The Benefits and Challenges of Open Source Software in Research

An important part of ensuring analyses are reproducible for other researchers is ensuring that the software used is accessible. Analysis softwares can be categorised into proprietary, freeware, or open source. The most open of these is open source software, in which the source code is open for any developer to work on and improve upon. Then is freeware, where the source code is closed and cannot be modified publicly, but the finished product can be downloaded and used by anybody. Finally, the most closed is proprietary software, which can only be used by users who have purchased licences or subscriptions to the software.  

FAIR data principles state that open data should be interoperable. This means they should run on any machine. Proprietary file formats produced by proprietary software are therefore not interoperable, and cannot be reproduced by researchers without access to those softwares. Freeware can be downloaded and used to reproduce an analysis given that the operations have been clearly laid out, but the details may be hidden under a GUI, and more complex or bespoke analyses may not be possible as the source code cannot be altered, and file formats may or may not be interoperable. Open source software always produce interoperable outcomes, because the source code itself is interoperable by anyone on the internet. However, more open software options often come with the difficulty that they are less user-friendly and have a steeper learning curve than proprietary software. For example, SPSS is a “point and click” style software, whereas R requires users to write their own analysis script. Some proprietary softwares, such as Stata, are industry standard, and therefore learning to use new software would require an extra effort from users. What this means is that although the analysis techniques themselves become more open and accessible with open source software, and the software itself is available to anyone with an internet connection, the number of people with the skills to use and interpret that code is smaller.  Furthermore, as documentation for open source software is produced on a voluntary basis, it can have inconsistencies, making troubleshooting difficult for new users. 

This event examines the conflict between these ways of doing analysis, and what that means for open research. It asks whether the putative openness of open source software and reproducible code are actually a barrier to broader openness to the people who do not have the coding skills to interpret it. 

This event will be hybrid.

A light lunch will be provided.

Panel:

Chair: Dr Jonathan Cardoso-Silva, Assistant Professor (Education), LSE Data Science Institute.

Dr Barbara McGillivray, Senior Lecturer in Digital and Computational Humanities, King's College London.

Dr Steven Sam, Lecturer in Computer Science, Brunel University London.

Dr Adam Crymble, Associate Professor in Digital Humanities, University College London.