Q321 Mr. Evans: Can I ask what changes you have introduced in the company since making that error
[ed. : regarding a gang rape video that was not taken down fast enough] to ensure it does not happen again?
Mr. Walker: A number of different things. The mistake that was made had to do with the way the individual reviewer coded the video so that additional flags that came in were not immediately escalated to the beginning to the queue. We have made that much harder to do. It is now a double trigger, it needs to be reviewed twice. I do not want to go into the details because it would allow people who are trying to game the system to avoid the technologies. Where there are other signs of content being inappropriate that allows for a secondary review. There are a number of other things I would be happy to talk about in a private session. (House of Commons, 2008, Ev 121)
Xanga has patented its flagging system, that it launched on May 1st, 2006 , the patent is entitled “SYSTEM AND METHOD FOR FLAGGING INFORMATION CONTENT ” The patent was filed on April 27, 2007 and is a direct reaction to the “explosion in the amount of User-Generated Content (UGC) being created” and the fact that this “is exceedingly difficult to police such a massive amount of content. A problem that will only grow over time as the UGC industry matures.” This was not an understatement; it was during 2005 that the term UCG was picked up in mainstream usage. And if we look at the startups of the earlier mentioned video sharing sites; Vimeo and Yahoo! Video started in 2004. YouTube, 56.com, Blip.tv, Dailymotion, Google Video & Tudou started in 2005. iFilm was founded back in 1997, but was taken over by MTV in 2005. Youku started in 2006 and Veoh in 2007. The content that was generated on all these platforms was in need of some kind of moderation that was suited for that amount of content, such as flagging by the community. Resulting in user-controlled content. The reasons that this is not done by hired moderators is summed up in the patent:
“(…) full time moderators are very expensive.”
“(…) such moderators can not respond quickly enough to police content (…)”
“(…) it is difficult to “scale up” and quickly hire as many moderators as needed (…)”
The second and third reason are inherently connected to each other, the amount of moderators can never measure up to the amount of users and their activity. A recent example, taking place in China, not only gives an (unrealistic) indication of how it could work;
“In addition to its massive firewall and intrusive software, the government employees thousands of paid commentators who pose as ordinary Web users to counter criticism of the government. Known derisively as 50 Cent Party members, these shapers of public opinion are often paid a small sum, 50 Chinese cents, for every posting.
In another sign that Chinese officials are trying to assert more control over the Internet, the city of Beijing wants to recruit 10,000 volunteers by the end of the summer to monitor Internet content, according to the Beijing government’s Spiritual Civilization Office.”
But also provides an additional argument against hired moderators, for this approach is highly sensitive to censorship. YouTube uses a similar argument, against pre-monitoring content, in the report by the House of Commons.
But what’s left out being mentioned here is that the task lies outside the community itself. Leaving us with the question what the users have to say in the moderation process (in the case of moderators being hired). The second question would be which side gets to decide who’s fit to be a moderator. Just as the amount and speed of the content, it’s also a matter of judgment. The judgment of a few moderators against that of a whole community to decide whether a video is right or wrong.
But a decentralized flagging system that gives it all in the hands of the community also has its downsides such as fraud and other misuses. This patent claims to have formulated a fraud-resistant flagging system by adding prioritizing hierarchy amongst the users in the form of adding ‘flagging weight’ in accordance with accuracy of the user’s flags. My main objective towards this patent is an argument I have used earlier, who gets to decide whether a video is accurate or not? Which in their case is left up to moderators and/ or ‘flagging review module’ (by which they mean the software).