Final Project
editFor my project, I proposed and added a section to the youtube wiki page, elaborating on content id. Here is the edit:
In 2007 (http://online.wsj.com/article/SB118161295626932114.html), Google Inc.’s YouTube began testing on video identification technology in the hopes that if it could identify works posted to YouTube without the consent of the owner, the technology could swiftly remove them. With backing from AudibleMagic’s audio-recognition technology (http://news.cnet.com/8301-10784_3-9797622-7.html?part=rss&subj=news&tag=2547-1_3-0-20), in August Google launched in beta form the YouTube Video Identification tool (http://googleblog.blogspot.com/2007/10/latest-content-id-tool-for-youtube.html). Upon moving out of beta and into final form, this technology was renamed Content ID as it is known today (http://youtube-global.blogspot.com/2010/12/content-id-turns-three.html), a technology that allows “copyright holders to easily identify and manage their content on YouTube” (http://www.youtube.com/t/contentid_more), whether their content is identified in its entirety or only partially (http://www.youtube.com/t/contentid).
The Content ID tool works by creating an ID file from content that the owners submit into the system and corresponds to pieces of content that get stored in a database to be checked against pieces of audio or video uploaded to YouTube. If a match is found, a Usage Policy is carried out given the preference of the content owner. The three usage policies are Block, Track, and Monetize, where a block renders the uploaded video un-viewable, a Track yields statistics about the video, and Monetize runs ads in conjunction with the video. (http://www.youtube.com/t/contentid_more)
In 2009, a string of tests which included various tweaks such as reverse playback, resampling, and amplification to the video, “I Know What Boys Like”, was conducted on the tool to gauge its effectiveness. While the Content ID tool was noted for its pervasiveness and ability to pick up various amplification changes, some uploads still managed to get by the tool. (https://www.eff.org/deeplinks/2009/04/testing-youtubes-aud)(http://www.csh.rit.edu/~parallax/) That said, while the Content ID tool is effective in matching uploads to the database, the tool is not perfect and is constantly fine-tuned to prevent user uploads from slipping past the system (http://www.youtube.com/t/contentid_more).