We're talking about it in this thread and I'm not aware of it having failed recently (other than the fact that it's always a fail, even when it works, if you see what I mean).
True, Anderson survived in the most recent case so it didn't "fail" in that sense but it didn't prove to be a great move because Bell didn't score any runs the next day. So it didn't do anything or wasn't proved to be a success. Had Bell scored runs I doubt people would have credited Anderson going in as a good move though.
In any case, important to note a nightwatchman going out before the end of play doesn't mean it is a failure, unless a batsman comes in and then goes out afterwards. So long as the batsman on the other end and the batsman due to come in are not out from when the nightwatchman comes in, it works.
I think whether a nightwatchman is a sensible idea depends on the specific batsman due to come in, the time left in the day, the match situation, the ability and batting style of your tailenders etc.
I can see why people don't like it - its a bit of a contradiction in that you send your openers out there and risk their wickets with a few overs left but aren't willing to risk your number 3 or 4. It also goes against the whole "man up and take it to your opposition, don't hide etc" mantra people like to go on about. However personally for one I think there are times when it is the right move, so long as you have a tailender who has the game for it (Gillespie was perfect, and Morkel from 3-4 years ago also had the game for it).
The other aspect I love is it is a concept so unique to cricket, and has been used for so long. Its just one of the intricacies of the game we love.