Second International Conference on Open Source Systems 2006

by James A J Wilson on 14 June 2006 , last updated

Archived This page has been archived. Its content will not be updated. Further details of our archive policy.

Introduction

Introduction

Doubtless the Italians regard Como in much the same way as the English regard Macclesfield – a drab Northern town associated with silk manufacturing and not much else. They probably assumed that hosting a conference there was a good idea because there would be few distractions. Certainly, judging by the confused expression on his face, my taxi driver from the airport seemed baffled by my desire to go to the Hotel Miralago, near Como. The hours ticked by as he chose to take the scenic route, via the Milan ring road, gesticulating at other road-users. Eventually we reached the town, rattled towards the lake, turned a sharp left, and found ourselves within the hotel district. Delighted by his achievement, my driver indicated that I should point at which hotel I had told him to drive to. This took longer than he had anticipated. Resorting to his satellite navigation system with a derisive snort, he demanded an exact address. Another 15 minutes elapsed as we circled around Como, threatening pedestrians. At this point, I decided to call the hotel and check exactly where they were situated, or if they had moved. Assuring me that they were still in Cernobbio, and not Como, I hung up and informed the driver of his error. Another 15 minutes, and my driver was beginning to direct his ire towards me, rather than the bemused citizenry. Seizing my phone and realising that he did not know the hotel’s telephone number, he handed it back with menaces and I dialled. Re-seizing the phone to remonstrate with the hotel’s concierge, a short verbal altercation ensued. He then reassured me in broken English that I was an idiot for telling him that the hotel was in Como, not Cernobbio. Five minutes later, we were outside an attractive lake-front hotel in Cernobbio.

Neither Como, nor Cernobbio, nor the Lake itself bore the slightest resemblance to Macclesfield.

The conference commenced the following morning. A wide variety of topics were covered during the three days of activities. The conference itself made full use of the Thursday and Friday, whilst the Saturday was given over to various workshops. I attended the TOSSAD workshop on Governmental, Educational, Usability, and Legal Issues towards Open Source Software Adoption in an Enlarged Europe. The first business of the conference was business.

Business

The issue of open source software (OSS) business models continues to vex and fascinate, with uncertainty over sustainability a lingering issue. Tony Wasserman (Carnegie Melon West), the first keynote speaker, indicated the resurgent confidence in OSS in the American market, citing the highly profitable takeover of JBOSS earlier this year. He spoke of slow adoption in large corporations, contrasted with the opportunities OSS presents to new start-ups thanks to low initial costs. He ran through the most prevalent business models for OSS and the most significant obstacles to it, defining the seven criteria for success as:

  • functionality
  • operational software characteristics
  • documentation
  • service and support
  • software technology attributes
  • adoption
  • development process

He noted that companies were starting to adopt open source development processes, and advised those wishing to persuade sceptics of the benefits of OSS to point to those companies already using it, such as IBM and Accenture.

Patrick Finnegan and Jeremy Hayes (University College Cork), and Björn Lundell (University of Skövde) attempted to broaden the terms in which OSS business models are generally considered, pointing out that a business model implied a way of doing business, not just an explanation of how money is made. A business model consists of a company’s culture, practices, management structure, and many other aspects beyond simply whether they make their money by offering support for OSS or via dual-licensing. Their panel explored notions of consortia of small companies exchanging ideas and forming networks, and trying to include their customers within these networks.

Open-sourcing Products

The keynote speakers Bertrand Meyer, Till G. Bay, and Bernd Schoeller (Federal Institute of Technology in Zurich),presented an interesting paper on their experiences of open-sourcing the EiffelStudio development environment. Eiffel was described as a bondage and discipline programming language, generally used for large industrial applications, but also a useful teaching language due to its highly-structured nature. The decision to make the software open source seems to have been made partly for defensive reasons – open source having become the standard for compilers and other tools and middleware products; and partly to help revive the language by broadening its developer base and market share. A dual licence model was chosen, with a stripped-down version designed to encourage the use of the language in education, and the full version maintaining income from industrial users. A big drive by the project team has been to utilise the pre-built community of computer science students, getting them to develop usable tools and libraries for the language as part of their degrees. This tactic seems to be having a good deal of success considering the recent development of some useful graphics libraries. The project had four warnings for others considering making their software open source:

  • It must be easy to build (at home, alone)
  • Fear regressions (e.g. do not change the build file format)
  • Have a development plan and publish it. A roadmap is useful so that people know where they’re going
  • Eat your own dogfood – i.e. always work with the changes you’ve just made, and have a nightly build.

