F/O,
I would say that each situation has to be treated separately. I'm not sure about the idea of landing on a road, though. The only way you're going to know where the road is, is if there are either cars on it (bad idea to land - I'd rather kill myself in a field than kill myself and a dozen drivers on a road), or streetlights (are they far enough apart that your wings are going to fit between them?)
In most cases, I think I would go for a dark area, and hope it turned out to be something I could land on. And then I'd go for homeguard's idea of switching on the landings light... and if I don't like what I see, switching it off again!
No doubt that this is a big problem. I seem to remember from my ATPL groundschool that, in commercial aviation, you deal with risk as follows: take the chances of a situation arising. Then take the chances of a successful outcome from that situation. Multiply them together, and if that comes out as less than one in a million, it's an acceptable risk. With night flying in a single-engined aircraft, we have to admit that the chances of a successful outcome following an engine failure are small. But the chances of the engine failure happening in the first place, in any almost-modern aircraft, are also small - so the total risk, although too big to be ignored, is actually not all that big.
At least, that's the way I justify it to myself!
FFF
---------------