AUTHOR
Richard Mills
ABSTRACT
Social News websites deploy voting systems whereby their community of users rank and sort large volumes of content collectively without explicit organisation or editorial control. The primary focus of this paper is the website reddit.com. On reddit, individual users can submit items of content (either links to web resources or text posts) and also vote (up or down) on the items submitted by other users. This voting system is also applied to comments on Posts. On any given day tens of thousands of users participate in content submission and voting; with the aggregate of this activity producing a ‘front page’ list of the 25 items which are seen by a large number of visitors to the website. Through the voting interface large numbers of users make a collective decision about which posts are displayed on the front page. Furthermore, each of these posts is accompanied by a democratically mediated discussion which is contributed to by thousands of individuals.
This paper will present research on reddit.com conducted using a variety of methods. As with any Social Media site reddit’s web servers contain precise and accurate records of all activity taking place on the website. Reddit’s administrators have provided one month of voting data from their servers for the purposes of research (henceforth ‘back-end’ data). This back-end data allows for questions relating to patterns of behaviour on the website to be addressed. However, back-end data has been anonymised with respect to post identity; making it unsuitable if we wish to consider qualities of the content being ranked. Reddit (again in common with many Social Media applications) provides an Application Programming Interface (API) which can be used to extract and store data. This paper will present details of the data extraction process, and offer examples of research questions which are more suited to back-end and front-end data respectively.
The interfaces deployed by Social News websites bear some similarity to those of Social Bookmarking websites. The major dissimilarity between Social News and Social Bookmarking is the degree to which Social News websites focus attention on the website’s front page; and absence of tagging on Social News sites.
This paper will present the following analyses related to back-end and front-end data obtained directly from the website.
- The hypothesis that Social News websites focus attention on their front page is supported by a Power Law (with exponential cut-off) distribution of votes between posts. This is seen as key to allowing a sense of website identity to emerge among its large user-base.
- The lifespan of a post is analysed in great detail, with results suggesting that early voting (within the first hour) is a strong predictor of the post’s ultimate success of failure.
- The relative attention received (in terms of votes and also hits) by posts in different areas of the website is analysed; suggesting that the front page is indeed where most of reddit users’ attention is focused.
- Patterns of User behaviour are studied; and it is revealed that people use their functionally identical accounts to participate in diverse ways.
- Several strategies are also identified whereby users seek to maximise the impact of their votes in determining what the front page will look like.
Content analysis has also been employed in an effort to understand what the broader significance of reddit is with respect to public discourse on the Internet. The background to this is the literature about public discourse on the Internet (e.g. Benkler, 2006; Negroponte, 1995, Sunstein, 2002; Noam, 2003). Case studies are being conducted on reddit’s coverage of the Wikileaks diplomatic cables (November/December 2010) and the current Egyptian protests. The primary aim of these case studies is firstly to determine how coverage of the stories on reddit differs from that of conventional news outlets. Preliminary results suggest that many posts on reddit link to conventional news outlets, with these articles being discussed through the commenting system. There are however also posts which speak to aspects of the story not covered by conventional news outlets; and text posts by users which espouse particular points of view; ask questions of the community; or which call on reddit users/visitors to engage in some form of collective action related to the ongoing story. A large proportion of posts about these subjects address the conventional news’ coverage of the story directly; often these are highly critical. This suggests that Social News websites may be of particular importance to issues surrounding the convergence of old and new media (Jenkins, 2006).
The method employed in these case studies involves looking at the profile of submissions on the topic, and comparing this to the voting reception each submission receives. A secondary aim of these case studies is to establish whether reddit’s voting interface is capable of expressing conflicting points of view; for example if I see a pro-wikileaks post on the front page does this mean I am unlikely to also see an anti-wikileaks post appearing here? What proportion of posts about the subject are for or against the topic, and is this proportion reflected in the number of posts of each persuasion that reach the front page? These questions relate to issues of fragmentation and polarisation of public discourse on the Internet (e.g. Negroponte, 1995).
Finally, this paper will present the results of an ongoing experiment which looks at how well reddit’s voting interface performs on the task of sorting large volumes of content such that the ‘best’ content (as judged by reddit users) has the highest rank and will be seen by the most people. This experiment also looks at whether an item’s voting response on reddit effects reddit users’ perception of the content; and if so how quickly this relationship is learned/un-learned.
REFERENCES
Y. Benkler. The Wealth of Networks: How Social Production Transforms Markets and Freedom. Yale University Press, New Haven, 2006.