Some of the questioners were a little doubtful about the idea of a development roadmap setting out exactly what the project was hoping to achieve and by when, suggesting that this might put off newcomers or give a misleading impression that it’s all going to get done without much input from outside developers. The panel recognized that this was a potential problem, but thought that the clarity of objectives gained by having a roadmap made it worthwhile.

Lunch and Dinner

Lunch was on your own, as the conference programme put it. And so it proved to be. As the Grand Hotel, Como, enjoys a privileged position somewhat removed from the hurly-burly of Cernobbio, delegates unwilling to afford the hotel’s 5-star hospitality were required to travel along the lakeside to the nearest patisserie (about 20 minutes at a gentle run) to obtain food. This may have been an intentional ploy to ensure that delegates did not spoil their appetite for the conference dinner in the evening.

The conference dinner more than made up for the slimming lunch breaks, consisting of a generous nine-course banquet within the confines of Villa Geno, a grand waterfront villa reached by taxi boat. The conversation flowed as freely as the wine, and delegates had several hours to get to grips with each other’s projects at leisure. Making the most of the opportunity, I spoke with, amongst others, Andres Baravalle (Open University) about the OU’s ambitious plans for OSS (see Niall Sclater’s presentation to the OSS Watch Open Source and Sustainability Conference about choosing Moodle as a Virtual Learning Environment) and his own research into the respective Web-based press coverage of OSS in Italy, the UK, and the USA. His research determined that, despite the relative difficulty of finding OSS migration test cases in the UK compared with Italy, this cannot be accounted for by an overtly negative or apathetic press.

OSS in Regional and European Government

Whilst the conference was an international affair, and attended by delegates from all continents (except for the usual exception of Antarctica), there was a strongly European flavour to proceedings. A number of papers addressed the adoption and use of OSS in regional and pan-European government.

The one global study was an interesting one, produced by a team from Istanbul Bilgi University (Harald Schmidbauer, Mehmet Gençer, Bülent Özel, and Vehbi Sinan Tunalioğlu). It concluded that the prominence of OSS in a given country had less to do with GDP per capita or Internet access and use than with educational incentives and information campaigns. The study’s prominence measure was the number of double-logged Google counts of “open source”. This finding should give encouragement to organisations such as OSS Watch, whose brief includes raising awareness of OSS and the issues surrounding it.

Patrick Ohnewein, the coordinator of CoCOS (Competence Center Open Source) South Tyrol gave a paper entitled A Proper Ecosystem for Free Software, which looked specifically at the OSS situation in South Tyrol, an autonomous region with a small population of about half a million and an unusually high rate of employment in its bureaucratic government. Ohnewein explained how South Tyrol’s public transport system (SAD) was an early adopter of OSS, switching to Linux in 1993, and how the rest of the regional government has been migrating since. In 2005, all the region’s Italian schools moved to F/OSS only. OpenOffice’s file formats (.odt, .ods, etc.) became the standard in 2006. CoCOS was formed in 2005 to explain to the population why they should be using F/OSS. Apart from the apparent cost benefits, the ability to translate software into the local language, Ladin, has been welcomed. As Ladin has only about 40,000 speakers, such a translation was considered uneconomical by Microsoft, but culturally important by the government. CoCOS have also provided a South Tyrolian equivalent to SourceForge, in response to the original becoming too slow. Ohnewein concluded his paper sagely, by remarking that we can’t force an ecosystem, only provide an infrastructure.

Bardhyl Jashari and Filip Stolanovski (Metamorphosis Foundation) presented a paper to Saturday’s TOSSAD workshop entitled Challenges and obstacles: Usage of Free and Open Source Software in local government in Macedonia. The Macedonian government entered into a strategic partnership with Microsoft in 2003 which largely restricts OSS to web servers and applications. The country’s 85 local government units do not fall within this partnership, although they too use Microsoft products almost exclusively. A survey of local government units was conducted, which found that the most common reasons for not using FLOSS were a lack of technical support, and a lack of knowledge about implementing and maintaining the software. The conditions that would need to be met before migrating were the availability of appropriate training and support, and the certainty that the migration would lower the total cost of ownership. Other concerns that were expressed include a lack of functionality in OSS (particularly when compared with tailor-made applications), and backwards compatibility.

