The following is a more detailed description by Ed Groth of the signals that we're looking for.
We're looking for very short pulses - less than 10 nanoseconds. In this time, light travels 3 meters.
If both Harvard and Princeton see the same pulse from the same object, we expect a difference in arrival time at Harvard and Princeton because the pulse had to travel a little farther to get to one observatory than the other. The exact time difference depends on the geometry of the observation, which depends on the position of the object and the time of day as the Harvard-Princeton baseline rotates with respect to the object's direction.
The biggest time difference that could conceivably occur would be when Harvard is looking directly away from Princeton at the source or when Princeton is looking directly away from Harvard. Then the light going to one observatory has to travel an extra distance equal to the Harvard-Princeton distance. Of course, this is not a realistic situation since one observatory would be looking slightly above the horizon and the other slightly below! But it does give an upper limit to the time mismatch.
Princeton University's Fitz-Randolph Observatory is located at 74° 38.8' West longitude, 40° 20.7' North latitude, and 43 meters altitude. Harvard University's Oak Ridge Observatory is located at 71° 33.5' West longitude, 42° 30.3' North latitude, 185 meters altitude. With the rather messy formulae that account for the fact that the Earth is an oblate spheroid, the straight line distance between Oak Ridge and Fitz-Randolph is 352.33 kilometers, which takes 1.1752 milliseconds for light to travel.
If we treat the Earth as a sphere with mean radius 6371008.8 meters, the straight line distance is 352.04 kilometers, which is a light travel time of 1.1743 milliseconds.
A more realistic observing geometry would have the extra distance of one path relative to the other be something like half the above, or maybe 0.6 milliseconds. If we actually see a coincidence within a millisecond, then we would want to verify that the pulses were received with a delay exactly equal to the extra distance from one observatory to the other within some uncertainty window.
The naive uncertainty window would be the pulse width, 10 nanoseconds. But there are errors in the timing, etc. GPS errors in position amount to about 10-100 meters, so taking 100 meters as an upper limit, this amounts to about 300 nanoseconds. There are delays in the cables and the electronics. Since we have the same electronics as Harvard, we might ignore the electronics delay and assume an unknown difference in cable lengths of about 100 meters, which adds another 600 nanoseconds (speed in a cable = 1/2 speed of light). So altogether we're looking at about 1 microsecond of error in the delay.
In other words, for us to conclude that both Harvard and Princeton might have seen the same short pulse, the pulses must have been seen with the correct delay to within about a microsecond. The delay will be about a millisecond or less.
If we actually see something like this, we would of course try to measure all the delays in the electronics and cables to cut down the coincidence uncertainty due to our lack of knowledge of the delays. However, I think the limit from the GPS will be hard to get around.
Once we did all that and the coincidence still passed, we would have just one event that might be a sign of an extraterrestrial laser or it might just be a really unlucky random event!
Caveat: Of course, the source might have sent a pulse train (that's what one expects in any case). If we can just barely see the pulses, then Harvard might see one and Princeton might see one earlier or later in the train. But we really have to be looking for the same pulse. Having a very narrow time window is what makes this technique work at all. Besides, if the extraterrestrials are aiming their lasers at us, they ought to be able to figure out what strength they need to make sure we can see all the pulses. In which case, we might have discovered something optical and extraterrestrial, but not intelligent!
Bottom line: We need to see something within about a half a millisecond of when Harvard sees something, then we need to do lots of analysis, then we both need to see it again before getting excited!