Abstract
Videos stored on YouTube served as a valuable set of communicative resources for publics interested in the Occupy movement. This article explores this loosely bound media ecology, focusing on how and what types of video content are shared and circulated across both YouTube and Twitter. Developing a novel data-collection methodology, a population of videos posted to YouTube with Occupy-related metadata or circulated on Twitter alongside Occupy-related keywords during the month of November 2011 was assembled. In addition to harvesting metadata related to view count and video ratings on YouTube and the number of times a video was tweeted, a probability sample of 1100 videos was hand coded, with an emphasis on classifying video genre and type, borrowed sources of content, and production quality. The novelty of the data set and the techniques adapted for analysing it allow one to take an important step beyond cataloging Occupy-related videos to examine whether and how videos are circulated on Twitter. A variety of practices were uncovered that link YouTube and Twitter together, including sharing cell phone footage as eyewitness accounts of protest (and police) activity, digging up news footage or movie clips posted months and sometimes years before the movement began; and the sharing of music videos and other entertainment content in the interest of promoting solidarity or sociability among publics created through shared hashtags. This study demonstrates both the need for, and challenge of, conducting social media research that accommodates data from multiple platforms.
Notes
This article was originally published with errors. This version has been corrected. Please see Erratum (http://dx.doi.org/10.1080/1369118X.2013.769779).
Following Gillespie (Citation2010), although we recognize ‘platform’ as a complicated and ambiguous term, we use it here to describe online services, such as Twitter and YouTube, that operate according to particular terms of service, commercial interests, technological affordances and constraints, and social and cultural norms.
A common set of keywords was adapted for the peculiar characteristics of each service. For example, while the use of #hashtag syntax is native to Twitter, it is less commonly found in YouTube descriptions, titles, or tags.
Keywords were: #occupy, #ows, move your money, ows, occupy, occupy movement, occupy together, occupy wall street, we are the 99, zucotti. False positives for the search term ‘occupy’ were nearly eliminated by use of proximity word functionality in Radian6.
All URLs shared on Twitter are shortened at least once. For example, the short URL http://t.co/jxXSNAsr redirects to http://www.youtube.com/watch?v=u7OPv3y216Q&feature=youtu.be. However, third-party software, such as Echofon or Tweetdeck, may introduce additional rounds of shortening and, over time, links that are highly circulated among Twitter users may be shortened dozens of times.
Each video on YouTube is identified by a unique, case-sensitive, eleven-character ID, for example, 70QzGFWumZQ.
The exact search terms differed slightly between Gnip PowerTrack and Radian6 to account for differences in each tools and in the particular data being parsed. For example, hashtag syntax (e.g. #occupy) is native to Twitter and appears less frequently on YouTube.