But more interestingly, the above question begets another one. What would be the reasons for educational institutions to adopt open source?. Cost does not seem to be a clear one. Another reason to try open source could be better interoperability, robustness, features, etc. But ICT managers may favour familiarity with the system rather than migrating to a new system, even if it is potentially better from a technical point of view. And not all open source software is technically better than its closed alternatives.

So maybe we should look for the reasons to adopt open source somewhere else. The main characteristic of open source, if done properly, is that it promotes the formation of a community of people who can get involved with the project. This is arguably a good thing in education: engaging students and letting them play with the tools. Open source projects also function as a tutoring environment, where students can learn good practice from more experienced users and developers.

Thus, maybe the main reason for promoting open source in education is that it will help students learn more. But how? In fact, the ‘why’ leads to the ‘how’. If open source is to be successful in education, it cannot be just a matter of policy or, in general, a top-down approach. Its community and playful nature needs to be acknowledged and promoted amongst students, be it through local Linux User Groups (LUGs), creating fun projects (write a computer game in a collaborative way), or joining external projects.

Stanford rationale for offering courses free of charge is that research and teaching transfer is an important part of their mission, and that the SEE is an important step in making this possible. The materials include lecture videos, transcripts and course handouts.

MIT offers 1800 courses covering all subjects from Engineering, Sciences, Humanities and Arts. Susan Hockfield, President of MIT, explains the decision behind OpenCourseWare as ‘MIT faculty are passionate about their teaching, and they are keen to see their work benefit global society’. With an estimated cost of $10,000 to $15,000 per published course, this shows how seriously committed MIT are.

It shouldn’t be a surprise that top Universities have started offering their teaching materials for free on the Internet. In fact, it seems historically inexorable and necessary. In our Western societies with mandatory schooling until the age of 14 or 16, graduate students well in their 30s, and continuing education, it is easy to forget that ‘By 1424, Cambridge University library owned only 122 books—each of which had a value equal to a farm or vineyard’. It was largely thanks to Gutenberg’s printing press (ca. 1439) that knowledge became available to the wider society, not to talk about enabling the scientific revolution.

While cynics may argue that one goes to University for a degree, not an education (and a degree is precisely what open courses do not provide), the value of widely available high-quality teaching materials is hard to dispute. And not only for countries with worse access to education, or families who cannot afford college, but also for scientists struggling to keep up with new methods, techniques and results being produced at an ever faster pace.

Thus, open courses may be just another symptom of a social revolution that requires better sharing of knowledge, be it under the name of free or open source, open development, copyleft, open access or open standards.

Massiel was born as part of a short collaboration between OSS Watch (UK) and Libresoft (Spain), and informed by our Profiling Communities Expert Workshop. In short, Massiel is so far an attempt to extend  FLOSSMetrics, providing scripts for higher level analysis of databases created from projects’ svn logs and mailing lists (we have imported some code from CVSAnalY to do this). We are also exploring with Andrea Wiggins the use of workflows like those in myExperiment.

So why bother? Because even though OSS Watch is not a research group, data to back our advice to projects and institutions is very useful. This is important, for example, to avoid confirmation bias. That is, to avoid cherry picking information that confirms what we think about software projects and communities, while snubbing facts that contradict our beliefs.

The Profiling Communities workshop that we ran on July 11 addressed interesting questions. How can research be useful to evaluate open software communities? What is available in terms of data, tools, scientific methodology and algorithms? What would be interesting to measure?

The bit about open research just means that instead of running the study behind closed doors, and then one day publishing the results in a subscription journal, we make the data, methodology and intermediate results available right from the beginning.

This is more convenient for dissemination, but also a more effective collaboration framework for this project. The collaboration started when Israel Herraiz from Libresoft came to OSS Watch for a 1 month internship. In this case, starting a Google Code project was an easy way of setting up the infrastructure for the project (version control for the code, an issue tracker, mailing lists…), allowing it to survive past the 1 month internship, and at the same time enabling contributions from other people.

In addition to the purely utilitarian side of it, it could be argued that open research is better value for society than the more widespread subscription journal alternative. Scott Aaronson’s “Review of The Access Principle by John Willinsky” notes (with a good sense of humour) that researches and their institutions are using their time and resources to produce a wealth of invaluable scientific knowledge, that they give away to private publishing conglomerates. And this knowledge, produced by the scientific community and often funded with public money, remains closed and expensive, not only to the general public, but to the scientific community itself.