A different survey was conducted by Bülent Özel (Istanbul Bilgi University), Julia Velkova (Bulgaria Information Society), and Manon van Leeuwen (FUNDECYT, Extremadura), looking at F/OSS in European National Programmes. This concluded that whilst some of the European public administration bodies showed a willingness to try OSS, adoption was in practice generally limited to local and regional administrations (as was the case in Macedonia). Public administration was found to be less flexible and open to innovation than the educational and commercial sectors, with many staff wary of being the first to adopt new technology. Many in public administration were unaware of the extent to which they were locked in to specific vendors, and they lacked a decision model to assist with upgrading or purchasing new software. One-year budget cycles were also found to be problematic, as they made it hard to properly estimate cost savings over time against immediate (and difficult to determine) costs. The study noted that where employees were not involved in the decision-making process, there was a high resistance to change.

Communications in OSS Development

Alongside the practical developers and those interested in governmental use of OSS, the conference also featured a number of proper academics – sociologists studying the behaviour patterns of open source developers. I had hoped that this might involve building mazes for them, occasionally picking them up and turning them around to see how they dealt with disorientation. In fact, they were more concerned with developers’ communication patterns, and how social networks evolve.

Sandra Slaughter and her team at Carnegie Mellon University have been looking in particular at how an individual’s position in a social network determines their success. This heavily statistical paper found that a handful of key developers who were centrally located in a project’s social network were driving communications within the project, apparently conforming to Robert Michels’ Iron Law of Oligarchy. It was suggested that the research might be of additional interest if carried out at intervals over time, to see how this situation changed, rather than just taking a snapshot.

James Howison (Syracuse University) took this diachronic approach in his well-presented paper on Social Dynamics in FLOSS Communications. He looked at many OS projects, but specifically at the development of Squirrelmail. His research involved the study of email postings relating to bugs that prompted responses, rather than the who answered who approach taken by Slaughter. Howison opened with the premise that centrally-led decisions are more efficient; distributed decisions are more rewarding for participants, and worked with the hypothesis that open source development would, in practice, prove to be a fairly centralised process. Surprisingly, his finding indicated that there were multiple core nodes in several social networks, and in the case of Squirrelmail, periods of centralisation were followed by very democratic periods of communication, preceding shifts in leadership. This was not the case in all projects, although there was nevertheless a greater degree of distribution than the initial hypothesis had projected. Most projects had no changes at the centre, but also participants who did not stick around, often staying for only 10% to 20% of a project’s life span.

A third social project was described by Jae Yoon Moon (Hong Kong University of Science and Technology) and her team, although this was still in its early stages. They intended to examine the impact of prior collaborative ties on open source project team formation. The initial findings suggested that people are indeed more likely to join projects with other prior collaborators in them.

Moving away from OSS a little, Mehmet Gençer (Instanbul Bilgi University) and his team were studying the Organisation of Internet Standards using social network analysis, identifying the subgroups involved in determining standards and their relationships with one another. Their results showed major groups centred around key standards tracks. There were two almost entirely separated super-groups: one related to Internet Protocols and security extensions, the other to network management, email, domain names, and so on. Gençer argued that the analysis could be applied to similar systems, and might be able to help identify structural bottlenecks and thus assist workforce allocation to projects such as the development of Debian distributions.

Learning and Teaching OSS Development

A significant strand of the conference and TOSSAD workshop concerned the pedagogy of OSS development, i.e. not how OSS can be used in education, but how students could be educated in OSS.

Sulayman K .Sowe, Ionnis Stamelos, and Ignatius Deligiannis (Aristotle University of Thessaloniki) have been implementing A Framework for Teaching Software Testing using F/OSS Methodology in their introductory course to software engineering at Aristotle University in Greece. Preliminary findings suggest that their undergraduates made successful bug hunters, and many also made code fixes, despite not having to as part of the course. The use of OSS methodologies made an interesting contrast to traditional teaching based on closed source methods. 90% of the students said that they would continue to contribute to their chosen projects after graduation. This is obviously an encouraging finding with regards to the future growth of OSS.

Enn Öunapuu (Technical University of Tallin) opened the TOSSAD workshop with a paper on Determining Best F/OSS Training Methods. He ran through the various learning intelligences (visual, aural, kinaesthetic, etc.) and means of considering learning (as activity, cognitive, situative), before explaining how he intended to develop an adaptive learning system that responded to the observed learning preferences of an individual, which could be assessed by means of a questionnaire. This seems like a nice idea, although the project is still in its early stages. It is intended to assist the teaching of OSS development methods, but might be of broader application if proven to work.

