectors. I'm assuming the detectors aren't uniformly sensitive; perhaps they're most sensitive to waves travelling in a direction parallel to one arm of the detector and perpendicular to the other, and completely insensitive to waves that are perpendicular to both (i.e. it would affect them both the same and cancel out, or perhaps not affect either of them at all). With multiple detectors at right angles to each other, you can get a pretty good idea which way it came from.

Combining the two methods could give you greater accuracy, and also help to rule out spurious signals that are not gravity waves.

This makes me wonder what you would see if you had a sufficiently accurate directional gravity wave detector and let it run for a long time. Sooner or later, you'd get an actual image, like a telescope.

Triangulation doesn't work in this case, since how would you expect that the waves aren't curved by gravitational electromagnetic field?

I'm guessing they can already do that by corroborating with other techniques.