Introduction

1. An ontology of user-options for control on YouTube

2. Research hypothesis, framework and methodology
2.1 Hypothesis: “Adding a user-controlled meta-moderation system to the
YouTube flagging system would make it more democratic.”
2.2 Theoretical Framework
2.3 Methodology

3. YouTube’s regulation model: flagging as part of the whole
3.1 International / legal level of regulation
3.1.1. Notice and takedown policy
3.2 Level for self-regulation
3.2.1. Flagging issues
3.3 Level of automation in content regulation
3.3.1 YouTube video ID technology
3.3.2 Automated Flagging
3.4 YouTube’s model concluded

4. Community’s viewpoint on the flagging system on YouTube
4.1 Methodology
4.2. Flagging problems as experienced by the community
4.2.1 Anonymity & accountability: who flagged my video?
4.2.2 Accountability: I can’t contact YouTube
4.2.3 Motivation: why was my video flagged?
4.2.4 Flagging campaigns
4.2.5 Vote bots
4.2.6 Censorship: the claim to free speech
4.3 Findings

5. Case Study: LGBT community on YouTube
5.1 F(l)agging gay content on YouTube
5.2 Strategies
5.3 Conclusion of case study

6. Flagging on other video sharing sites
6.1. Categories
6.2. Visibility
6.3. Handling of copyright
6.4. SafeSearch

7. Hierarchical moderation systems
7.1 Slashdot (1997): meta-moderation
7.2 Wikipedia (2001): user levels
7.3 Digg (2004): undigging
7.3.1 Funny or Die (2007)
7.4 Additional comments

8. Proposal: Combining flagging with rating
8.1 Model
8.2 Possible points of critique
8.3 Possible benefits

9. Conclusion

References

Appendix I : list of YouTube blogposts related to flagging

I. Blogposts on youtube.com/blog addressing or related to flagging;

Nov 13, 2008 “Flagging at YouTube: The Basics”

Oct 27, 2008 “Addressing Youth Violence on YouTube”

April 17, 2008 “YouTube Policy Enforcement Changes”

Nov 06, 2007 “Improvements to Video Flagging System”

Aug 06, 2007 “Choose Your Own Thumbnail Image”

Oct 19, 2006 “Greetings from the YouTube SQUAD”

“Please Be Decent and Kind”

Oct 08, 2006 “How Flagging Works”

June 24, 2006 “Comments Bug”

Dec 14, 2005

Aug 11, 2005

Benkler, Yochai. “The Wealth of Networks: How Social Production Transforms Markets and Freedom” Yale University Press, 2006.

boyd, danah. “Why Youth (Heart) Social Network Sites: The Role of Networked Publics in Teenage Social Life.” MacArthur Foundation Series on Digital Learning – Youth, Identity, and Digital Media Volume (ed. David Buckingham). Cambridge, MA: MIT Press. 2007.

boyd, danah. “None of this is Real.” Structures of Participation in Digital Culture (ed. Joe Karaganis). New York: Social Science Research Council, 2007, pp. 132-157.

David, Shay. “Toward Participatory Culture” Structures of Participation in Digital Culture (ed. Joe Karaganis). New York: Social Science Research Council, 2007, pp 176-199.

Deuze, Mark. “Media Work” Cambridge: Polity Press, 2007.

Fuller, Matthew, ed. “Software Studies: A Lexicon” The MIT Press, 2008.

Golder, Scott and Bernardo A. Huberman. “Usage Patterns of Collaborative Tagging Systems.” Journal of Information Science, 32(2), 2006. p. 198-208.

Goldsmith, Jack and Tim Wu. “Who Controls the Internet? Illusions of a Borderless World.” Oxford University Press, 2006.

House of Commons: Culture, Media and Sport Committee. “Harmful content on the Internet and in video games” Tenth Report of Session 2007–08, Vol.II. Oral and written evidence. Ordered by The House of Commons to be printed 22 July 2008.

Jenkins, Henry. “Convergence Culture: Where Old and New Media Collide” New York University Press, 2006. p.257.

Lange, P. G. “Publicly private and privately public: Social networking on YouTube”. Journal of Computer-Mediated Communication, 13(1) 2007, article 18.

Lessig, Lawrence. “Code: Version 2.0” New York: Basic Books, 2006.

Marlow, Cameron, M. Naaman, D. Boyd, M. Davis, Position Paper, Tagging, Taxonomy, Flickr, Article, ToRead. Collaborative Web Tagging Workshop at WWW 2006. Edinburgh, Scotland, May 22, 2006.

