Meeting on European perspectives for open source software

18 May 2001 - Brussels - Breydel I 9/256 - 15:00-17:30

Participants:

Experts:

Mr Toon Knapen, Consultant, Si-Lab, Belgium

Mr François Etienne, CNRS-IN2P3 (Data-Grid Initiative), France

Mr Jean-Pierre Laisné, President, Linbox, France

Mr Marc Peltier, VP Services, Mandrakesoft, France

Mr Gerard Vandome, Object Web Initiative, INRIA Rhône Alpes, France

Mr Norbert Munkel, Alliance Manager, SuSE, Germany

Mr Hartmut Pilch, FFII / Eurolinux, Germany: Mr. Harmut Pilch was not present, Mr. Jean-Paul Smets spoke on his behalf.

Dr Leena Suhl, University of Paderborn, Germany

Mr Jan-Oliver Wagner, Intevation GmbH, Germany

Dr Mauro Giorgetti, T6, Italy

Mr Massimo Lo Iacono, Consorzio Pisa Ricerche - gruppo CREA, Italy

Mr Francisco de Urquijo, Chief Strategist, Ximian, Mexico

Ing. Jeroen Baten, Stone IT, Netherlands

Dr Malcolm Herbert, Red Hat Europe, UK

Mr Michael Gough, National Computing Centre, UK

Mr Richard French,Sr. Vice President & General Manager, Open Source Development Network, USA

Excused: Per Öster, KTH Open Source Lab, Sweden; Dr Roland Alton-Scheidl, Public Voice Lab, Austria, Regine Schickentanz, T-Nova Deutsche Telekom, Germany

European Commission:

Jacques Bus, Head of unit INFSO/E2

Philippe Aigrain

Peter Ritzmann

Peter Diry

Michel Lacroix

• Introduction by Jacques Bus

Presentation of the objective of the meeting: obtain input on research and other policy orientations that would be useful to stimulate the open source software world. Recalls the current state of preparation of the 6^th framework programmme, and the basic structure of the Information Society Technologies priority within it. Asks for inputs to address beyong technology to be developed also the kind of structural support adequate for open source software, and possible orientations including for regulatory policies.

• Presentations and position statements

Norbert Munkel, SuSE: [Addressing the list of issues for discussion in order] Assessment of range: the IT infrastructure and everything with a broad base of potential users. Important topics to be addressed: inter-operation between software components, security issues, generic components and platforms of embedded software. Opportunities: increase the local content (the share of value chain). Missing to achieve the benefits: knowledge of open source in opinion leaders. Key technologies, process, skills and economic challenges: none except that open source is not yet coming big. Conditions: patents are a general threat to software industry in general, not just open source. Financial incentive for projects in the public sector using open source approaches would be useful. Efforts to directly steer co-operation also.

Jan-Oliver Wagner, Intevation: Main missing elements: many independent collaboration services and infrastructure. Not jut SourceForge. Independent infrastructure needed to avoid risk of shutting down (example: shutting down of SourceXChange). Public libraries have responsibility in storing and giving access to free software. Make opne format requirements for software used in the public sector. More companies to follow free software strategies consistently. Free software companies have service business models. The switch from mainly-license-fee business model to service-only will be a general change in software economy. Key technology, process, skills, and economic challenges: making software is a communication problem more than a technology problem. Free software is based on broader communication opportunities: close co-operation between developers of different companies. Developers communicating with users. One of the major challenges for the economy is to establish new structures of communication. Conditions: benefits of free software are the absence of proprietary locks, the fact that the results are always available, whether they are useful or relevant of course not guaranteed. 100% funding, rapid grants justified. Discussion on the reasons for the 100% funding being needed. A. Lefebvre (France Telecom): large corporations do not have the same funding problems than smaller entities. Funding single talented people. Modalities of selection: stick to the simple heuristics of science, judge the methods of working and communication of people. Long-term perspective always important for free software. Obligation to release under free software licenses. Provide for and ask to use independent collaboration infrastructures.

Jean-Paul Smets: On how to stimulate more businesses. US Small and Medium Business Act does not have a counterpart in Europe. Public administration can often act through procurement policy to make it more open to small entreprises and free software solutions. Malcolm Herbert: regulation is maybe not needed, this is a business practice problem. JPS: tenders do exclude some open source solutions. Now intervenes as representing Harmut Pilch (FFII/Eurolinux). Recalls the history of the foundation of Eurolinux . Funding: David Faure has made one the best browsers while being an unknown student. Many free / open source developers are like him, and the programme should be fit to fund such individuals as he was at the time. Other conditions: Juridical terrorism created by software patents and the new copyright directive are a major problem. Software archives are needed (dating of free software). Eurolinux operates a service but it should be a universal service: it can be privately operated under a clear mission. Problems in financing and structuring networks of SMEs active in open source (similar to clusters in the textile production). Research funding should not create incentives to produce proprietary software rather than open source.

Richard French/OSDN: Responses from SourceForge to comments previously made on risks of absence of multiple and independent collaboration hosting entities: SourceForge does not remove software even if used. We get lawsuits daily. We believe in the first amendment protection for software. DMCA is a big danger. Open source is not American or European, it is world wide. Source Forge itself is built worldwide, it is a critical cultural component. SourceForge will put a mirror in Europe in the coming weeks. Major issue for the growth of open source is an infrastructure issue. This infrastructure will cost a lot of money: we are spending a lot of money. We are spending 250,000 $ a month just on bandwidth. Is SourceForge is not profitable on its own? No, but other indirect schemes fund it. Infrastructure is not just technology and money, it has a critical cultural component. Michael Gough: the whole success depends on usage of open source outside its community, its external credibility. The National Computing Centre was historically called the National Program Library. So the concept of a library does apply here, but it is not just a warehouse. It must include validation, maintenance, possibly certification.