The softly-spoken Mehmet Gençer (Instanbul Bilgi University), in his second presentation of the conference, spoke of the proposed MSc Programme in Open Source Information Systems at Bilgi, being set up by himself and his colleagues Bülent Özel and Chris Stephenson. He explained that interest in OSS in Turkey is increasing rapidly, fuelled primarily by commercial demand: 25% of Turkish IT job advertisements now ask for Linux experience. In contrast to other European courses in OSS, this was intended to be strongly interdisciplinary, being accessible not just to software engineers, but also to those in information law, political science, economics, and business administration. The team felt that this interdisciplinary approach was necessary for open source to grow in the longer term, and it demonstrates that OSS is increasingly being conceived in broader terms than simply as a software development method.

Migrating from Proprietary Software to OSS

Whilst technically-minded people such as sysadmins and software developers are generally happy to consider OSS for applications, if OSS is really going to rule the world, commoditizing all before it in a rampage of total cost of ownership assessments, it needs to be accepted by Joe Public, or at least Joe Office-Worker. The feasibility of weaning the masses off the opium of proprietary software was addressed by a number of delegates.

Bruno Rossi, Barbara Russo, and Giancarlo Succi (Free University of Bolzano-Bozen) attempted the Introduction of Open Source Software in the Public Administration, looking at users’ experiences of switching to OpenOffice from Microsoft Office. The motivation for the migration was, naturally enough, financial. Their findings suggested that there was little difference in the nature of the work produced using OpenOffice (perhaps unsurprisingly), but that there were still some usability problems with the OS alternative. They recommended that organisations not use brute force to push through such migrations, but rather allowed the use of both packages together for an extended period.

One of the conference’s more charismatic speakers, Jean-Paul Van Belle (University of Cape Town), presented a paper entitled Critical Success Factors for Migrating to OSS-on-the-Desktop: Common Themes Across Three South African Case Studies, written with Daniel Brink, Llewelyn Roos, and James Weller. This began by discussing the particular situation in South Africa that made the country receptive to OSS, before enumerating the success factors that the case studies drew attention to:

  • a financial motivating factor
  • top management support
  • user awareness and communication
  • detailed planning
  • training
  • a pilot project and partial migration
  • good support.

The influence of the Shuttleworth Foundation was also a positive factor in the success of OSS in the country. Allegedly, Microsoft has responded to the threat of Linux adoption in South Africa by offering all schools their software for free (as in beer).

Selene Uras (University of Cagliari) introduced delegates to Polaris40S: a best practice for training and adoption of F/OSS in SME [small to medium-sized enterprises]. This involved training the management and staff of several IT companies in F/OSS methodology and skills, and then measuring their responses over nine months. The results were very encouraging, with 82% of the developers involved in the project providing a positive judgement. The difficulty of initially understanding F/OSS culture was identified as the most critical obstacle in the way of exploiting the opportunities enabled by open source methods and working practices.

Chris Stephenson (Istanbul Bilgi University) delivered an amusing account of his Computer Science department’s adoption of OSS in despite of university politics and entrenched interests, and its unintended but positive consequences for the university. The motivations behind the switch were cost, system integration, and to some degree, it seems, simply to differentiate themselves from the IT department. As OSS is free, the department did not have to get approval for purchases via the Byzantine central purchasing system (it appears that elements of Byzantine culture still live on in modern Istanbul). Other sections of the university noticed the speed and low cost at which Computing Sciences could get things done, and started to go there for IT assistance rather than to the official IT department, but this then led to unrealistic expectations being placed on Computing Sciences that they could do things for nothing, such as automating the entire university’s library system. The conclusions reached in this paper were that switching to OSS increased the flexibility and problem-solving capacity of the Computer Science management team, and also that, after a long period of resistance, it was possible to convince university management that cost savings and painless transitions were indeed possible with OSS. Stephenson’s advice for others was that one should not run into presenting technical aspects and [a] cost-savings perspective without a comprehensive picture of cost of ownership and [a] transition road map.

Oliver Strauss and his colleagues at the Fraunhofer Institute for Industrial Engineering advocated the use of wikis to support OS adoption in small and medium sized companies. He explained that a wiki was the ideal collaborative discussion and comment forum for colleagues to help each other get to grips with OSS adoption, a process which requires and produces information. As wikis involve users, this would be likely to better motivate them and prevent them from being discouraged or frustrated by the migration process. Strauss did point out that users would need to accept and use the wiki itself if it were to be of any help.

One of TOSSAD’s objectives is to help developers improve the usability of OSS, to which end a survey was designed to gauge users’ reactions to Open Office, Linux, and Firefox. Daminano Verzulli (PDA Communication), Uros Jovanovic (Xlab), and Görken Çetin (Scientific and Technological Research Council, Turkey) concluded that the usability of these programs was high, although Linux still needed to improve so as to be properly usable on the desktop for newcomers. A reduction in the learning curve when migrating from MS Windows was called for. Linux’s failings were perceived not so much to be due to the complexity of the installation process, but rather due to the lack of straightforward start up help and documentation. They observed that this was not a F/OSS specific problem, but rather an issue relating to all software production.

