LWN.net Weekly Edition for November 3, 2005

A study on free software in British schools

There have been a number of stories recently about the adoption of free software in public school systems around the world. Certainly free software has a lot of attributes which make it well suited for that role: it is relatively secure, open to curious minds which wish to look inside it, freely available for students to copy and use at home, easily adapted to local languages, and easier on a school's (typically stretched) budget. Of course, not everybody agrees that the use of free software is cheaper; certain proprietary software companies, in particular, are trying to cast doubt on that assertion. So the administration of a school contemplating a switch to free software might well wonder: will it truly save money?

The British Educational Communications and Technology Agency decided that it needed an answer to that question. So it took a detailed look at 48 British schools - 33 which were not using free software, and 15 which were - to get a sense for their relative costs. The result of this work is now available as a glossy report [PDF], suitable for printing on heavy paper and handing to a school administrator near you.

The study divided software usage into three broad categories: (1) servers, (2) class and administrative computer operating systems, and (3) classroom and administrative applications. The total costs were summarized in each category, taking a broad view. Costs include hardware and software, but also support - both purchased from outside and provided by internal staff. Training was also included. In other words, the study took into account all of those factors which, according to the critics, make free software more expensive than the proprietary alternatives.

The bottom-line result is quite clear:

Unsurprisingly, the study found that the best immediate results came from the use of free software on server systems. There are more obstacles to deployments on administrative and classroom systems. In some cases - especially for school administrative functions - the necessary applications are not yet available (the study notes that projects like SchoolTool are working to provide those applications). There is also some opposition to free applications from people who are trained in other packages. Tellingly, most of this opposition seems to come from the teachers, not the students:

Teachers and administrators, like most adults, have a certain tendency to get set in their ways and stick with what they already know. Children can be more flexible. What these schools are seeing corresponds with your editor's own experience: children have no problem working with free software, and, if exposed to it, will take to it readily. Just don't (speaking from experience here, again) expose your children to Battle For Wesnoth, or their homework will suffer.

In summary: this report is a good thing, as far as it goes. The flood of hostile "total cost of ownership" studies is unlikely to slow in the near future, so it is good to have contrary evidence from relatively unbiased sources. There are, however, no end of reasons, beyond the financial ones, for using free software in public schools, but this report ignores them almost completely. At the lower school levels, free software can be made available to students without licensing hassles or sanctimonious lectures about not making copies. At higher levels it can teach the students much about software itself, encourage them to experiment, and demonstrate how cooperative work can yield benefits for everybody involved. A strict focus on costs may provide a favorable picture, but risks creating the impression that cost is the only reason for using free software. In the context of the public schools, more than in many other situations, it is important that people understand that there is far more to free software than "free of cost."

Comments (14 posted)

Open source compliance insurance

The folks at Open Source Risk Management have, for some time now, been working on indemnification insurance for free software. The idea behind this offering is that businesses which are worried about an SCO-style lawsuit can purchase insurance turning that risk into a regular, predictable business expense. This sort of service may well turn out to be a hard sell, however; SCO's experience seems unlikely to inspire many copycat acts. The risk of successful, copyright-based legal attacks against free software currently seems to be quite low.

Patents may yet prove to be a different story, though.

Meanwhile, OSRM, in conjunction with Kiln plc and Miller Insurance Services, has come up with a new idea: sell insurance to companies which fear GPL compliance problems:

This is, in other words, a 180-degree change from the previous OSRM offering. The previously-offered indemnification policies addressed concerns that free software could be infringing on copyrights through the inclusion of proprietary code. Now, instead, we have insurance to benefit those who might just infringe the copyrights associated with free software. This situation is, certainly, more likely to come about; here's another quote from the press release:

This statement, certainly, is a strong statement in favor of the enforceability of free software licenses.

One might argue that this sort of insurance policy presents a moral hazard. A company which hopes to ignore the requirements of the GPL could ship a product without source, secure in the knowledge that, if somebody calls them on it, they can fall back on the insurance policy to mitigate their losses. The plan's fine print must certainly have language excluding deliberate acts of noncompliance, but proving that a specific infringement was willful could be a challenge.

Others might say that there should be no inadvertent infringement of the GPL; any such infringement constitutes, at best, an extreme lack of due diligence. Consider, however, the much-publicized cases where various wireless routers have used GPL code in noncompliant ways. In these cases, the final vendor - the one whose name is on the product - often has little knowledge of what software was used by the obscure, far-eastern supplier who actually made the product. As more GPL rulings are handed down, it seems likely that resellers will start asking more questions of their suppliers, but surprises still seem possible. This particular risk - being betrayed by a supplier - seems like a legitimate thing to insure against.

So OSRM and its partners might just find a market for this particular offering. If, in the process, they make businesses feel more comfortable about using free software in their products - and, perhaps, even helping with the further development of that software - it should be a good thing for the free software community as a whole.

Comments (3 posted)

EFF: Halloween on the hill

As reported on the EFF site: the broadcast flag is back, bigger and badder than ever. The new halloween document [PDF], otherwise known as the "Analog Content Security Preservation Act of 2005," would impose no end of restrictions. "The unprotected analog outputs of computers will be, in perpetuity, restricted to either DRM-laden standards, or to a 'constrained image', 'no more than 350,000 pixels'. Analog video which has been branded as 'do not copy', will last for only ninety minutes only in the digital world - and will be erased, literally frame by frame, megabyte by megabyte, from your PC, without your control. You'll watch a two hour film, and as you watch the final half hour, the first few scenes will be being dissolved away by statute."

Comments (16 posted)

The send-a-link feature

The recent discussion on improving LWN's readership led to one clear action item: the addition of a feature which would allow subscribers to create special links which they could use to point out interesting articles to non-subscribers. These links would bypass the normal subscription gate, allowing articles to be read while they are still current.

That feature has been implemented, and is now active. There is no limit on the number of links a subscriber may create, and no limit on how many people may read an article via a given link. A few caveats do apply, however:

• For the time being, only "project leader" subscribers have the ability to create subscriber links. This restriction is meant to be temporary; its main purpose is to slow the initial use of the new feature while any remaining bugs are shaken out. It would, however, be interesting to hear what people think of keeping subscriber links as a differentiating feature for the high-level subscriptions.

• Subscriber links can be made for individual articles; just look for the "send a link" line in the left column. These links cannot be made for entire Weekly Edition pages, however.

• We reserve the right to turn off the subscriber link capability for specific articles; the annual timeline is a case where we might do that. No decisions have been made on that point, however, and the mechanism to implement an exclusion policy has not yet been implemented.

• We reserve the right to turn off the whole thing if it looks like the feature is being abused and hurting subscription sales. We do not expect things to go that way, however.

Privacy stuff and details: for each link, we track who created it and the number of hits it receives. That information will go away some time after the link expires - which happens when the relevant article becomes freely available. The links are constructed in such a way that they will continue to work forever. Currently, following a subscriber link leads directly to the article in question; in the future, we might throw in some sort of encouragement to subscribe.

We are most interested to see how this new feature - which was driven by requests from our subscribers - works out.

Comments (18 posted)