| Analysing the Economics of Open Source |
| Written by Mal Minhas, CTO, LiMo Foundation |
| Friday, 18 September 2009 16:50 |
|
An important underlying dynamic in the mobile industry to date is that investment costs have been rising faster than the corresponding return on that investment. As user expectations have increased, so has the software burden of developing ever-more sophisticated handsets and the corresponding backend infrastructure. This investment return dynamic has been a fundamental driver for the adoption of open source software in mobile devices though there are, of course, many other important factors such as the open scrutiny of code and architecture which reduces security risks and wider access to engineering skills that improves quality of software and provides access to innovation.
LiMo Foundation, as many others in the industry, is a firm believer of the benefits of using Linux and open source software. The recently announced LiMo R2 platform is built around key community open source software components such as GTK, D-Bus, GStreamer and BlueZ. Since the source code for these components is publicly available, it is possible to download, analyse, produce and capture metadata from that code. This is what the ohloh web service does. Using ohloh, it is possible to examine the history and evolution of an open source project in considerable detail and determine amongst other things, who contributed to the code and when. This ability to create data about source code is another significant benefit of open source and it opens the door for quantitative analysis based around measuring source lines of code (SLOC).
To date, the economic benefits of open source in a mobile context have not been subjected to detailed quantitative analysis. At LiMo, we wanted to make the basic economic case for mobile open source software. We recently conducted an examination centred around the three key motivating drivers for mobile platform providers to use open source code within their platform:
The core of the approach was to use Barry Boehm's CoCoMo II model to determine the raw cost of software in each case. The RESTful ohloh web service APIs were used to generate accurate figures on the evolution of SLOC in major open source projects of importance within a mobile context. The resulting white paper was recently published by the LiMo Foundation and presented at OSiM 2009 in Amsterdam a couple of days ago. In addition, the Python-based tool we created to assist our analysis has been open sourced. Looking at the indicative figures for raw costs alone, they run into $millions in many cases. A key observation from looking at the data is that community open source projects continue to evolve and move on and in a sense, are never really finished. This means it is vital for a platform provider to have an effective strategy in relation to maintenance, particularly of community open source code. Acquiring an OSS project is like acquiring a garden - one needs to think about how to maintain it. On the other hand, OEMs are driven by commercial imperatives, particularly around shipment dates. Moreover, there are practical limits with regards to staying in sync with upstream given that the OEMs need to undertake a software freeze process once they are close to shipment. The study is best viewed as an initial positioning paper which establishes the scale of the ceiling costs involved in using open source in mobile devices. It is not exhaustive by any means. We have yet to model the involvement of the community in aiding maintenance. Our hope is that it stimulates a debate about practical ways in which commercially driven mobile platform providers can work with the wider open source community. The operational detail of that engagement is what we're going to focus on next. We also hope that the tool proves to be useful and we welcome suggestions for evolving and enhancing it. We believe that the mining of open source for open metadata is an aspect of openness that should definitely be elevated.
|
