Date of Award

12-2009

Document Type

Dissertation

Degree Name

Doctor of Philosophy (PhD)

Legacy Department

Management

Advisor

Grover, Varun

Committee Member

Moore , DeWayne

Committee Member

Klein , Richard

Committee Member

Wasko , Molly M

Abstract

With the emergence of Free/Libre and Open Source Software as a significant force that is reshaping the software industry, it becomes more important to reassess conventionally held wisdom about software development. Recent literature on the FLOSS development process suggests that our previously held knowledge about software development might be obsolete. We specifically highlight the tension between the views embodied by the Linus' Law and Brooks' Law.
Linus' Law was forwarded by Eric Raymond and suggests that the FLOSS development process benefits greatly from large numbers of developers. Brooks' Law, which is part of currently held wisdom on software development, suggests that adding developers is detrimental to the progress of software projects. Raymond explains that the distributed nature of the FLOSS development process and the capacity of source code to convey rich information between developers are the main causes of the obsolescence Brooks' Law in the FLOSS development context.
By performing two separate studies, we show how both views of software development can be complementary. Using the lens of Transaction Cost Theory (TCT) in the first study, we identify the characteristics of the development knowledge as being the main factors constraining new members from contributing source code to FLOSS development projects. We also conceptualize of these knowledge characteristics as being analogous to what Brooks' described as the ramp-up effect. We forward the argument, and offer empirical validation, that managing these characteristics of knowledge would result in an increase the number of contributors to a FLOSS projects.
The second study is concerned with the impact if having these new members added to the development team in a FLOSS project. Using the lens of Organizational Information Processing Theory (OIPT), we forward the argument, and offer empirical validation, that more contributors can be detrimental to progress if the committers of a FLOSS project are overwhelmed. Our findings also suggest that large development teams are indeed possible in FLOSS, however, they must be supported by proper source code design and community structures.

Share

COinS