Driving Behavioral Change for Information Management through Data-Driven Gree...
Alone Together: Patterns of collaboration in free and open source software development
1. James Howison
PhD Dissertation Defense
iSchool @ Syracuse University
4 December 2008
Alone Together:
Patterns of Collaboration in
Free and Open Source
Software Development
Advisor: Kevin Crowston
Committee: Bob Heckman,
Carsten Østerlund, Don Harter (B-school)
Inside Reader: Steve Sawyer
Outside Reader: Francesco Bolici (Remote)
Chair: Sumitro Bannerjee (B-school) CC Credit: http://www.flickr.com/photos/baggis/
2. Free (Libre) and Open Source
Software Development
• FLOSS development is a canonical
model of distributed work
– Interesting for itself, but also for adaptation
• Research to date on Motivations OR
Production but not both together
3. Scope:
Community-based FLOSS
• FLOSS is not one phenomenon
– A license is not enough
– Many hybrids with existing organizational forms
• Interested in the Something Else, therefore
studying the “pure” form:
– Distributed, no center or face to face
– Unpaid and non-commercial
– Sourceforge hosted, set of collaboration
technologies: Email, source code repository, issue
trackers …
4. Overall Research Questions
1. How is successful FLOSS production organized?
2. How does this organization interact with
motivation, and thereby recruitment and retention
of developers?
3. What are the implications for the adaptation of
the FLOSS model of organization in other
environments?
5. Structure
• Unfolding arc:
– Discovery: Participant Observation
• BibDesk (RQ1 & RQ2)
– Replication: Archival Study
• Fire & Gaim (RQ1)
– Explanation: Rational choice model
• RQ2 and RQ3, via conditions for model
11. Discovery Findings
1. Individual work with personal
motivations
2. Layering
3. Deferral
CC Credit: http://flickr.com/photos/jvk/
12. Theory: Interdependency
• Fundamental
• Fixed
• aprioi
• Emergent
• “emergent property
in social systems”
• (Actually socio-
technical systems)
March and Simon (1958),
Mintzberg (1979),
Thompson (1967), Van de
Ven et al (1976), Malone
and Crowston (1994)
Shea and Guzzo (1989),
Wageman (1995), Wageman
and Gordon (2005), Rico and
Cohen (2006)
17. Replication: Fire and Gaim
• Specific RQs:
– What proportion of work was individual?
– Any evidence of productive deferral?
• Fire and Gaim
– Multi-protocol instant messaging clients
– Community-based open source
– Similar task and collaboration infrastructure to
BibDesk
18. To the Archives!
The evidence is here, somewhere.
http://www.flickr.com/photos/hamadryades/
19. Method
1. Identify Task Outcomes
2. Search for Relevant Documents
3. Recognize Actions and Participants
4. Code Action by Contribution
5. Classify Task
20.
21. Types of Contribution
• Management Work
• Review Work
• Production Work
– Core Production Work
– Polishing Production Work
• Documentation work
• Supporting work
22. Classification
1. Find Production Type
– Solo, or
– Polished Solo, or
– Co-work.
2. Add other tags
+ Reviewed, and
+ Supported, and
+ Managed.
27. An image of FLOSS production
• Work is done in Tasks that are
– Individual
– Short
– Layered
– Spontaneously supported
• Complex work is often deferred
– Until it is easier
Other types of work build on this base
28. Answering overall
Dissertation RQs
• RQ1: organization of FLOSS
– The image presented today
• RQ2: interaction with motivation
– Co-evolved socio-technical fit between motivation,
organization and collaboration technologies
– Rational Choice Model explores expectancy-
valance motivational model
– Co-work multiplies risks
29. Adaptation (RQ3)
• Empirical studies and Model help build a
framework used to assess difficulties of
adaptation. eg:
– Layerable, with ‘stackable’ incentives
– Ultra-Low Instantiation and Distribution Costs
• Using the same technologies and licenses not
sufficient!
30. Contributions I
• Information Systems
– Empirically grounded socio-technical theory of IS
development and distributed teams, IT artifacts
play important roles.
• Organizational Science
– Identification and description of a novel
organizational form
– Empirical evidence of low-interdependency
complex work
– Deferral may have wider applicability
31. Contributions II
• Methodology
– Concepts and Method for reconstructing work from archives
• Teaching
– Set of annotated narrative cases of FLOSS work, useful for
teaching those working with communities
• Open Source practicioners
– Give “Community managers” a lens to understand their own
communities
– Those starting projects: a (partial) guide to existing
experience
32. Thanks especially to:
• The BibDesk developers and users and the rest of the Free
and Open Source communities
• Kevin, my advisor
• My committee (Bob, Carsten and Don)
• My 2 writing groups (Anu and Indira & Saira and Isabelle)
• The FLOSS research team (Kevin, Bob, Hala, Chengetai,
Andrea, Yeliz, Kangning, Qing, Mike, Lina, Steve and Eileen)
• All the iSchool colleagues, junior and senior, who’ve
challenged and inspired me.
• Jennifer, Maureen, Ellen and Bridget!
• Becks and Joe
• Becky (so, so cool), Kal and Drogo
• Judes & Bobs, Kate & Andy (never far away, really :)
And, of course, the NSF (03-41475, 04–14468, 05-27457 and 07–08437)
fin