Novek, Beth Simone. “Wiki Government: How Technology Can Make Government Better, Democracy Stronger, and Citizens More Powerful” Brookings Institution Press, 2009.

Pold, Søren. “Button” Software Studies Lexicon. Ed. Matthew Fuller. The MIT Press, 2008.

Rassi, Termeh. YouTube: Examing a Revolution” p. 40 gnovis journal _ Fall 2007 _ Volume 8, No. 1.

Rosen, Jeffrey. “Google’s Gatekeepers” Published November 28, 2008.

Taylor, T.L. “Pushing the Borders: Player Participation and Game Culture” Structures of Participation in Digital Culture (ed. Joe Karaganis). New York: Social Science Research Council, 2007, pp. 112-131

Tisselli, E. “thinkflickrthink: a case study on strategic tagging”. Accepted, to appear:
Communications of the ACM Journal.

“Us/Now”. Documentary. Dir. Ivo Gormley. Banyak Films, 2009. Transcripts of interviews:

Wesch, Michael. “An Anthropological Introduction to YouTube” a YouTube video by Michael Wesch and the Digital Ethnography Working Group. Posted on YouTube on July 26, 2008.

With this thesis I’ve tried to democratize YouTube’s flagging system by suggesting a reputation logic amongst the user community. Thus providing a platform that not only embodies the voice of the community as a whole, but also emasculates the dissident viewpoints which are variously labelled as ‘biased’, ‘ideological’ or ‘extreme’. By analyzing flagging as part of the whole moderation system, I came across some very interesting aspects such as the different interpretations of free speech and censorship, the strength and weakness of anonymity and appropriation of technology.

The tools that YouTube presents to the community can be interpreted as a form of corporate convergence, by which they shape the different flows of media. The way the community appropriates these tools (the flagging feature in particular) is a grassroots convergence that shows a strategy and demand for greater participation. Exemplifying how technology can be a determinant factor in participation, on which point I disagree with Jenkins. Jenkins shows a technological deterministic view by contrasting ‘participation’ with ‘interactivity’; according to him ‘interactivity’ is determined by technology and has a predetermined outcome and ‘participation’ is determined by social and cultural factors and has a more open outcome, more fully shaped by consumer choices. I have shown that users can appropriate given technological tools and make them their own in order to gain more control and participation on YouTube.

I do agree however with Jenkins’ concept of convergence and how this changes the roles between the corporation and the community. Taking this one step ahead on YouTube by proposing a redesigned model of the existing flagging system in which I add a reputation system and a meta-moderation logic, which potentially eliminates the existing issues and creates a democratic open environment. And thereby empowering and stimulating the grassroots convergence in participating and shaping the community, which ultimately benefits YouTube as well.

Increasing the community’s commitment
The change of rules is inspired by the critical feedback from (part of) the community. Addressing their issues by changing the system could means that this could build up greater loyalty amongst the community through this responsiveness.

Accountability
It is one thing to tap into the power of the distributed flexible volunteer ethic, but corresponding systems of recognition and accountability must be in place. This could be such as system, according to my opinion.

Chance of game play
Game play is not necessarily a negative development. The outcomes can possibly enrich the community and the platform even further. It is unavoidable that through time the demands of the users will change and the system needs to be adaptable to that.

“The interaction of people with information systems is recurrently marked by play and experimentation, as people test the limits of their settings and manage the consequences of unexpected interactions and altered contexts. Digital social structures disrupt the boundaries that define social communities, but the reassessment of context and performance that accompanies this is endlessly generative.”

There will always be a part of the community that will test the boundaries of what is given and explore possibilities. Because the outcome of this is unpredictable, it could also proof to be beneficiary to the community.

Disrupting the ecology of YouTube.
Although this thesis has a high hypothetical degree to it, if it were to be applied to YouTube, one of the biggest pitfalls would be that it would force many user of the YouTube community out of their comfort zone by changing the rules of the game.

“We are in a critical moment of transition during which the old rules are open to change and companies may be forced to renegotiate their relationship to consumers. The question is whether the public is ready to push for greater participation or willing to settle for the same old relations to mass media.”

The risk of game play.
There is also a chance that the strategy of reputation backfires because of game play. It is not unlikely it would create a competition amongst users for achieving the highest rank. And as we’ve seen happening with existing aspects of YouTube, this system of reputation will also be tested on its boundaries and perhaps appropriated in ways we cannot predict.

