Opening Note

As we have mentioned in recent issues of the C3 Weekly Update, May 8 and 9 we are holding the annual C3 Retreat, which is an opportunity for all of you to join us in Cambridge to talk about the work we’ve been completing, about new developments on the Consortium’s agenda, and where we’re heading in the future. During this time, one of the projects we’ll be discussing is the work we’ve been doing on YouTube. I wanted to take a brief moment this week to provide an introduction to this project, which we’ve talked about informally on both the blog (for instance, here and here) and in Editor's Notes in previous Weekly Updates.

We set out to try to gain a better understanding of what YouTube “is." The site is among the most well-known and widely used commercial websites built around user-created content to emerge so far, and it is quickly on its way to becoming a common-sense object. Like MySpace, Wikipedia and Facebook, YouTube has come to occupy a central place in our shared understanding of popular uses of the Internet. At the same time, however, just exactly what is summed up or referred to when the name “YouTube” is invoked is unclear. As it rapidly approaches status as a somewhat generic term for ‘online-video,' it remains a multiplicitous object, defined by particular and individualized uses. It is positioned as a major player in a new industry; as a disruption to ‘big media’ hegemony and expertise; as a site of ‘wacky and wonderful’ vernacular expression and user-led innovation. Truthfully, YouTube functions as all of these things - YouTube's functions as both a 'top-down' platform for the distribution of commercial popular culture and a 'bottom-up' platform for vernacular and 'amateur' creativity.

This tension is especially productive for understanding the role YouTube plays in the shifting landscape of popular culture. It is understood as a site that can make the products of commercial media widely popular, challenging the promotional reach the mass media are accustomed to monopolising, while also a site where “challenges” to commercial popular culture might emerge (be they user-created news services or generic forms such as vlogging, appropriated and exploited by the LonelyGirl15 team). Understanding YouTube’s status as a platform for popular culture therefore requires an analytical understanding of the interplay between these two roles - as both distributor and originator of popular culture.

Our ambition to contribute, however tentatively, to a more comprehensive and complex understanding of how YouTube works as a site of popular culture at a particular historical conjuncture presents epistemological and methodological challenges to media and cultural studies. Such an approach requires dealing with both (local) complexity and (global) scale, within the reality of constant change and growth. The methods of cultural and media studies are particularly adept at dealing in the former, but at least in an empirical sense, we are frequently confounded by the latter. Scale at the level which YouTube represents seriously tests the limits of the explanatory power of even our best grounded or particularist accounts, especially if we want to intervene in or influence debates outside our own disciplines. Working from a media studies perspective, if we determined at the outset that we were interested in exploring 'remix culture,' or music fandom, or DIY cooking shows, or any number of other niche uses of YouTube, we would be sure to find sufficient examples among the more than 75 million videos (and counting) that are currently available in the YouTube archive. But doing so would not tell us nearly so much about how YouTube works as it would about how that particular niche genre or subculture works. Therefore, it is not enough for our purposes identify in advance a small, localised sample of texts or groups of users on the basis of an existing set of scholarly or subcultural enthusiasms, or a predetermined political stance.

But, perhaps unsurprisingly, approaches to YouTube that attempt to comprehend it as a 'system' have, so far, been restricted to the 'hard' end of social science - usually, from computer science and informatics, employing methodological tools like network analysis. See, for instance, Meeyoung Cha, Haewoon Kwak, Pablo Rodriguez, Yong-Yeol Ahn, and Sue Moon's "I Tube, YouTube, Everybody Tubes: Analyzing the World's Largest User Generated Content Video System, as part of the IMC'07 proceedings October 24-26, 2007, in San Diego. Also see, Phillipa Gill, Martin Arlitt, Zongpeng Li, and Anirban Mahanti's 2007 piece "YouTube Traffic Characterization: A View from the Edge," from the Proceedings of the 7th ACM SIGCOMM Conference on Internet Measurement in San Diego, pp. 15-28, and Martin Halvey and Keane T. Mark's 2007 piece "Exploring Social Dynamics in Online Media Sharing," from the Proceedings of WWW 2007 May 8-12, Baniff: Alberta, Canada, pp. 1273-1274.

These studies are used, for instance, to reveal content patterns, explore the popularity “life-cycles” of videos across the site, and certain behavioral patterns of viewing. This approach draws heavily on the architecture of the site itself, however, resulting in maps of the system that tend to reproduce the ontologies imposed by the design of the website, and therefore allowing less obvious, more hybrid and context-specific findings that are the object of critical cultural analysis to fall between the cracks.