OSS usability was also of concern to Ellen Reitmayr (Relevantive AG, Berlin), Björn Balazs (Apliki, Berlin), and Jan Mühlig (Relevantive). Recognizing that OSS organisational structures and working practices posed new challenges to ensuring good usability, they looked at case studies of projects in which members of the OpenUsability Intitiative had become involved. The aim of the OpenUsability Initiative is to foster the establishment of User-Oriented Design processes (according to the terms of ISO 13407). Prior studies had identified that OSS developers have a tendency to disregard feedback from common users and to judge the contents of wish lists on a technical basis. Results from the new study suggested that usability experts should find development leaders and collaborate with them, or, failing that, they should look for the common goals of a project and gain developers’ agreement on these. Usability experts should, in particular:

  • report results and suggestions to developers early and often
  • identify theories about user habits
  • survey actual user expectations
  • gather user profiles
  • persuade developers to limit the target user base to a feasible group

Additional Notes on Software, Licences, and Organisations

Software Tools and Development Projects:

  • ASLA (Automated Software Licence Analyzer). Timo Tuunanen introduced this software that retrieves open source licences from source code, so as issues such as code re-use and licence compatibility may be properly addressed. It may be extended via new licence templates and has about a 90% success rate at present. LIDESC is a similar tool.
  • ROSE (A Robust Open Source Exchange for OSS Development). Amit Basu (Cox School of Business, SMU) envisaged an exchange enabling developers to find collaborators and share knowledge and code, working on a payment model such as Agoric systems. To avoid repeating the failures of the 1990s, Basu is working on a system that enables both anonymity and accountability.
  • Gregory L. Simmons (Unversity of Ballarat) is developing an OSS metadata schema to help overcome the high degree of conceptual dissonance between projects (at present people understand their terms of reference differently from one another). This would more efficiently enable developers to join appropriate projects.

Opinions Concerning Licences:

  • Andrés Guadamuz González (University of Edinburgh) argued that the proposed GPL v3 had lost clarity and gained length, and that it contained explanatory paragraphs that were out of place in a legal document.
  • Darren Skidmore (University of Melbourne) admitted that there are a lot of OSS licences, but nevertheless felt that more licences would probably be required to cater for particular stakeholders, especially due to changes in jurisprudence, new participants in the OSS industry, and the globalisation of software usage.

Organisations:

  • ObjectWeb. Jean Pierre Laisne introduced this consortium of OSS companies working with governments as middleware providers. He stated that a benefit to governments of using OSS was that they could employ local companies for support and development. Another advantage was that the use of open standards ensured compatibility for long-term projects.
  • NESSI. Stafano De Panfilis introduced NESSI, the European Technology Platform. NESSI seeks to influence investment in European research, promoting open standards and interoperability.
  • IDABC (Interoperable Delivery of pan-European eGovernment). Barbara Held introduced this EC body explaining and promoting OSS.
  • IFIP (International Federation for Information Processing). This venerable organisation, established in 1960 under the auspices of UNESCO, signalled their intention to promote OSS through lobbying and publications.
  • Free IT Foundation. This recently-formed foundation arranges funding for projects licensed under GPL-compatible licences. It is associated with IFIP.
  • AICA (Associazione Italiana per l’Informatica ed il Calcolo Automatica, the Italian Computer Association). Giovanna Sissa introduced AICA, the body in charge of managing the European Computer Driving Licence (ECDL) programme in Italy. AICA have become the first such body to offer the ECDL certification in OSS, so as to promote vendor neutrality. It is undoubtedly good news for the open source movement that general users are now being given the option to learn their computing skills using OSS.

Conclusions

It is now almost a month since the conference. I have had time to digest the presentations and the nine-course dinner. What conclusions can be drawn?

  • That research into OSS is beginning to come of age. Several delegates mentioned that research into OSS relied too heavily on anecdotal evidence in the past, and that it lacked case studies and quantitative analysis. From the papers presented, this would appear to be changing. As OSS matures, so too does the research it engenders.
  • Some current academic research may in time feed back into OSS development methodology, influencing the way lines of communication are managed in particular.
  • Progress is being made towards a viable, enterprise-ready, completely open source desktop computing system.
  • Strategies for encouraging and managing mass migration to such a system are being honed.
  • Awareness and uptake of OSS continues to grow.

Further reading

Links:

Related information from OSS Watch: