allows users within an enterprise to share files, to create ..... from a separate application. ..... webapps/ how-to-mea
We are all Lurkers: Toward a Lurker Research Agenda Michael Muller, David R. Millen, N. Sadat Shami, and Jonathan Feinberg IBM Research One Rogers Street, Cambridge, MA, USA 02142 {michael_muller, david_r_millen, sadat, jdf} @ us.ibm.com +1-617-693-4235 ABSTRACT
Most analyses of knowledge repositories focus on the role of knowledge-creators, as we did in [28]. Users who do not contribute knowledge are usually termed "Lurkers," and are sometimes called "free-loaders" because they take information from the shared resource, but do not contribute anything in return [10, 15]. There are several reasons now to question this perspective.
Most knowledge repositories focus on the role of knowledge-creators. Recent commercial and research reports have begun to highlight the role and contributions of lurkers, or "non-public participants." We examined the work of Lurkers in an enterprise file-sharing service, and we compared their lurking behaviors to the lurking behaviors of users who uploaded files (Uploaders), and users who contributed metadata about files (Contributors). As predicted, Uploaders engaged in more frequent lurking activities, followed by Contributors, with Lurkers as least active. Surprisingly, the structure of the lurking activities was very similar across all three roles, suggesting that there are no distinctive task-appropriate forms of lurking that reflect the goals and strategies of Uploaders and Contributors. We use these results, and the published research literature, as a source of research questions in collective intelligence.
In the business research literature, recent reports by Li and Bernoff [20], Porter [37], and McDermott [21] have highlighted the commercial importance of users who do not create primary content. Their lurking behavior, properly measured, can make indirect contributions, including evidence of the usefulness and perceived value of each resource (i.e., consumption of a resource becomes a metric for its perceived worth). Behaviors that occur before and after lurking can also provide value. For example, a summary of search terms can give an advance notice of what people are interested in. This kind of information can help knowledge-creators to serve the needs of their readers, or can help an organization to anticipate future needs or compete with other organizations to fulfill those needs earlier or more effectively. More controversially, an individual user’s lurking data can be used to customize services for that user, including the delivery of advertisements and other targeted content [17, 33]. Privacy-protection rules may affect the appropriateness of these commercial strategies in some cultural or national settings.
Author Keywords
Collective intelligence, Lurker, Non-public participant, Social software, Social media, File-sharing ACM Classification Keywords
H5.3. Group & Organizational Interfaces: CSCW. INTRODUCTION
This paper addresses the workshop theme of "Designs of new software tools or proof-of-concept prototypes supporting CI in task forces, communities; or in-depth evaluations of tools already deployed that support CI in organization" through an analysis of lurking behaviors in an enterprise file-sharing service called Cattail [27, 28]. We briefly review earlier studies of this system and similar systems. We then focus on the role of lurkers in such systems, including an empirical comparison of the lurking behaviors among users in several different roles. We use these results to motivate a research agenda focused on lurkers and lurking behaviors in collective intelligence.
Within CHI and CSCW, the work of Preece and colleagues [30, 31, 38], and of Tahakashi and colleagues [45], has shown the motivations and subtle contributions lurkers, called "non-public participants" by Nonnecke et al. [30] For example, some lurkers deliberately decline to contribute for altruistic reasons, principally the desire not to clutter an already-filled shared information space. Lurkers may also feel that they lack the standing to make useful contributions, believing that the act of adding new knowledge is appropriate for people in some roles, but not in other roles. In a partially-convergent research programme, Takahashi et al. explored how lurkers first find information from the shared resource, and then share that information to other, indirect users outside of the bounds of the shared resource [45]. In these ways, lurkers add to the effectiveness of the knowledge-creators and of the shared repository as-a-whole, through increasing the number of
Position paper submitted to CSCW 2010 workshop, Collective Intelligence.
1
L H G
D A
I E F B
J
K
C
Figure 1. One aspect of the Cattail user interface, showing the files related to a particular user. Grey ovals obscure the names of users to protect their privacy. A. List of files uploaded by the user. B. One file has been selected for a more detailed view, or for download. C. More detailed information about the selected file. D. A menu of other possible actions. E. Link to a view of files shared to the user from other users. F. Link to files that have been shared from the user to other users. G. Link to recent events in the user’s social network. H. Link to a list of all public files (which can be viewed in order of popularity or recency). I. Link to collections with which the user is associated. J. Ability to share a file to other user(s). K. Ability to add a file to a collection. L. Ability to search all accessible files (public plus private files on which the user has permission) by tag or content. people who are influenced by the information in the shared resource. The Cattail File-Sharing Service: A Site to Study Lurking
We studied the activities of lurkers in a research prototype enterprise file-sharing system. Unlike most productiongrade services, this prototype had been instrumented to capture not only the conventional write-based events (uploading, commenting, tagging, etc.) but also view-based events (downloading, viewing, searching). The Cattail file-sharing system is a centralized service that allows users within an enterprise to share files, to create metadata about those files, to search for files, and to download files (Figure 1).1 More specifically, users may engage in the following types of activities: •
Upload a file
•
Contribute metadata on a file: o Comment on a file o Share a file (recommend a file to another user)
1
For the purposes of this position paper, Figures 1 and 2 have been copied from [28]. Please note that the lurker analyses in this CSCW workshop position paper have not been reported previously.