Chance of oligarchy
With the chance of game play and the consequences of this, it is feasible that a certain amount of users seize power in this newly evolved democracy and turns from being ruled by the people into ruled by the few.

Tyranny of the majority
The phrase ‘tyranny of the majority’, used in discussing systems of democracy and majority rule, is a criticism of the scenario in which decisions made by a majority under that system would place that majority’s interests so far above a minority’s interest as to be comparable to tyrannical despots.

Is this a way of giving the power over to the users?
One could argue that it only applies a part of the moderation system and leaves out the legal and corporate part as earlier mentioned. The user can help enforce the rules of the house, but they are not allowed to either make them or break them. It is however not unlikely that the user community come to see of the platform as their territory, a phenomenon that is also witnessed in the area of player production in games;

“when designers take user engagement seriously, users sometimes come to feel they have a real stake in the product. The issue, then, is not just how to encourage player-produced content. It is also to fundamentally think about, fully acknowledge, and integrate into the game structure the engagement and strong commitment of player-producers and the sense of investment that often follows. Raph Koster, chief creative officer for Sony Online Entertainment, has written an imaginative exercise entitled “Declaring the Rights of Players” (Koster, 2000). It is striking in the ways it envisions virtual world users as citizens of that space, with due attendant rights. As a thought experiment, it is a great example of the kinds of things we might have to consider when we reformulate passive consumption into active engagement.” (Taylor, 2007, p.125)

It should be emphasized the new system is in order to democratize YouTube. And the idea of the users being citizens of that space, with due attendant rights fits right in there. Not only of having the right to free speech, but also the right to be held accountable for their actions. Leaving the company deal with corporate, legal and marketing aspects of this democracy in order for the platform to exist.

The model that I am describing below is an extension on existing systems and form a starting point for discussion on what would work and why. Names, labels and number indications serve this purpose and are made up and not specific to YouTube. Setting up an open protocol/standard not unlike this one would increase transparency, accountability and democracy on YouTube.

A set up of a reputation building model on YouTube would look like this:

1. 0 – 150 points newbie
2. 150 – 500 points resident
3. >500 points citizen
4. >1000 points honored citizen

A newbie can only connect to other users, post videos, video responses and leave comments. Allowing him/her to position him-\herself in the community and get acquainted with the system. A newbie is not able to vote to discourage the practice of flagging campaigns by sockpuppets and vote botting.

A resident does not differ from any YouTube user in the system as it is now. He/ she has access to all existing options, including voting.

A citizen is registered for meta-moderation, a new feature in this model, which allows the user to remove or confirm one flag per video.

An honored citizen is able to lock a certain video, similar controversial articles on Wikipedia, from users flagging it.

Scenario #01
A video gets flagged by five different users, the video gets hidden and checked by five different citizens or one honored citizen. For every flag one citizen is needed.
If it is decided that the video is indeed not in accordance with the TOS the video stays hidden and the account has to hand in 5 points from their account and the flaggers receive 15 points on their account. If however, the video is wrongfully flagged, it would stay up and all the flaggers have to hand in points and the accused account receives points. In order to discourage the overall practice of false flagging, positive rating is always adding more points than negative rating is taking away. This means that the intention to falsely flag a video could turn out to be beneficiary to the attacked account, which contradicts the initial intention.

Scenario #02
A video reaches an average positive rating of higher than four stars after 1000 users have rated it. The video receives immunity, after it is checked by (honored) citizens. The user that posted the video will receive 15 points on his/ her account, while the people who rated it do not receive points. This is to avoid users from rating all the videos five stars.

Very important in such a system is that all the benefits are non-monetary. It would generate a feedback loop between participation and community recognition, similar to Slashdot’s system of meta-moderation. “The system [of Slashdot] rewards both participation on the site and community judgment of that participation” .

Democratizing a complex area such as the moderation of any social network is a “thorny terrain in which spontaneous cultural participation intersects with complicated corporate and legal interests.” (David, 2007, p.115) For not only is it their interest, it is also the company’s responsibility to uphold any legal and corporate interest. That is why I start with democratizing the part of the system that is depended on the community, the flagging aspect.

[Cara Mertes questions the promise to ‘democratize television’ of a new network format Current:] What are you talking about when you say ‘democratizing the media? Is it using the media to further democratic ends, to create an environment conducive to the democratic process through unity, empathy and civil discourse? Or does it mean handing over the means of production, which is the logic of public access? (Jenkins, 2006, p.252)

Not only do I see the latter as a necessary predecessor of the first statement. In that sense I must confess that I am a bit of a technological idealist. But, as Mertes says, you can create the environment, but whether the users are ready for this and will use it to this potential is a preferable, though unpredictable outcome. One cannot force users into a civil discourse, but can lead the way.

