Papers

This issue is the heart of the study, "Pooling Open Source Software", which was financed by the Commission's Interchange of Data between Administrations (IDA) programme: How could EU suppport a kind of software clearing house in which various administrations could "donate" software for re-use.

With the cost of e-government soaring in Europe, and estimated to increase by 28 percent to 6.6 billion euro in 2002, sharing software could help save money as well, and lead to across-the-board improvements in efficiency of the European public sector. Money can be saved not only on re-using existing solutions (“not re-inventing the wheel”), but also on maintenance and upgrades if a wider community of developers share knowledge and provide improvements.
"Good practice is built on proven solutions that work," said the EU Commissioner Erkki Liikanen. "Software and concrete applications that work in practice are an important element of these. They could be usefully employed as a source of inspiration for member states to develop good and interactive public services in the future."

The title of the study “Pooling Open Source Software” (or POSS) is based on the suggestion that e-government software should be issued by administrations under an open source licence, allowing other administrations to consult and transform the software code freely (for example, to respect local language or regulations) and to re-distribute it to all their users without restrictions. As most software distribution related issues are not technical, but legal, the study explored the various possible “licences”(or contract allowing one administration to re-use the software developed by other administration), their compliance with copyright regulations and other possible issues as patent law and impact on commerce and industry.

The main recommendations in this field are:

• The donor of the software must have the right to distribute it. This is dependent on contracts with employees or withproviders and the possible use of existing “components” to elaborate the software.
• The donor of the software must adopt a “licence” and the choice of this licence is his sole responsibility, but he should select an existing standard licence model and should not try to draft its own specific “public sector licence”.
• Three standard licences are recommended with a “selection guide”, depending on the way the software has been developed and on the government policy: they are known as the BSD, GPL and MPL licenses.
• The future pooling service must create a complete contractual framework, to define the roles, duties and rights of the donors, users and the service itself, to indicate the applicable (European) law and competent court if needed.
• To prevent donor liability issues, it is recommended that the pooling service should submit cases to one external specialised legal adviser, not only to benefit from external liability coverage (as legal advisers are professionally insured) but to ensure consistency and avoid the “over protection” that should happen if each donor’s legal departments had to consider the case.

Another issue is more technical and related to the public sector software development: Is it “done for sharing”? Indeed, trying to pool any existing software as open source just to benefit from free maintenance and upgrades is an illusion, if the software does not respond to precise development conditions:

• a reasonable number of persons “sharing the same problem”;
• initial but flexible repartition of “ownership / leadership” between diverse persons, from diverse organizations;
• documentation everywhere;
• a roadmap (navigate into the code as in a website);
• a common, easily understandable trunk + functional modules / no monolithic cod;
• software organised in many relatively small pieces of code (in order to facilitate individual ownership);
• clear identification (and declaration) of which parts are “mature/stable” and which parts have “room for improvement” (according to the “release often” principle);
• Launch permanent discussion forums on requirements, objectives, priorities for further development.

In consideration of all these conditions, the decision to “go open source” may represent a serious initial investment for the donor (on existing software) or a serious development cost increase (on new software): a multiplication of the initial cost by 2 (or even by 3 if the development aims
to be multi-platform) that will be recovered long term (over 5 years).

To write down the POSS pooling service requirements, Unisys processed a hundred questionnaires and interviews in public administrations all over Europe. The study conclusions were that the sharing should not concern only software, but knowledge in general and all information that is required to deliver “best practices” to citizens and enterprises. In addition, the service must operate in collaboration with existing services in member states and not generate a new “mammoth administration”.

The POSS expected services are:

• a multilingual portal with a clear site map, pan-European content and standard data format;
• registered member management;
• software submission, description, classification and download management…);
• links to national initiatives and addition of any other link related to projects (soft-ware) and members;
• library management to attach various documents related to software (technical documentation, user guides, knowledge and competences regarding an European framework of open standards);
• news, forums and mailing lists;
• opinion surveys;
• software hosting for downloading (FTP) directly from the POSS;
• a scientific committee. Even if no formal guarantee or quality audit is possible, the conformity to formal quality criteria (documentation, manuals) and available test scenarios, as well as the the recommendations of experts corresponds to a frequently expressed need;
• legal advisory service, and contractual framework management (including the registration or logging of all contracted downloads, by authenticated users, software, time stamps, agreed licence types).

Based on the above requirements, the Unisys report presented an example of all the software tools and components that should be integrated to construct the service, based on tools used by other
similar sites and that are proven to be compatible (Operating system, Web and FTP servers, Database, e-Mail, Statistics, Link checker, Scripting language). It is significant to note that most of these components are also publicly available as “open source software”.

The service delivery maintenance and roles were also detailed in the Unisys report, as most of the POSS cost will be spent in maintaining it long-term rather than just creating it. The system administration, the software selection, evaluation and classification, the news, links, events and content management, the members registration are all human driven tasks that have their cost: pooling knowledge is a service, not a software…

A quality service will require the same level of investment as any other pan-European service with constraints concerning the quality, the number of languages, the availability and the neutrality. Based on a similar project (in Canada) the global TCO (Total cost of ownership) of the service was estimated at 6 million euro over five years, with no immediate return on investment, as the initial effort to evaluate and “adapt” existing software to open source pooling will be substantial and as new projects take time to reach maturity. But, as was said before, the valorization of the
6.6 billion euro spent each year in software 2002 justifies some investment…

For more information on the report or the IDA programme, and a full download of the study, visit the EU IDA Website: http://europa.eu.int/ISPO/ida/