Google YouTube copyright infringement filtering system - How will it work?

Google / YouTube was in federal district court in New York on Friday defending a lawsuit brought by Viacom and several other media companies. The New York Times story says "Google hopes technology will be in place in September to stop the posting of copyright-infringing videos on its YouTube site, a company lawyer told a judge presiding over copyright lawsuits yesterday."

This is like Lindsey Lohan saying, OK so I did drive drunk a couple times, and did drive with a suspended license, but I am in rehab now and I won't do it again. So, everything is OK now. Lets just drop this and move on. Hmmm...not so fast.

Lets think about a few questions; How will the filtering system work? How will users subvert the systems? What if the system actually works? Will advertisers be interested in a YouTube without the good stuff? Even if it does work will the advertising economics work? First a little more background, then lets take those questions one at a time.

C/Net News story says "If in fact Google puts this (system) in place, it is obviously way too late," Solomon said. "But we encourage Google to come forward and do what other companies have already done and treat all the content providers fairly. Not just the favorite few who have agreed to share advertising revenue with YouTube."

During the court proceeding, lawyers from both sides estimated that pre-trial discovery could take more than a year. That means there's a chance that YouTube's copyright issues may not be resolved until late next year.

Viacom is suing for $1 Billion for past copyright infringement. What Google/YouTube does in the future is irrelevant to the facts of this case. The laws against copyright infringement are strict and the penalties are stupendously large...$250,000 per instance. Viacom only needs to document 4,000 instances of copyright infringement to collect $1B in damages. YouTube probably does 4,000 streams of "The Daily Show with Jon Stewart" in one day.

How will Google's copyright filtering work? I haven't seen it, but based on what I have read it works by creating a "digital fingerprint" of the original video or music, and comparing it to every file uploaded to YouTube. Music and video uploaded to the Internet is converted to a digital file which is basically a series of ones and zeros. Digital fingerprinting simply records the series of ones and zeroes, then uses pattern matching technology to find instances of matches, even small sequences found in short clips.

What does this mean for music and video producers? It means they must submit to Google an original copy of every piece of content they own. They also must provide proof that they own the copyright. Then they need to trust Google to screen out any offending uploads. The copyright owners will still need to monitor YouTube to make sure the system is working...and serve Google with "take down" notices when it fails.

How will users subvert the system? People who upload music and video are usually pretty tech savvy and take delight in beating any system. They will do the obvious things; change the name of the file, chunk it up into short clips, add characters, sound, or other video to confuse the system, stretch or wave the video to make it appear different, and a bunch or other tricks. Think about the bot filtering technology on blogs and web sites that require you to type in a series of letters to confirm you are a real person. A computer bot can't read the squiggly lines and waves in the letters but humans can easily read them. The same sorts of techniques can be applied to music and video to fool algorithmic filters.

What about "sampling" and "fair use"? The copyright law allow for "sampling" and "excerpts" under the "fair use" provisions. These laws are very imprecise and open to interpretation. No computer generated filter will be able to determine what is "fair use" and what is copyright infringement.

What if the copyright filter system actually works? Will YouTube be interesting if there is no "good" content? How many videos of college kids drinking and dancing will you watch? Will the power users bother to upload music and video if they can't put up the good stuff? Will YouTube viewers care about watching it?

Will advertisers be interested in a YouTube without great content? YouTube is very expensive to operate. The hosting costs, bandwidth, infrastructure, and people cost millions. Advertising is the only real source of revenue. Will advertisers be willing to pay for placement on YouTube? Will they be comfortable displaying their brands next to random, perhaps raunchy, content? Even if they do, will the economics work? Low ad rates against expensive content don't work no matter how you scale it.

There is a great future in Internet video...at some point. YouTube may be the early pioneer that makes all the mistakes and clears the way for a "fast follower" to come in and do it right. Or, YouTube may figure it all out and make the necessary changes fast enough to survive and thrive. There is a long history of early innovators being eclipsed by fast followers. Some of the issues discussed above are reasons why it happens so often. The jury is (literally) still out on YouTube.

Subscribe - To get an automatic feed of all future posts subscribe here, or to receive them via email go here and enter your email address in the box in the right column.

Comments

[kid mercury]

great post as always, don. IMO the key consideration is that underdog niche video publishers are not the copyright crybabies; it's the established film and music industries that are. and once these niche video stars emerge, advertisers will not have any problem in putting ads around them -- and the independent niche video stars will be happy to take a majority cut of the revenue.

if the copyright crybabies want to whine and sue, fine. the youtube game is not really designed for them anyway.

[G]

Kid Mercury - you miss the whole point of why advertisers will flock to YouTube - your argument would be analogous to a prime real-estate company building a lurxury mall stacked with the best a facility can offer with high rent - and then put up a sign that says "we are going to build the biggest fanciest mall for 99 cent stores". - where is YouTube's competitive advantage at that point?

[kid mercury]

not sure i follow your analogy -- are you saying i'm saying 99 cent stores are the niche video publishers? if so, i dont think they are like 99 cent stores; perhaps more like creators/distributors of high quality niche products.

[Brian Evans]

Interesting article that Mark Cuban wrote last month about this...
http://www.blogmaverick.com/2007/06/12/gootubes-content-protection-programs-that-aint-it/

[Mark Z.]

Small video producers are aiming to become big video producers because this is where (they think currently) the money is. Happens all the time. As long as they are small, they look for distribution. Once they are big, they try to protect their success, which means selling out to the big wigs, which means playing the copyright game...

[Yusuf Motiwala]

Great post Don. I often visit youtube if I’ve missed any of my favorite TV episodes. Hence, after filtration youtube may not get as many hits from people like me. I can definitely imagine some loss of traffic.

Google has been talking about fingerprinting but I too wonder how it will work. We all know that any pattern matching or hash in compressed space is definitely not going to work (transcoding, reencoding, etc) and pattern matching in uncompressed video may not be practical and IMO, even that is defeatable too. So subverting a digital fingerprinting based detection scheme will definitely happen.

A relevant post is here http://www.signedmedia.com/blog