The starting point of my proposal, which consists of four main points, would be addressing the issues of free speech and censorship. Videos will no longer be removed without leaving a trace and all comments rated ‘-6’ or lower that are now hidden for the user community should offer the option to be temporarily unhidden in order to read them. This same logic of ‘unhiding’ could be applied to videos in addition to a banner stating the reason why the video was flagged.

The second point would be by taking away the negative connotation of flags, the ‘I don’t agree with this’ aspect, by extending it with an ‘I do agree with this’ option. This could be achieved by a technological convergence of the effects of flagging and rating.

The third point would be proposing a reputation system to form as a non-monetary incentive for users to participate in the discussions that take place at YouTube. A key difference with the existing system is that not only videos are rated and/ or flagged. The user’s account builds up a reputation through the means of receiving or handing in points. The amount of points determine the privileges a user has.

The final point is derived from the third point, because the reputation system would for the basis for my proposed system of meta-moderation by the users. Both of which, I shall discuss in detail in the proposed model that follows.

Both Digg and Wikipedia’s software architecture tries to associate expertise with community authority; the latter defined as the ability to shape the participation of others within the online space. At Slashdot, expertise and authority are more granular, valid for a short time. In the book ‘Towards Participatory Culture’, Shay David makes the point that expertise in many online communities is not related to a single individual;

Community expertise resides in two competing dynamics:

(1) the legitimation of “aggregate” opinion, as opinions tend toward an equilibrium, even on controversial issues
(2) the openness of the community to dissenting opinions, with the potential to change the
aggregate consensus.

In successful online communities, these dynamics produce confidence in the knowledge-making process, rather than confidence in any particular instance of expert opinion. The system is legitimate to the degree that it represents a properly constituted authority in the eyes of the community of users. (David, 2007)

An expert in the YouTube community I would say is a user that knows which issues are at play amongst the community. Relating to David’s point on expertise, one could say that a diverse group of random active users can be part of a possible moderating hierarchy.

By looking at the existing hierarchical moderation systems, I see hierarchy amongst users as a proven non-monetary incentive for users to participate. Only ‘Funny or Die’ does not make use of this, but they place themselves (the staff) in the watchdog position. And this particular platform deals with a basic binary question of whether the viewer thinks a video is funny or not. In that sense Digg, but especially Wikipedia and YouTube are platforms containing broader discussions and opinions.

The controversy around the ‘Bury Brigade’ is somewhat similar to the reactions about the misuse of the flagging system on YouTube. Issues like accountability and transparency come up in both these issues. Whether or not the discussion is justified, it’s interesting that both these widely used services stir up such a discussion. Which can be because these users feel left out of the system because of the lack insight in the system and their the minor participative role they play in it. It is not similar to CL Pet controversy on Craigslist, because the people in that discussion blame each other instead of the system. The only aspect on which they blame the system is for not putting their TOS clear enough.

By not only introducing a certain form of hierarchy, but also by positioning the community in the role of the watchdog YouTube would take over certain qualities of both Slashdot as well as Wikipedia. Which results in a system of meta-moderation of flags – flagging of the users will be by the users, for the users – thereby taking the next step in the grassroots convergence on YouTube towards a more open, democratic platform.

On the website www.funnyordie.com, videos are rated on a scale of 0% to 100% funny by the user community. It’s a very simple system, which shows a close resemblance to Digg, and is based on whether the viewer thinks a video is funny or not. Every use can rate a video only once by clicking on either the ‘funny’ icon or the ‘die’ icon. When the ‘die’ icon is clicked, the video will stop and start to play a scene from random action movie in which someone gets killed. When the ‘funny’ icon is clicked nothing happens, except for when it’s clicked after the movie has ended. In that case it starts to play a random clip of a guy playing airguitar.
Videos can reach three different levels. The first level is ‘Chosen One’; this is decided by the staff and these videos have diplomatic immunity, so votes don’t affect them. This is comparable to the articles on Wikipedia that are fully protected from editing. Except at Funny Or Die this decision is not made by a higher level of users (‘administrators), but by company staff. The second level is called ‘Immortal’, which is “a badge of honor.” Achieved when the video has at least 100.000 views and is rated over 80% funny, earning it a immunity for other votes.

“Does this mean every Immortal video will be the funniest thing you’ve ever seen? Maybe not. But this a democracy and sometimes people vote for things that aren’t good (see: White House). So think before you vote.”

