Content ID has problems

Content ID doesn’t help prevent pirate scams on YouTube

This week YouTube announced it was tweaking the algorithms used by its Content ID system in order to reduce the likelihood of erroneous takedowns.  They also revised their dispute procedures to require rights holders to file a DMCA in order to enforce claims.  Frankly, their system has always favored anyone who “disputed” a claim.   As with the DMCA,  YouTube’s previous procedure required that the rights holder to  file a motion in court in order to enforce a disputed takedown.  Most of us don’t have the money or time for that.  Perhaps this new procedure will make the process more transparent and generate good buzz for YouTube, but as a practical matter,  it doesn’t really change much.

As I mentioned in an earlier blog post explaining how to use the Content ID, my real question is why didn’t YouTube address the other HUGE issue left unaddressed by the Content ID system?  Why can’t YouTube Content ID algorithms be adjusted to catch the massive numbers of “fake” uploads that pirates literally use to advertise the availability of illegal downloads on other websites?

They claim to be improving their “matching quality:”

At the heart of Content ID is the matching technology that identifies partners’ content among all the videos on YouTube. Earlier this year we introduced a significant improvement to how the matching happens. We continue to work on ways to make the matching more precise through better algorithms and a more comprehensive reference library.

However, no mention is made of improving their technology to detect these dummy uploads.  Pirates upload these files to YouTube and use specific search keywords to make them easy to find. Most are ten minutes in duration and contain a still from the pirated film with text overlays advertising the pirate link and a direct link in the description.  While writing this post I used my own Content Management Account to search for the recent release “Arbitrage” using the search term “Arbitrage full movie.”  You can see from the screen grab below that these dummy files weren’t hard to find.

YouTube Content ID doesn't help prevent scams

Fake “Arbitrage” uploads on YouTube advertising links to pirate downloads.

When you go to the actual post you find a typical dummy upload.  The pirate link is provided as a text overlay and in the description.

The keywords for this upload included: watch Arbitrage complete movie free,Arbitrage full movie + HD,watch Arbitrage complete movie online,Arbitrage part 1,Arbitrage Part 1 Full Movie + Free,watch: Arbitrage full online movie,Arbitrage :complete movie,Arbitrage free HD movie,watch: Arbitrage part 1 movie + free,watch Arbitrage online

YouTube Content ID doesn't help detect pirate scams

Dummy upload for “Arbitrage” advertising link to pirate download.

When you click on the link you end up here:

YouTube links to pirate site

Pirate site that uses YouTube to drive traffic.

If you look at one of these pirate’s YouTube page, you’ll find dozens of dummy uploads.

Content ID doesn't prevent YouTube from linking to pirate downloads


While it’s likely that thousands of these dummy files are uploaded daily on YouTube, in my experience they go undetected by the Content Management System.  As I wrote in a post on earlier this year:

Pirate thieves are entrepreneurs at heart.  Money is what drives them.  Since they can’t upload and/or monetize content they own via Youtube they resort to the next, best thing.  They use Youtube as an advertising vehicle–a convenient gateway–to connect “customers” to their illegal websites.

I’m not sure why YouTube fingerprinting ID technology  can’t automatically detect these files (most are videos comprised of a ten minute freeze frame).   Why allow pirates to use YouTube for free advertising?  To add insult to injury, notice too that when one navigates to one of these dummy uploads YouTube conveniently lists additional dummy uploads in the sidebar:

Fake YouTube pirate links

More dummy uploads offered by YouTube

As it stands, the only way a rights holder can detect them is to manually search and remove…a time-consuming process that few can afford.  If folks at YouTube really want to “fix” their Content ID system, cutting off this lifeline to pirate websites would be a start.