Community Dynamics in Open Source Software Projects: Aging and Social Reshaping

TeLLNet

Lehrstuhl Informatik 5 (Information Systems)

Prof. Dr. M. Jarke I5-FL-MMYY-1 This work is licensed under a Creative Commons Attribution-ShareAlike 3.0 Unported License.

Community Dynamics in Open Source Software Projects: Aging and Social Reshaping

Anna Hannemann and Ralf Klamma RWTH Aachen University

Advanced Community Information Systems (ACIS) [email protected]

TeLLNet


Prof. Dr. M. Jarke I5-FL-MMYY-2

Motivation for Study Settings

  Address interdisciplinary projects (Bioinformatics) –  Biology meets Computer Science –  High disparities in level of development experience –  Better approximation for end-user integration in

community information systems (Lead User1, Open Innovation2, etc.)

  Analysis of long-tail: based on mailing lists  Dynamic analysis: community evolution

–  Demographic perspective –  Social structure perspective

1 von Hippel, E. “Lead users: a source of novel product concepts”, 1986

2 Chesbrough, H. “Open Innovation: The new imperative for creating and profiting from technology”, 2003

TeLLNet



Open Bio*

  BioJava, Biopython, BioPerl   Similar problems, infrastructure, organizational issues  Open Bioinformatics Foundation   Long-term: over 13 years Project* #Messages #User in ML #Commits #Developers LOC

BioJava 11951 2208 8267 94 290608

Biopython 16108

1138 16868

143 249566

BioPerl 31755

2824 12848

139 383351

* [Data on May 20, 2013]

TeLLNet



Newcomers vs. Survived Users BioJava

TeLLNet



Newcomers vs. Survived Users Biopython

TeLLNet



Newcomers vs. Survived Users BioPerl

TeLLNet



External Factors

 High attention to Bioinformatics due to sequencing of human genome

 Cross-project influence: rich get richer

  Personal aspects: –  doing PhD for 3 years –  being in a project with room for OSS

TeLLNet



Population Ecology

  Year of birth t0i: date of the first message from user i to project mailing list

  Age group (x; x+1): all currently active project participants participating in the project for min x and max x+1 years

 Currently active: at least one posting to the mailing lists in current year

  Survival rate (x; x+1)è(x+1; x+2): percentage of active users in age group (x; x+1) in the last year, who still active in the current year

TeLLNet



Population Ecology, Example 2010

  Age groups –  (0,1) people started in 2010 –  (1,2) people started in 2009, still active in 2010 –  (2,3) people started in 2008, still active in 2010 – …

  Survival rates –  |(1,2)|2010/|(0,1)|2009 –  |(2,3)|2010/|(1,2)|2009 –  |(3,4)|2010/|(2,3)|2009 –  ...

TeLLNet



Demographic Forecast

P0 = 0;1( )! (1;2)"# $% & 20%

P1 = 1;2( )! 2;3( )"# $% & 40%

Pm = x; x +1( )! (x +1; x + 2)"# $% & 90%,'x >1

  Power-law distribution of survival rates  Rebirth phenomena

TeLLNet



Conclusions and Discussion (1)

  Survival pattern: –  Prediction of minimal number of newcomers required to

support the same level of participation

–  Longer than three years survives only 7.2% – Who saves over three years, stays “forever”

 No maximal participation duration –  Number of “oldies” increases continuously –  Possible seclusion against newcomers

newcomert+1! 0;1( )t *0.2+ 1;2( )t *0.4+… x; x +1( )t *0.9,"x >1

TeLLNet



Social Network Analysis   Social Network (1 for each year)

–  Nodes: Email Participants –  Relations: Same thread

  Shortest path   Diameter   Node betweenness   Largest connected component   Density   Transitivity   Edge betweenness clustering

Biopython Network for 2012

TeLLNet



Social Measures

TeLLNet



Dynamic of Diameter

2001 2002 2003 2004 2005 2006 2007 2008 2009 20105

6

7

8

9

10

11

12

Years

Dia

mete

r

Dynamic of Diameter

BioJavaBiopythonBioPerl

TeLLNet



Dynamic of Max Betweenness

2001 2002 2003 2004 2005 2006 2007 2008 2009 20100

0.1

0.2

0.3

0.4

0.5

0.6

0.7

Years

Maxi

mal B

etw

eenness

Norm

aliz

ed

Dynamic of Maximal Betweenness

BioJava

Biopython

BioPerl

TeLLNet



BioJava, Social Network, 2006

TeLLNet



Biopython, Social Network, 2005

TeLLNet



BioPerl, Social Network, 2004

TeLLNet



BioJava, Core Evolution

TeLLNet



Biopython, Core Evolution

TeLLNet



BioPerl, Core Evolution

TeLLNet



Development Evolution, Biopython

TeLLNet



Conclusion and Discussion (2)   Core evolution

–  Evolves strongly –  Core generations (ca. 5 year periods) –  Dangerous for the whole project –  Defines organizational principles –  Can be predicted by combination of diameter and max

betweenness   Threats to Validity

–  Evolution step size (year to year, release to release, etc.) –  Scientist driven OSS –  Construct validity: quality of data; network construction –  Internal validity: observation – explanation

Community Dynamics in Open Source Software Projects: Aging and Social Reshaping

Business

jarke i5fl

tellnet lehrstuhl informatik

current year survival

years project

active project participants

percentage of active

survived users biojava

survived users biopython