Such studies tend to rely, for instance, on the categorization and tagging systems YouTube provides, which enable uploaders to describe and sort their videos by content, theme and style. Both of these strategies pose particular challenges for categorizing or analysing the content and popularity across the site. The categories YouTube provides categories, with titles such as “Pets & Animals” and “Cars & Vehicles," at best offer a (very) general framework for organizing content across the site. They suffer from some of the limits second-order ordering stratgies (see pp. 18-19 of David Weinberger's 2007 book Everything Is Miscellaneous: The Power of the New Digital Disorder). They’re necessarily broad and unable to contain much information about the videos themselves - they tell us little that is useful about genres, aesthetics, or the modes of communication that are associated with them. Videos on the site are further organized through a user-defined tagging system, which provides findability and a greater degree of categorization. This system is not only used descriptively but often strategically by uploaders. As the business model of the service continues to reward page views, applying ‘popular’ tags to content, marking videos as a response to popular but unrelated content, and other methods of ‘gaming’ the YouTube system have emerged. As such, these tags should not be treated as a matter of fact system for describing content within the system.

We set out, then, to find a middle ground between the specialized reading approach of cultural studies and the broad sweep of the ‘hard’ sciences. Searching for this middle ground, we adapted the methods of content analysis to examine YouTube in a manner approaching the breadth of quantitative methods as developed for communication studies and throughout the social sciences while providing the capacity to engage in the critical-interpretive analysis afforded by media and cultural studies. Our methods are explained in more detail below, but in short, we undertook a survey of more than 4,000 of the 'most popular' videos from a period of several months, using methdological tools derived broadly from content analysis. As humanities-oriented researchers, this activity gave us a way to order a relatively large body of raw material in a way that would allow us to identify patterns across the sample, as well as to interrogate clusters of individual texts using our much more familiar qualitative methods - textual analysis, ethnography, forensic analysis; but, again, without selecting these smaller sub-sets of data in advance. This strategy has been useful for identifying controversies and mapping aesthetic characteristics across particular cultural forms.

Accounting for Popularity

Investigating how the site accounts for “popularity” requires assessing more than just the videos which are “most watched." Content is organized across the site according to a range of different measures, from those which have been viewed the most times to those which have been uploaded most recently. To balance depth of analysis with the ability to cover a broad sample of the content, we set out to assess YouTube’s ‘common culture’, looking at the videos which appeared at the top of a range of YouTube’s popularity measures.

Between March and December 2007, we capture an image each day of the "Most Viewed," "Most Responded," "Most Discussed," and "Most Favorited" pages on YouTube, recording details such as the name of the uploader and how many times the video had been viewed, favorited, commented upon or responded to. We captured this content at 11pm US Eastern Standard Time each day. The regionalization of YouTube in June 2007 divided the service into 19 different geographical regions. Each of these maintains its own lists of content relative to the region and the respective time zone. There is no region for the US, however; the US is included as the “Global” region. While this might suggest content drawn from across the globe, content from North America predominates the sample. Comparative analysis between material collected in each of the regional categories and some sense of the differing numbers of videos uploaded within each region would be incredibly fruitful to understand the international status of YouTube.

Each of these four categories represents a different measure of popularity. Whereas Most Viewed most closely resembles the aggregate measures of attention utilized by mainstream media industries as a way of counting “eyeballs in front of the screen.” The category counts only full views and counts views from external embeds once per IP address according to tests run in 2007 by a video analytics company (TubeMogul, 2007). Each of the other three measures provide some account of popularity based on activities that signal a degree of participation in the YouTube community -- if nothing else, all of them require the user to have an account. The Most Favorited cateogory measures the videos popular enough to be added to a user’s profile and Most Discussed the videos that generate the most comments, whereas Most Responded records the video viewers were prompted to post a video response to, either by filming their own material or laying a trace to another video in the system. So while the Most Viewed category relies on basic quantitative measurements of 'eyeballs,' the categories Most Responded, Most Commented, and Most Favorited move our assessment of popularity beyond the quantitative assessments of attention over time that have predominated notions of success in the broadcast era.

To get a richer understanding of the types of content on YouTube, we selected three months out of this sample (August, October, and November) and developed a coding system to categorize the videos according to textual and extratextual features. Sampling six days over two weeks in each of these months, we examined 4,320 videos, coding for things such as the most likely origin of the content (whether it was produced inside or outside of the traditional media system, for instance), the nature of the uploader, and the type or genre of the videos (whether scripted television footage, say, or a personal vlog; a fan mash-up, or promotional content of some kind). The C3 team working in conjunction with Dr. Jean Burgess (who some of you met at the retreat last year) and Eli Koger at the Queensland University of Technology coded these videos over December 2007 and January 2008.

In the time since, we’ve been analyzing this data, looking at which sorts of content, uploaded by whom, were popular where within YouTube, in turn developing a picture of YouTube as much sensitive to the range of content available on the service as it is the particularities of specific types of material. The results of this analysis we’ll present at the Partner’s Retreat in May. We hope to see you there.

Glancing at the C3 Blog

Follow the Blog

Don't forget – you can always post, read, and carry out online conversations with the C3 team at our blog.

