who's there? a stakeholder analysis approach for hybrid oss
TRANSCRIPT
Who's there?
A stakeholder analysis approach
...for a Hybrid OSS developer ecosystem
Hanna Mäenpää, Terhi Kilamo Mikko Nurminen
Myriam MunezeroFabian Fagerholm
Tomi Männistö
Open Source Software
• Community driven development
– Freely accessible source code
– Self-organized developers
– Meritocratic
– Consensus
– Open communication
A hybrid OSS project
Commercialstakeholders
business goals,
needs of customers
Independentdevelopers
own problems, intrinsic reasons,
ideology
Software product
Hybrid OSS projects
Commercialstakeholders
business goals,
needs of customers
Independentdevelopers
own problems, intrinsic reasons,
ideology
Software product
Host
The research gap
RQ a1: How to identify new and stay aware of present stakeholders in an open environment?
Linåker et al. (2015):
Multi-platform applicationdevelopment framework
– 20 years of development
– 1.1 million downloads (2014-)
– 5 000 customers
– Active developer community
– Stable, business driven stakeholder ecosystem
– Influential independentdevelopers
Open community
Workflow coordination
Software development
Quality assurance
Qt Company's teams
Release engineering
Software development
Quality assurance
Customer service
The hybrid organization
Open community
Workflow coordination
Software development
Quality assurance
The hybrid organization
Qt Company's teams
Release engineering
Software development
Quality assurance
Customer service
Management challenges
Sustainability
● Acquiring new contributors from outside the company● Increasing involvement of existing
Conflict of motivations
● Release authority● Development priorities – whose say gets through?
Reserach Questions
RQ1: How to provide a visual overview of the active developers of the Qt software?
RQ2: What value does this visualization provide for community management?
Single case study, empirical and descriptive
1 Interviews with the community manager.
2 Logs from openly available development tools: Qt 5.0 development during Jan-March 2015
3 ..and more interviews with the community manager.
Methods and data
284 active developers Four tasks:
(A) Submitting work issues
(B) Being assigned for work
(C) Performing code reviews
(D) Delivering software source code
Qt 5.0 Jan-March 2015
Developers of the Qt software (1-2 roles)
(A) created a work task(B) was assigned to one(C) performed a code review(D) authored a code increment
The core developers (3-4 roles)
(A) created a work task(B) was assigned to one(C) performed a code review(D) authored a code increment
139 affiliations
- Employees of The Qt Company
- Members of partner organizations
- Independent individuals
- Commercial organizations
Qt 5.0 Jan-March 2015
Requirements engineering and code authorship
A = Created a work taskD = Committed code
Large groups that provideinput for productdevelopment.
A = Created a work taskD = Committed code
Large groups that provideinput for productdevelopment.
Stakeholders from versatileorganizations is desired.
Requirements engineering and code authorship
Core developers: External influence
ABCD, ABD, ACD, BCD
Developers who actively run theProject.
Mostly (but not solely!) employeesof the host company.
Sizes of certain groups: Following the developer's entry paths
Size of group can reflecteffectiveness of communitymanagement activities:
Acquiring new members (A,B,D)
Sizes of certain groups: Following the developer's entry paths
Size of group can reflecteffectiveness of communitymanagement activities:
Acquiring new members (A,B,D)
Increasing involvement of existingmembers (AB, AD, BD)
Sizes of certain groups: Following the developer's entry paths
Size of group can reflecteffectiveness of communitymanagement activities:
Acquiring new members (A,B,D)
Increasing involvement of existingmembers (AB, AD, BD)
Names of individuals:
John Doe, Intel
Code reviewers
Deeply knowledgable.
Long path of personal learning.
Work independently.
55% The Qt Company's employees.
→ High level of external involvementis desired.
Warning sign: Losing the support of external stakeholders
ABCD, ABD, ACD, BCD
Loss of external contributors in any group signals a threath forsustainability of the community.
Is the company influencingdevelopment too much?
Summary of findings:
Measuring external influence:
● Versatility is foundational for hybrids
● Community drivenness can be measured
Balancing acts:
● Quantity and quality of involvement
● Which tasks require more work power?
● Are we attracting new developers?
● Are the community building activitiessuccessful?
Warning signals:
Conflict of motivations?
Are workflowcoordination tools fit forpurpose?
Promising, yet limited approach
● Single case study, no generalizability
● Limited set of data, representative of only a short moment in time
● No automation, temporal analysis impossible
● However: case is representative, visualization method simple
● FUTURE: Governance is configurational - what is relevant and for whom?