Yep, a theoretical test that makes absolutely no sense for a smartphone camera. Or when was the last time you mounted your smartphone on a tripod?
OIS is actually counterproductive when used on a tripod. Not sure if L920 detects this and turns it off (unlikely).
Yeah I doubt it turns it off automatically.
Getting back to DxOMark, the way they score, they measure various things and average them. Problem is they are basically doing daylight testing only. Okay, so let's say they got 66 for the 920 and 72 for iphone/GN2. Now, if we do the same set of tests handheld in lowlight, we probably get something like 66 for the 920 and 10 for iphone/GN2. And let say you use your camera in low light about half the time so you weight them equally. Now you get (66+66)/2 = 66 for the 920 and (72+10)/2 = 41 for iphone/GN2. You get my drift? Of course, iphone/GN2 users won't be using their cameras all that often in low light...because they can't! lol