Where and when can open source thrive? Towards a theory of robust performance

From AcaWiki
Jump to: navigation, search

Citation: S Levine, M Prietula (2010) Where and when can open source thrive? Towards a theory of robust performance. Proceedings of OSS 2010. Open Source Software: New Horizons (RSS)
DOI (original publisher): 10.1007/978-3-642-13244-5_13
Semantic Scholar (metadata): 10.1007/978-3-642-13244-5_13
Sci-Hub (fulltext): 10.1007/978-3-642-13244-5_13
Internet Archive Scholar (search for fulltext): Where and when can open source thrive? Towards a theory of robust performance
Download: http://dx.doi.org/10.1007/978-3-642-13244-5 13
Tagged: open source (RSS)

Summary

The paper, presented at OSS 2010, is framed as helping address an unaddressed aspect of the literature on open source on the performance of open source models. Performance, in the authors conceptualization, seems to translate roughly into the percentage of goals addressed by an open model so their question can roughly be seen as analogous to asking what portion of problems can be addressed by open source.

The authors use a very broad definition of open source that goes far beyond just software and seems to encapsulate almost any model that is characterized by (a) open access to contribute and consume, (b) involved in the creation of products of economic value (c) involves interaction and exchange activities as central and (d) involves purposeful but loosely coordinated work.

The authors claim to bring data that is qualitative in nature and to use that to construct a agent-based simulation. Details of the qualitative data are extremely thin. The model includes variable associated with agent and population based characteristics. These are deterred in terms of likelihood to contribute and are based on three basic types (cooperators, reciprocators, and free riders). The model is essentially a large number of social dilemma type games. The models also include variables measuring the demand for the goods being produced and a spectrum of rivalry from completely rival goods to non-rival goods.

The basic-takeaway from the paper seems to be that in situations of low (or near-zero) marginal cost (i.e., low rivalry), free-riding has a very minimal effect because it doesn't cost much. The free-riders, in this sense, do not directly affect the availability of the resource. As a result, even with only a small number of cooperators in the population (e.g., 13% which is what the authors cite as empirical observable in human populations), performance can be quite high (40% of goals being served).

The authors' simulation finds decreasing marginal returns to cooperation (Prop. 1b) and a trade-off effect between cooperators and reciprocators in the population (Prop. 1d).