François Etienne, Datagrid/IN2P3: middleware components for transparent access to "infinite resources" in computing and data storage resources. Driven by high-energy physics, bio-informatics and earth obversation secondarily. Awareness of open source was initially not strong but all software was distributed. A memorandum of understanding is being elaborated to clear intellectual property issues. Certification is an issue that must be addressed: labellised software repositories. Who will take care of the middleware after the project?

Francisco de Urquijo, Ximian/GNOME: At some point GNOME development went so fast that going to venture capital to pay developers was the only way to proceeed. I am going to focus on strategy. Software progress has not followed the exponential progress of hardware. In 1973, software progress should have exploded following the invention of the component. But it did not occur, due to imperfections in markets. The solution lies in the educational system, when research centres will be able to produce software solutions. 2 basic component are missing on top of OS, UI and generic office applications: an organisational operation system, a knowledge operating system.

Leena Suhl, University of Paderborn: 2 aspects. The education as an application area. Most European universities are public. They develop software (open source or not). There should be more exchange of this software and knowledge. Why so few women in open source software development? Programming 24 hours a day (or developing during the day for a pay, and doing it again in the evening as a hobby or a mission) is not appealing to women. Women have other, broader interests. Lack of usability is also a blocking factor, because women are not interested in solving a technical problem for the sake of solving it, but more about using tools to do things with them. But freedom and choice fascinates women as much as men, so these problems can be overcome with the proper organisation and new ways of working.

Mauro Giorgotti, T-6: problems encountered using open source software strategy in the FETISH project (middleware for tourism system information and service discovery). Service-to-service: the architecture is based on Jini (not OSS licensed). Project 100% funded, so we would like to realease the integration tools under lGPL. So our project is mixed between licenses. One layer will be generic open source services, one other. I do not agree to some statements, like SMEs have constraints in access to European funding. It seems to me that the European Commission already cluster projects and specifically supports them. These resources are there but not used. Time to market is one blocking element. Not enough communication. It is not a structure or funding problem. What is missing are a clear quantitative analysis of the quantification of usage of open source software. This is really a problem for integrator companies like us.

Malcolm Herbert, RedHat: Biggest challenge: what will be the economic model of the open source world? Will hardware dealing companies like Dell really back up open source?

Jean-Pierre Laisné, Linbox: Independent centres, or portals are really needed. The debate about functionality should be in the open not in restricting. Why think in terms of centralising: there can 100s of different sites, even supporting specialised projects. On the business: sustainability is a critical issue in the open source software world. To build good software, you need time. But do companies such as ours have it if they are threatened in their short term survival? We are not creating profit but value: tools for people to communicate, to create. Provision of services is the solution but also a problem: you need as many people as projects. Large government organisations are more advanced than companies in their acceptance of open source. Discussion on whether mixed economy entities (with co-operative statute) can be a solution to the sustainability problem.

Willam Cobert, VA Linux, Europe: open source is global. We are ready to contribute to efforts in Europe.

Toon Knapen, Si-Lab: the aims should be more focused on process rather results. Exhibit directly the service aspect of open source software. Best practice and take-up measures or initiatives like Europort.

Marc Peltier, Mandrakesoft: on the business issues. Some larger Linux companies have been created with venture capital funding. Thanks to that, they have been able to fund developers (contribute them to projects). The public programmes could do the same. [PhA: how to select people? Discussion on whether "host fellowships" would be an adequate scheme.] Investors no longer have the sense of humour. The business model is very hard . Traditional services Linux companies don't work: who is buying services? large organisations and they are hostile to risk, and perceive anything but leading installed base as a risk. Inter-operability and migration from proprietary are the critical enablers. Patents are really a big risk: we are really convinced that they can kill all momentum for software innovation in Europe.

Jeroen Baten, Stone-IT: Regulatory measures on open file standards in civil organisations would be welcome, they would break open the market, and enable the best in the market to win. Reading a .doc attachment is a critical user requirement. Import filters for proprietary formats are a perpetual nightmare. Open file formats being compulsory will just give a fair chance to open source software.

Patrick McGovern: On the comments on SourceFourge: on code being on one place, all the code is open source and people can create mirrors as they wish. Patents and juridical risks: similar concerns than the other speakers. Assessment on the range of software: open source and related services will be everywhere in the next years. The present growth rate is enormous. But the business model challenge is a real one. Jeroen Baten: the service model is definitely sustainable, we are a profitable 35 people company. Richard French: there is whole mental revolution needed on what is the reasonable growth target for a software service company compared to product software. Intevation: we are also profitable, but not with a closed source product type of margin.

Jacques Bus summarises the discussion:

• Regulatory: tendering, open standards for files, worries about patents and copyright enforcement

• Infrastructure: communication, development, archiving

• More focus of measures to support process

• It is to the open source software world to convince the "rest of the world"

• Role of universities, education obviously important

• Gender issues: seems to particularly important in this field

• On smaller entities and their co-operation: support to clustering

• Fellowships: useful but with selection done through organisations being selected to host them.

Follow-up: the minutes will be circulated for comments on 28 May 2001. An analytical report will then be drafted and published on the IST Web site after approval by participants.