Open source describes a software where the code is placed in the public domain. Hence anyone can download it and use it or alter it. When I first heard about it I thought it sounded like a great idea. But on reflection I have serious doubts.
Peter Gill informs me that “official” code can only be changed by the developers. However our copy of LRmix has been altered by us and there is no visible evidence for this in the output.
The primary argument, that initially attracted me, was openness with respect to the defence in court cases. However there is no reality to this. It is almost impossible to detect a bug from the code. In our own case we track down any faults during development by noticing an unusual result and slowly narrowing down the part of the code suspected. Eventually we do go to the code, but look at a small part that we have identified as the potential source of the unusual result.
Peter Gill agrees: see https://drive.google.com/file/d/0B18mPWHZjsgYRkhMOTRGU2JuMTA/view
For STRmix we do disclose the code to the defence – hence this openness can be achieved without open sourcing. This has been requested twice. Once in Regina v Tuite which is a case in Melbourne, Australia. In that case Nathanial Adams, a student from the US, was sent to inspect the code. His conclusion was that he really needed to run samples and compare the expected result with what was obtained. We agree.
The second case was in Scotland. Allan Jamieson demanded the code. When we agreed a strange series of events followed that included very large delays by Jamieson, and increasing demands first to share with colleagues and then to use the code in any case. We agreed to this. Jamieson never took up the offer. Nor did he ever answer the question: “Can you read Java?”
We did detect two errors in Lab Retreiver but not from the code, but by running samples.
The other argument we have had put to us in collaborative development. I think that this is unrealistic. The only instance we have is Lab Retreiver which grew out of LikeLtd. This was more a simplified partial clone than a development.
Nothing is free. Open Source software is usually developed by academics funded by some research grant. In some cases the developers charge for the training. This sets up a situation where a potential user can obtain the software and use it without training or skimp on the training to save money. In most cases the software is simply an interesting sideline for these academics and they have no real investment in training and supporting labs during validation and implementation. There is no commitment to continuity of development, training and support. In all cases the supporting documentation is very light. Most come with a disclaimer that the software is made available but training and validation is the responsibility of the user.
In forensic science this is the absolute opposite of what we want. We want well trained users who understand the principles of the software and both validate and use it intelligently and professionally.