We conducted a principle components analysis of users' actions in an enterprise file-sharing service. We describe four factors, their attributes with respect to social action and awareness, and their implications for design.
Usage Of Enterprise File Sharing Service Muller Chi 2010
1. Patterns of Usage
in an Enterprise File-Sharing Service:
Publicizing, Discovering, & Telling the News
Michael Muller, David R Millen, & Jonathan Feinberg*
IBM Research / Collaborative User Experience
IBM Center for Social Software
Cambridge, MA, USA
* Jonathan Feinberg is now at Google.
IBM Research 1
2. Agenda
• File-sharing in an enterprise
• Factor analysis of file-sharing usage patterns
– Upload & Publicize
– Annotate & Watch
– Discover & Tell
– Refind
• Interpretation
– Both more and less social than anticipated
– Design what and for whom?
– Who is missing? Future analyses
IBM Research 2
3. File-Sharing in an Enterprise
• Cattail, an enterprise file-sharing service
– Centralized
– Authenticated for all access
– Simpler than peer-to-peer or public networks
• Cattail statistical profile
– Users 88270 (many job roles, 80 countries)
– Uploaders15943 (18%)
– Files 120288 (diverse formats)
– Downloads 728509
• Related work
– Dropbox / UD Dropbox (Schwartz, 2007)
– Apocrita (Reynolds et al., 2005)
– Studies of file-sharing (Christin et al., 2005; Lee, 2003; Rader, 2009;
Voida et al., 2006; Whalen et al., 2008)
IBM Research 3
4. Cattail Features and Capabilities
• Capabilities
– Upload – Share (recommend) *
– Download * – Collect *
– – Annotate *
– – Watch *
* Can be done on own files or on files that were uploaded by others
A F
H
B
P
C D
Q
G
E
B
IBM Research 4
5. Factor Analysis
• Review of the method of factor analysis
– Start with too many variables
– Find subsets of variables that are highly correlated with one another
summarize each subset of variables as a factor
– Continue the analysis in terms of the smaller number of factors
– Figures of merit: Eigenvalue & variance-accounted-for
• The 10 variables in our factor analysis
On own file On file of other
Upload
Download
Share
Collect
Annotate
Watch
IBM Research 5
7. Factor Analysis Results
• Upload Publicize (15935 users, 18.1%)
Upload + Share-own + Collect-own
– Interest in an audience for own files
• Annotate Watch (9780 users, 11.1%)
Annotate-own + Annotate-other + Watch-other
– Adding value / reviewing rating
– Social awareness
• Discover Tell (85303 users, 96.7%)
Download-other + Share-other + Collect-other
– Interest in an audience for others’ files
– Personal utility of collections
• Refind (5149 users, 5.8%)
Refind
– Download own files - Cattail as an extension of the user’s local drives
IBM Research 7
8. Interpretation: How do people use file-sharing in enterprise?
• Both more social and less social than anticipated
– More social
• Sharing, Collecting for use by others
– Upload Publicize
– Discover Tell
• Similar to “Information Curators” finding (ECSCW 2009)
• Apparently easy to organize (confer Rader, 2009, in public networks)
– “Yours, mine, and… ours” appears to be possible in enterprise sharing
– Less social
• Refind
– Public-access files with no viewers a sharing-failure case?
• Users appropriated the technology for unanticipated purposes
(Dourish, 2001; Kujala Kauppinen, 2004)
– Impression management (see also Thom-Santelli et al, 2008)
– Curating (ECSCW 2009)
– Refinding
IBM Research 8
9. A UI for each Factor? Or combined UI?
Factor Upload Annotate Discover
Refind
Publish Watch Tell
Upload 4301 13269 5143
-
Publish (4.9%) (15.0%) (5.8%)
Annotate 4301 9368 2054
-
Watch (4.9%) (10.6%) (2.3%)
Discover 13269 9368 4640
-
Tell (15.0%) (10.6%) (5.3%)
5143 2054 4640
Refind -
(5.8%) (2.3%) (5.3%)
15935 9780 85303 5149
Total
(18.1%) (11.1%) (96.7%) (5.9%)
• Promote files + • Structured • Promote files +
collections discussion collections • Synchroniza-
• Impression threads / file • Impression tion (private
Potential
management • Has my management disk £ Cattail)
Features
• Ratings by annotation • Ratings by • Ego-centric
readers been read? By readers filters
• Search terms whom? • Search terms
IBM Research 9
10. A UI for each Factor? Or combined UI?
Factor Upload Annotate Discover
Refind
Publish Watch Tell
Upload 4301 13269 5143
-
Publish (4.9%) (15.0%) (5.8%)
Annotate 4301 9368 2054
-
Watch (4.9%) (10.6%) (2.3%)
Discover 13269 9368 4640
-
Tell (15.0%) (10.6%) (5.3%)
5143 2054 4640
Refind -
(5.8%) (2.3%) (5.3%)
Same features Many users
15935 9780 85303 5149
for Total different
two (18.1%) (11.1%) (96.7%) engage in more
(5.9%)
factors than one factor
• Promote files + • Structured • Promote files +
collections discussion collections • Synchroniza-
Three factors
Three factors
• Impression threads / file • Impression tion (private
deal with
Potential
deal with groups management • Has my management disk £ Cattail) of
awareness
Features
of artifacts • Ratings by annotation • Ratings by • Ego-centric
actions by
readers been read? By readers filters
• Search terms whom? • Search terms others, with
impression
IBM Research management 10
11. Summary of Contributions
• Four factors in file sharing
– In the enterprise (and beyond?)
– Patterns of sociality
– Patterns of individual utility
– Appropriation of technology for new purposes
• Implications for design
– Unified user interface, supporting
• Work with collections of artifacts
• Impression management
• Ratings by readers
• Awareness of the work of others
IBM Research 11
12. Who is Missing? Future Work
• Analyses so far
– How do people use file-sharing in the enterprise?
• Four factors in usage (this note)
– “Information curators” (ECSCW 2009)
– Focused on people who changed the database
• Upload original content
• Contribute: Share (recommend), Annotate, Collect…
• (also Downloaders)
– Our four factors are primarily about visible contributions to the system
• Lurkers in file-sharing
– Lurkers are the modal users (75%)
– How do lurkers use file-sharing?
– What do lurkers need from a file-sharing service?
– Should lurkers eventually become contributors?
– Hearing the “silent” users
IBM Research 12