Closing Note

"Everyone knows" that to make money off of digital remix productions is illegal, that it violates copyright. But how did "everyone" come to "know" this?

Digital appropriation began in the mid-1980s, when the first relatively affordable digital samplers hit the consumer market, and inner-city black DJs began to use these early digital devices to make new compositions out of the bits of older recordings. When digitally sampled sounds began to be featured on major music hits, they were not immediately judged to be copyright infringement. For the first five years of digital sampling's existence, between 1986 and 1991, many copyright holders whose works were sampled without permission sued sampling artists during this five-year period, but none led to a court ruling being handed down; all lawsuits were settled out-of-court.

Then, in December, 1991, a New York District Court judge, Kevin Duffy, ruled in the case of Grand Upright vs. Warner Bros. Records, which concerned the question of rapper Biz Markie sampling a few bars from Gilbert O'Sullivan's 1970s hit, "Alone Again, Naturally." Duffy defined unlicensed sampling as uncategorically and unquestionably equivalent to theft. His ruling stated that: 1) all unlicensed sampling is stealing, and artists who want to use samples in their work must obtain a license from the source work's copyright owners; 2) the right to determine how much to charge for a sample lies entirely with the copyright holder.

However, for a brief period, before U.S. courts swiftly decided the law's antagonism to sampling (and, subsequently, all forms of digital remix), there was a moment when a "Fair Pay" structure for digital remix was advocated by multiple voices in the legal community. I have identified eight articles that were published in law journals between 1989 and 1992 which recommend Fair Pay guidelines for sampling: one in the High Technology Law Journal (now The Berkeley Technology Law Journal) in 1989; one in the Loyola of Los Angeles Entertainment Law Review in 1991; one in the Harvard Law Review in January 1992 (published just one month after the Grand Upright decision, and composed before that ruling was handed down); and a series of four articles that appeared in the New York Law Journal in 1989. Here are two key themes shared by all of the Fair Pay proposals have in common:

• A concern for balancing the interests of copyright holders with the interests of digital sampling artists, which takes for granted that sampling artists have the right to develop their new art, which has been made possible by a new technology. The authors firmly assert that there is a "need to protect [both] the original works and the value of sampling....Cooperation between samplers and copyright owners based on mutual respect and financial interest is the answer" (Gordon and Sanders, New York Law Journal). "At a policy level, the balance is a struggle between the right of artists to control their own work and that of unencumbering the creative opportunities inherent in a new technology" (Mc Graw, High Technology Law Journal). "Private agreements are unpredictable and probably unfair in some instances – either to the sampled artist who obtains less or to the sampling artist who pays more than he might have....A requirement that sampling artists pay for the use of samples is fair...because [it] ensures both that new artists are not ‘stealing' and that artists will not be discouraged from producing new songs" (Harvard Law Review). "The ad hoc approach within the music industry...has unnecessarily inhibited the growth of a process of creating music that has many artistic and economic benefits" (Loyola Entertainment Law Journal). Let us note that the "ad hoc" system of determining licensing fees not only was not remedied by the Grand Upright judgment, it was awarded an infinite lifespan; the current method for determining how much a remix artist must pay in order to obtain a license to use a copyrighted work is determined on a case-by-case basis solely by the copyright-holding individual or corporation.
• A licensing fee scheduling that assigns different prices for different types of sampling, based on both quantitative and qualitative measures. The High Technology article calls for "an appropriate test for establishing substantial similarity between" between the copied work and the copying work." The Loyola article states that the "specific qualitative aspect that courts tend to focus on in determining whether an appropriation constitutes copyright infringement is the commercial value of the portion appropriated"; for example, the chorus of a song is often "valuable because it is distinctive and memorable." The Harvard piece proposes four criteria for ascertaining the monetary value of a sample: "the popularity of the prior work as a whole, the importance of the sampled portion to the prior work, the duration of the sample, and the importance of the sample to the new work." The maximum statutory licensing fee, the Harvard article also suggests, should be the equivalent of the fee required for a cover song (in 1991, this was 6.25 cents per phonorecord distributed).

According to current interpretations of copyright law, no nuances between types of digital appropriation exist. In the contemporary legal climate, built on the foundation of the 1991 Grand Upright decision, all artistic appropriation is piracy, theft, and infringement, no matter what portions of a prior work are copied or how those copied portions figure into the new work (in fact, the 2004 ruling in Bridgeport vs. Dimension Films stated that a two-second guitar chord, which was lifted from a previously existing song and altered to be all-but-unrecognizable, constituted copyright infringement). The only safe "loophole" for remixers is the one of zero financial gain: "If you remix, it is illegal, but if you don't make any money, no one will sue you" is now the widespread common understanding of copyright law.

The conclusion to this piece will appear as the Closing Note in next week's C3 Weekly Update.