“Free software” and “open source software” are notoriously loose terms

The advocates of “open source software” tried to make it a trademark, saying this would enable them to prevent misuse. This initiative was later dropped, the term being too descriptive to qualify as a trademark; thus, the legal status of “open source” is the same as that of “free software”: there is no legal constraint on using it.

Free software aficionados are usually quick to point out that ‘free software is software that fulfills the 4 freedoms in the Free Software Definition of the FSF’.

Recently, somebody wrote to the aforementioned mailing list saying that software is free or not irrespectively of what the FSF says. It just needs to fulfill the 4 freedoms. But then again, who decides whether the 4 freedoms are fulfilled? The problem is that when you say ‘This program is free software’, as Bill Clinton famously put it (in a different context),

It depends on what the meaning of the word ‘is’ is.

That is, the 4 freedoms are not a mathematical expression that can be evaluated unequivocally. They are not even in legal language that can be argued in court (as my colleague Rowan noted). Something similar happens with the Open Source Definition of the OSI.

Somebody replied in the LUG’s mailing list saying that all you need is to ask a lawyer who knows about licences, and he or she will tell you whether the 4 freedoms are fulfilled. But this is not good enough, obviously, as different lawyers may have different opinions, and as I said before, the 4 freedoms are not in legal language.

What is more, not even the FSF thinks that the 4 freedoms are a perfect expression of the idea they have about “free software”

Finally, note that criteria such as those stated in this free software definition require careful thought for their interpretation. To decide whether a specific software license qualifies as a free software license, we judge it based on these criteria to determine whether it fits their spirit as well as the precise words. If a license includes unconscionable restrictions, we reject it, even if we did not anticipate the issue in these criteria. Sometimes a license requirement raises an issue that calls for extensive thought, including discussions with a lawyer, before we can decide if the requirement is acceptable. When we reach a conclusion about a new issue, we often update these criteria to make it easier to see why certain licenses do or don’t qualify.

So if we cannot rely on the Definition of Free Software, does this mean that we cannot define “free software” at all? In fact we can, if we accept that free software is software released under a free licence. In this case, free form language gets hammered down into the legal mold, and a lot of ambiguity is removed.

Of course the burden is now on deciding which licences are free and which ones aren’t. Accepting that free licences are licences that the FSF say are free seems to cause a lot of discomfort to some people.

This is not so much of a problem with the open source community, who seems more willing to accept that open source licences are not those that are believed to fulfill the Open Source Definition, but those that the OSI certifies are open source (and that’s the approach we follow in OSS Watch too). The OSI has even registered the “Open Source Initiative Approved” trademark for specific software products.

I guess that the reason why giving the last word in terms of “free” to the FSF causes discomfort with some people is that the free software community is built on the idea of freedom as a paramount value, not only for software but for society as a whole, and that subordinating “free” to the FSF is giving away part of that freedom. In the end, “free” would be a badge awarded by an opaque team of lawyers.

At the same time, I don’t think that anybody would seriously consider as a better option to have any number of licences, each of which needs to be evaluated by each individual in order to decide whether they are free or not.

Both the OSI and the FSF have approval processes in place to decide not only whether a licence fulfills certain requirements, but also to make sure that it is not similar to an existing one and thus cutting down licence proliferation.

Licence proliferation is the enemy from within for open and free software, because it is possible to have licences that fulfill the open and free definition, but at the same time prevent different projects to combine their outputs and collaborate. The most significant example being the incompatibility between the GPL v2 and Apache License v2 (the GPL v3 is compatible with Apache Licence v2, though).

I believe that at this point the need to standarize licences, make them compatible and reduce their number outweights the risk of the FSF going awry, and hence I’m happy with the statement “Free software is software released under a licence approved by the FSF”.

Although this does not look very ethical, in fact what the seller did is (quite likely) perfectly legal. The Gimp is distributed under the GPL. This means that the software is free in a “freedom” sense, but not that it has to be provided for free. In fact, the GNU project’s position about selling free software is clear:

Many people believe that the spirit of the GNU project is that you should not charge money for distributing copies of software, or that you should charge as little as possible — just enough to cover the cost.

Actually we encourage people who redistribute free software to charge as much as they wish or can. If this seems surprising to you, please read on.

Free software is quite unusual in that it is written by somebody who uses his copyright ownership to grant others the right to do with it as they please, as long as it remains being free software. And this includes selling it for any amount of money.

In most cases free software projects release their programs for free. If they did not, somebody could buy a copy and then redistribute it at no cost. Free software business models focus instead on other revenue streams, e.g. selling support to customers. Unfortunately, the seller was not trading in this case on a product or service but on the lack of awareness of the buyer.