Let's go back to the facts. The events began in 2006, with a horrible episode of Italian children bullying an autistic schoolmate that was captured on a video and posted to YouTube. "The video was totally reprehensible," writes Matt Sucherman, Google vice president and deputy general counsel, Europe, Middle East and Africa, in a post on Google's blog, "and we took it down within hours of being notified by the Italian police."
Nevertheless, last week the Italian courts held Google executives criminally responsible for allowing the autistic child's privacy to be violated. Although the executives were absolved of defamation charges, three of the executives (one of whom no longer works for Google) were convicted of privacy violations. An appeal is in the works.
The Web almost immediately filled with support for Google. On the logistical level, Google's supporters argue, there's no way for executives of a site filled with literally years of constantly updated user-generated content to stand guard over every upload that the site sees. And on the legal level, Google's supporters argue that there shouldn't be a difference between Google's level of culpability and, say, the Postal Service's culpability for hate mail that gets delivered through its systems, or the phone companies for offensive calls. The U.K.'s former Information Commissioner has gone so far as to say that the Italian case gives privacy laws a "bad name."
But for all the legal discussion, this is ultimately a search problem -- specifically, a problem for video search. SEO experts have long known that the engines are far better at understanding text than at understanding video. Often, the engines' understanding of video clips comes from understanding the text surrounding a given clip, not from the clip itself. Thus, it's hard for engines to create an automated flagging and takedown system that could know when its video uploads aren't OK. Instead, YouTube relies heavily on its users to notify Google of objectionable content. As a result, it can take "hours" for an offensive clip to be removed -- as the Italian courts will readily tell you.
Because Google's legal troubles are search troubles, I see a silver lining in all this. The legal threats may be the excuse that Google was looking for to work harder on improving the video search capabilities it has. That work will save Google from future legal trouble; it will also help as Google looks to stay ahead of Hulu and Co. and remain the leader in online video content.
This wouldn't be the first time YouTube's legal challenges became long-term opportunities for improving video content identification in highly monetizable ways. Consider YouTube's growth from its early days as a copyright-suit target, to its emerging status as a sought-after partner among movie studios and music companies. As I've written about before, much of that transformation is due to of Google's development of digital rights "fingerprinting" technology, which automatically identifies pirated clips within YouTube and gives rights owners the option of either removing or monetizing the content they own. This technology didn't exist when Google first acquired YouTube.
Of course, as difficult as locating infringing video content may be, it's still a search for clearly defined information, pulled from a clearly organized database. Using automation to find content that's generally offensive is a much fuzzier -- and therefore much more difficult -- problem to solve. But that isn't to say that the base work of a content-flagging system isn't beginning to take form. Google Goggles can already identify an object and search for it, based on the picture the user takes through a mobile device; clearly, Google is developing mastery at identifying an object without the assistance of surrounding text. Meanwhile, parts of the security industry already have automated systems in place that know what suspicious activity looks like, so moving-image flagging devices do exist.
Of course, the kinds of developments I'm thinking of are years away from where we are now. But when they finally do happen, they'll be a boon for making YouTube a safer place -- and video search a far more useful feature -- in ways that both users and the Google monetization engine could use well.
Which is why a better video search experience just may be Google's silver lining to its legal woes.
Smoking crack much steve.... are you really suggesting that once search for video is improved google will be ble to run a blocking search for "bullying autistic schoolmate" and this video will be pulled by non human interaction.....
sorry but someone had to call BS on this article.
Dean, I was disappointed to read your comment and I guess naive -- I thought this column delivered terrific insight -- but your suggesting sans the crack comment, that this kind of video monitoring/removal service Steve suggests is impossible -- I bet someone said that about Google back before Google became Google.
Steve's point which I thought was brilliantly and simply stated, is that Google will now have to focus resources to apply learnings they have obtained via Google Goggles (as he astutely points out) to create the impossible so they can remove videos of this nature -- and these recent legal proceedings are pointing them towards this goal which will be a net positive on many levels even though the legal outcome appears to be quite negative on the surface.
Steve should be commended for pointing this out, no?