A slightly indignant article in the Wall Street Journal about how video uploaders are putting false tags on their videos to attract more audience.
This is yet another example of why unmoderated and self-policing may work for the Wikipedias of this world, but why it won't work for video.
The trouble is that we're years, if not decades, away from video recognition software that can accurately tell what's being displayed on screen and can therefore interpret the context and content of a video (although I bet Google have got some clever people working on it).
In the meantime, at best, we're dependant on transition identification and voice recognition software to spot the cuts and log the soundtrack of videos. This may work quite well on news reports but is useless for music, sports, features films, etc..
The reality is that the metadata is everything and the old 'rubbish in, rubbish out' adage is so true.