The third level is called ‘The Crypt’, which is reached when the video has at least 1.000 views and is rated under 20% funny. Once a video has reached either the first or second level of immunity, it can never be voted down to ‘The Crypt’. The other way around is also possible, videos in ‘The Crypt’ can be “resurrected” by voting them funny to get them out of ‘The Crypt’. What I personally find interesting about ‘The Crypt’ is what I already mentioned earlier about the rating of comments on YouTube and ‘disemvowelling’, the fact that the content is not removed from the website, but more sort of hidden and thus creating a collection of data that most of the users don’t want to see or read. First of, this approach both reveals the collective opinion of the community as well as silences anyone claiming to be robbed of their freedom of speech. And secondly, this avoids senseless discussions (for the “proof” is gone) about falsely flagged videos that are taken down and not mirrored. On this subject MIT started an interesting research project called YouTomb . This project by MIT Free Culture monitors a certain amount of videos taken down from YouTube for alleged copyright violation along with the metadata, including who issued the complaint and how long the video was up before takedown, with a particular emphasis on those videos for which the takedown may be mistaken. Revealing a critical view towards the YouTube flagging system.

Like the companies blog already mentions, it is suppose to be a democracy. But this does not mean that the whole moderation system depends on the user community as a blogpost from May 7, 2007 reveals. In which misuse of the program is reported about someone who created a program that generated fake votes. This shows that they’ve set up other ways, besides the voting system, to keep the site clear. But similar to Digg they actively promote user participation in order for the system to work.

“Digg is democratizing digital media”

At Digg.com logged-in users can give it thumbs up to an article he/she likes and ‘digg it’, contributing to the popularity of that item. “Once a submission has earned a critical mass of Diggs, it becomes ‘popular’ and jumps to the homepage in its category. If it becomes one of the most popular, it qualifies as a ‘Top 10’. If a submission doesn’t receive enough Diggs within a certain time period, it eventually falls out of the ‘Upcoming’ section.”
A user can also give it the thumbs down and ‘bury’ it. Opening a dropdown menu containing categories such as duplicate story, spam, wrong topic, inaccurate and ‘OK, This is Lame.’ The same logic is applied to the comment section of the site.
Digg offers the option to change your vote on a comment: “If you’ve voted down a comment that you now want to support, just hit the thumbs-up button once to neutralize your down-vote and then a second to record your positive vote time (and vice-versa).” This does not apply the same way to votes on an article.

“You can un-Digg a story by logging in and clicking My Profile. Find the story in the Recent Activity section (or click the History tab) and you’ll find an un-Digg button next to each item. You can’t unbury a story, however.”

It is not visible to other users which articles have been buried by a certain user, in contrast to articles he/she has Digged. The company does not reveal how the algorithm behind this whole process works exactly to prevent users from “gaming the system.” But it “takes several factors into consideration, including (but not limited to) the number and diversity of Diggs, buries, the time the story was submitted and the topic.” Due to the mystery of around the exact workings of the algorithm and the identity of the buriers, it has also been the topic of discussion amongst (part of) the Digg community that claim the existence of so-called ‘Bury Brigade’ misusing the ‘bury’ option in an organized, collective manner to bury certain stories. And critizing the ‘bury’ option for being undemocratic, because of its anonymous nature and unaccountability. Thereby showing a resemblance to YouTube’s flagging campaigns and the type of reactions by the community. In reaction to users gaming the system, Kevin Rose announced an update in the algorithm on the company blog in September, 2006. To prevent people from gaming the system he does not disclose how it exactly works, but the update includes a “unique digging diversity of the individuals digging the story. Users that follow a gaming pattern will have a less promotion weight.” Which means that it would take a more diverse pool of diggers to get a story on the homepage.

In the end digg is yours to control. No matter which stories are promoted, it’s up to the masses to digg or bury them. If you see content you disagree with – bury it. If enough people do the same, the system will automatically remove the story.

This could be seen as a point where co-existing forms of convergence briefly meet, the grassroots convergence demands attention for what they experience as a problem and the corporate convergence reacts by adjusting the algorithm. In its turn this adjustment received a lot of critique from the community as well. But convergence is not an end-point, but an ongoing process.

