Interesting.
I wonder if the same approach would work for the motion detectors (and cameras) used to detect red-light runners.
*muses and wanders off*
EDIT: nevermind; from what you are all saying here, there's little credence to this claim.
However: it looks like it'd work on a lumenosity detecting sensor (street lamp, as well as household lamps), and could be easily done from a parked vehicle on the street (bore sighter + rifle + scope or equally effective combination). If a house only uses a single lumenosity sensor for the whole house (which seems reasonable if they're all tied together), then it'd be trivial to defeat.
There are a slew of ways to defeat motion detection lamps, too, on a time-by-time basis. I know this from personal experience as a teenager (when I would attempt to sneak by the front lamp on the house to get in late at night and avoid alerting my parents). I've heard wearing blue works, but I have no experience there. I do know that most are calibrated to a high enough tollerance for movement that, if you move slowly (say, a slow 'creep') you can easily get past the ones I've seen with this approach.
On the other hand, the one on my house is on the 'fritz' - it'll go on and off throughout the night, and there's nothing between it and my neighbor's house, at least on its most sensitive. So it'd obviously be adjustable; in a windy area (where there's a lot of waving of bushes, trees, etc.), I don't imagine a person would have it set to a high sensitivity, so obviously it'd be easier to sneak by.
Another way to screw up a motion sensor's effectiveness would be to set it off, repeatedly, for several nights. Let it turn off, then trigger it again. Repeat. It'll either get the homeowner or the neighbor irritated enough to turn it off temporarily until their (presumeably) busy schedule allows them or a contractor to come out and 'fix' it to be less sensitive (resulting in a period of time where it may not be sensitive enough to work while they fiddle.