“Institutional Repositories Should Be Built on Open Source Software” is one topic in “Institutional Repositories: The Great Debate” which is being held in the current issue (April/May 2009 –PDF format) of “The Bulletin of the American Society for Information Science and Technology”.
Meanwhile over on Glyn Moody’s Open … blog an argument is being made that “Open Science Requires Open Source“. Here we read that:
“The central argument is important: that you can’t do science with closed source software, because you can’t examine its assumptions or logic (that “incomplete scientific record”). Open science demands open source.“
And who could possibly disagree?
Well I’d challenge such conclusions. I feel that we need to reflect on the over-hyping of open source software over the past decade: we should now, if you believed the hype, all be using open source office software on our desktop PCs, and those desktop machines would all be running Linux. But this doesn’t reflect the working environment for most people, with open source email clients now seemingly in need of saving.
Despite its failure to live up to the expectations of the evangelists, we are now seeing more effective use of open source software. But for me this is because open source software is now being evaluated on par with licenced software, and not because open source software is felt to have any natural advantages. I would argue, in fact, that uncritical acceptance of open source software in the past led to disillusioned end users and the ‘counter-culture’ approach adopted by some open source developers led to the software development which failed to have a community to ensure that the software was sustainable in the long-term future.
Despite the frequently cited examples of Apache, email server software and the like, is there evidence that open source software has a significant role to play beyond the server environment?
And in cases in which open source software is growing in use, such as Open Office on cheap Netbook computers such as the Asus EEE PC, isn’t it the case that the advantages provided by such software are in avoiding licence costs rather than in the other benefits which open source evangelists promote? Aren’t the benefits for most users to be found inthe amntra that the software is “free as in beer” rather than “free as in speech”?
At the JISC OSS Watch’s inaugural conference Jeremy Wray, Business Development Executive for Public Sector, IBM argued that it would be a mistake to compete in well-established markets such as office software, citing IBM’s failures in competing with Microsoft. Perhaps open source software should be positioned in more niche sectors such as institutional repositories and open science? And yet even here I have my doubts. If we are passionate about open access to research publications and open access to scientific data, then shouldn’t we be focussed on such issues and be neutral on the production mechanisms used to develop the associated software? And the argument that you need open source software to examine the assumptions and logic is flawed – source code can be made available for inspection without it being licensed under an OSI-conformant open source licence.
Yes, use open source institutional repository software and open source open science software. But do so because the software satisfies its intended purpose and is better than proprietary alternatives and not just because it is open source. And let’s not forget the associated risks of using open source solutions: many of the more widely used open source applications are bankrolled by large IT companies which are suffering from the economic downturn. And if widely used open source solutions start to suffer from a lack of ongoing inverstment, where will that leave the more niche solutions?
OK, I’ll rise to that bait.
I don’t think anyone – at least not these days – would claim that open source is going to take over the desktop soon. There are good historical reasons for this – free software originated as a hacker tool, and user interface design was not a priority.
But I think this focus on the desktop misses the point – as does the comment “is there evidence that open source software has a significant role to play beyond the server environment?”, which seems to imply that the lack of success on the desktop means that open source is doomed to spend its life hidden away in the back office.
The interesting thing about open source is that it is all around us, but invisible. Google is one example, but just about every Web 2.0 company (Facebook, Twitter, Delicious etc.) is based on free software. Aside from Microsoft’s Azure, practically every cloud computing solution is built on it (necessarily – who could afford the licensing fees otherwise?) But even these pale beside the use of Linux in particular in embedded solutions. Most of the intelligent devices that are appearing – from GPS devices to ADSL routers – have open source at their heart.
This use of free software to power new kinds of devices derives from three things: its low cost, its stability and its customisability. These mean that it’s incredibly easy to produce new system quickly and for little money.
That, I would suggest, is why open source is the best solution for scientific software. Not for ideological reasons, but simply because it gives you the power to do what you want to do more easily, more quickly and for less money. The same applies to using pre-existent open source solutions: they give you a flexibility that closed ones do not, not to mention a freedom from lock-in.
I don’t think any lack of investment in open source will be a problem, for the simple reason that free software is ultimately based on people hacking for the pleasure of it. The investment by companies is welcome, but not indispensable – as the early history of free software shows. If there’s an itch, someone will scratch it.
Hi Glyn, Thanks for the response.
When you say “just about every Web 2.0 company (Facebook, Twitter, Delicious etc.) is based on free software” the naive reader might think that their core services is based on open source software which others can download, tweak and install. Clearly this isn’t the case.
We are in agreement, I think, that open source has particular strengths in the Web and Internet foundations. The question for me is the role open source may have at the application layer. And it seems that many of the cloud services are owned by large organisations such as Google, Yahoo and Microsoft. Indeed Peter Murray-Rust of the Unilever Cambridge Centre for Molecular Informatics, in his blog post about his talk on Libraries of The Future did include “Rapid entry of major players – (GYM) Google, Yahoo, Microsoft” in his summary of where the world is going anyway.
There are clear dangers with such popularity, but the solution need not be a requirement for open source solutions, IMHO.
Hi Brian
Picking the bones out of all the Open, Free and Libre rhetoric (software, source, licences, systems, access, archives, universities…) is a forensic task, and the dangers of over-conflating the many themes at work are ever-present.
It seems reasonable to me that high-stakes scientific research has an extra-pressing need for assurance its software is reliable and consistent, free from accidental or malicious surprises: access to the source code lays that bare. If I ship even a simple binary that adds and subtracts, it might yet contain any kind of error, prank or virus; if I ship the source code, it hardly can. As you point out, that openness can be achieved without giving up any rights in that code.
But I agree that Open Source software is not the be-all of many types of application, including IRs: costs and benefits need realistic assessment as ever.
What we can and must continue to promote, however, are Open Standards and Open Formats (and probably Open APIs too): the cost and risk of letting software manufacturers lock our data in private, undocumented, proprietary formats – as they have done in the past – is not acceptable. What happens inside the black box of many an app need not generally concern most of us, as long as it does what we expect: what is essential is that we are free to export, access and use our data in another application, as and when we (not the software mfr) decide.
It’s certainly important to cast a careful eye over the hype of Open Source and Free Software, but I’d agree with Richard M. Davis that Standards and Formats are the most important, as these are what prevent vendor lock-in, and thus preserve an important freedom.
With the issue of ‘free as in beer’, this in a sense follows freedom from vendor lock-in, and as an amateur web developer I can’t emphasize enough how important it is that I don’t need to shell out on ArcGIS licenses and instead can go for QGIS (to take one example from my own experience). For this reason Open Source software will always have an important role to play on the desktop (whether running on Windows, Linux or something else).