An article on Wikipedia is not only created collaboratively, but also maintained that way. By users of Wikipedia, which can be divided into different types of users;

Regular user levels
- Anonymous users
- New users
- Autoconfirmed users
Special user levels
- Administrators
- Bureaucrats
- Stewards

The regular user levels are appointed automatically by software and special user levels are granted by the community. Each type of user has its own specific privileges or constrictions, such as the ability to (semi-) protect pages or edit these type of pages. This feature is needed when so-called ‘edit wars’ arise around one particular article due to controversy.

Applying the so-called wiki-logic to decision making processes has been explored by law professor and information law expert Beth Simone Novek in her book “Wiki Government: How Technology Can Make Government Better, Democracy Stronger, and Citizens More Powerful.” In this book she describes how she worked with a team of experts to help inexperienced citizens participate in patent examination. They used the Internet to create communities of self-selected experts to contribute their time and effort to the policy process. And thereby, exemplifying its title, showing how technology can be used to create a participatory culture of government.

On YouTube there is not an official hierarchy amongst user accounts, with the exception of users that are part of the Partnership Program . But this only offers the possibility to monetize videos and does not offer any privileges in the sense of favoriting. In this chapter I will briefly address some existing systems of hierarchical moderation. I will not give an extensive analysis on each of them, because overall they are quite well known. Therefore I will highlight the exemplary aspects that I think are of interest and/ or applicable to YouTube.

7.1 Slashdot (1997): meta-moderation

Slashdot’s system of meta-moderation is a pioneer in its field by using crowdsourcing as a way to detect trolls. An random, automated system selects a few users providing them with the a temporary privilege to annotate comments as being ‘offtopic’ for example. Providing the comments with a score ranging from –1 to 5. Other users can adjust their filter, so they will not see the trolling comments and other spam. The concept of crowd-voting has worked so well that sites as high-traffic as the NYT now use it .

SafeSearch is a feature in every Google/Yahoo!/Flickr account that can be turned ‘off’, on ‘moderate’ or ‘on’ by the user. By doing so, the user allows the engine to filter the content that shows up in the results. This ‘filter’ focuses on pornography and explicit sexual content. With the filter on ‘moderate’ it blocks explicit images/video only, while fully turned on it blocks explicit text and images/video. Flickr has it set on by default. Yahoo and Google have the default set on moderate filtering.

With Flickr and Yahoo the geographical location is a factor in adjusting the filter. “If your Yahoo! ID is based in Singapore, Hong Kong or Korea you will only be able to view safe content based on your local Terms of Service so won’t be able to turn SafeSearch off. If your Yahoo! ID is based in Germany you are not able to view restricted content due to your local Terms of Service.” Something we also saw in YouTube’s self-regulatory model, which is partly based on a global/legal level as well. Except that was not communicated so directly to the user community.

Yahoo! SafeSearch cannot guarantee that as well that all explicit content will be recognized and filtered out of the results. While “Google’s SafeSearch system seems to reflect a focus on avoiding underblocking (i.e. avoiding linking to sexually-explicit content) at the expense of increased overblocking (omitting listings of content that is not sexually-explicit).”

Google’s SafeSearch works through continual crawling of the Web by a filter and by incorporating updates from user suggestions. Google’s filter uses advanced proprietary technology that checks keywords and phrases, URLs and Open Directory categories.

“Google’s help site suggests that SafeSearch is primarily driven by automated systems — computer classification of sites that are sexually-explicit versus not, in contrast to the human review purportedly used by companies like N2H2 (ref: N2H2’s “Human Review Advantage”). For lack of human review of SafeSearch’s filtering decisions, Google readily admits that its system is not completely accurate.”

But even though it’s an automated system, it does leave open the option for the users to request removal of a certain page with the Webpage removal request tool. This allows Google to improve the SafeSearch filter.

An interesting aspect is that on Google and Yahoo! the filtered content won’t show up in the results, but in Flickr they show it distorted, like this:

The distortion of the image is comparable to the warning screen on age-restricted videos on YouTube as well as the hiding of comments through rating. It’s visible to the community when it has been flagged and the choice to watch the video/picture anyway is made by the user self.

This means that in Google and in Yahoo! this SafeSearch function works similar to flagging. It controls the content on a site, but instead of deleting unwanted content SafeSearch holds it back from the viewer if he/she doesn’t which to see it. And this time it is not the community that decides which content isn’t conform wishes, but an algorithm makes this decision.

In the flagging procedures of Blip.tv, Dailymotion, Veoh and Yahoo the copyright infringement flag redirects the user to another form/instructions outside the flagging menu. 56.com, Google Video, Spike, Tudou and Youku don’t include copyright in their flagging options. Although 56.com does give the “other” option, which could serve this purpose.