o
Create a named collection of files, and add a file to a collection
Make use of a file or metadata about the file (lurk) o Download a file o View metadata about a file o View metadata about another user o Search for files A file in Cattail may be classified as public or private. A public file may be viewed or downloaded by any user. A private file may only be viewed or downloaded by users who have access to that file. Access is automatic for the person who Uploads a file. For private files, access can be granted to other users through the action of Sharing the file to named users (the user operation of sharing performs two system operations: [a] it adds the shared-to user to the permissions list for the file, and [b] it notifies the shared-to user that s/he has access to the file, optionally with an informal message from the sharer). A view onto the sharing data for a file is shown in Figure 2. Any user with access to a file may share that file to any other user. In some cases, a user has shared a public file to specific other users in order to call their attention to that file. •
In an earlier report [28], we provided an overall analysis of the pattern of activities in Cattail, including the following four activity patterns:
MM
N N N P
Q
Figure 2. Detail of sharing information about one file. M. Information about the shared-from and the current sharedto users. N. Information about previous pairs of shared-from and shared-to users. P. Record of which users have downloaded the file. Q. Information about which collections the file has been added to (a single file may be included in zero, one, or many collections). •
Uploading one or more files, and publicizing them to other users
•
Annotating a file and monitoring it for actions by other users
•
Finding files that had been uploaded by others, and telling additional users about those files
files. Contributors may also perform the same actions as lurkers. •
METHOD, RESEARCH QUESTION, AND HYPOTHESES
Our dataset consisted of user actions in the Cattail enterprise file-sharing service during the period 26 November 2008 through 5 June 2009 (n=133708 users). There were 17982 (13.5%) Uploaders, 19103 (14.3%) Contributors, and 96623 (72.2%) Lurkers. These categorizations proceed from the broadest to the narrowest sets of behaviors. Thus, most of the 13.5% of Uploaders also Contributed and Lurked. Most of the 14.3% of Contributors also Lurked, but none of the people in the Contributors category performed any Updates. Finally, the 72.2% of Lurkers were “pure” Lurkers, and performed no Uploading or Contributing.
• Refinding one's own files We have also examined the use of collections of files in Cattail [27], highlighting the work of creators of those collections, whom we described as "information curators." The two reports tended to focus on the activities of file creation and discussion (see also [2, 19, 40, 47, 48] for other studies of file-sharing systems). As with many studies of social media, we paid less attention to the less visible activities of lurking -- i.e., activities that do not contribute files to the shared repository, and that do not create visible metadata about those files. In this report, we take a closer look at lurking behaviors in Cattail.
Research Strategy
Following the research and commercial insights of [2, 19, 20, 21, 37, 40, 47, 48], we compared the lurking activities of three categories of users of Cattail: •
Lurkers never deliberately add information to the database, but they do engage in “non-public” actions [31] such as downloading files, viewing metadata about specific files, viewing lists of files, viewing information about other users, and searching for files
•
Contributors do not upload files, but they do create metadata about files through actions such as commenting, sharing to specific other users, adding files to named collections of files, and adding tags to
Uploaders create files in the service. Uploaders may also perform the same actions as Contributors and Lurkers.
In this paper, we wanted to compare Uploaders, Contributors, and Lurkers on a core set of behaviors that are common to people in all three categories. We therefore excluded behaviors of Uploading and Contributing, because Lurkers (of course) perform none of those behaviors. We restricted our analysis to Lurking behaviors only, because people in all three categories were likely to lurk from time to time. Our question thus becomes: Do users in different categories engage in different Lurking behaviors? This narrowing of focus led to a set of 18 lurking activities in our study (see Table 1). Research Question 1: What are the patterns of lurking behavior?
3
Variable
Explanation
Download
The user downloads a file
Download-remote
The user downloads a file via a remote access (feed, blog, or post)
ViewFile
The views metadata about a file
ViewFile-remote-feed
The user obtains metadata about a file via a remote access through a feed
ViewFile-remote-blog
The user obtains metadata about a file via a remote access through a blog
ViewFile-Share-Email
The user is notified via email about a shared file, and follows a link in the email to view the file’s metadata
ViewFile-Version-Email
The user is notified via email about a new version of a file in which s/he has an interest (through the operation of uploading, commenting, sharing), and follows a link in the email to view the file’s metadata
ViewHome-Date
View the user’s files in recency order
ViewHome-Type
View the user’s files in order of type of document (e.g., pdf, text, slides…)
ViewInbox
View files shared to the user
ViewOutbox
View files shared from the user
ViewDyadSharedToMe
View files shared to the user from a particular other user
ViewDyadShared FromMe
View files shared from the user to a particular other user
ViewOneCollection
View the files in a selected collection of files
ViewAllCollections
View the names of all collections to which the user has access
ViewPerson
View the files associated with another user
ViewPublic
View a list of all files that have public access
Search
Search the files by tag and/or by content
Table 1. Lurking activities in the analyses.
It seems likely that Uploaders engage in lurking behaviors in support of their uploading activities (e.g., determining which files are available, and then new files to supplement or fill in empty categories). It also seems likely that Contributors engage in lurking behaviors in support of their contributing (e.g., finding files to collect, to share/recommend, to comment on, and/or to tag). Therefore, we propose Hypothesis 1: It seems likely that there will be different amounts of lurking behaviors across the three different user roles of Uploaders vs. Contributors vs. Lurkers. Hypothesis 2: It seems likely that there will be different patterns of behaviors (different combinations of lurking actions) across the three different user roles. RESULTS
We restricted our analysis to consider only Lurking activities. That way we could make direct comparisons of all three user roles in the activities that all three roles engaged in. Amount of Activity
Using an ANOVA, we found that Uploaders were most active in their lurking behaviors, followed by Contributors,
and finally Lurkers (p