An open source track at the ACCU spring conference 2004

by Sebastian Rahtz on 15 April 2004 , last updated 13 July 2005

Introduction

On April 15th, I attended one day of the ACCU Spring 2004 conference (http://www.accu.org/conference/). ACCU started as an association of C & C++ users, but now cover Python, Java etc as well. For this conference Andy Robinson of Report Lab (http://www.reportlab.com) organized a track about open source, so I went along to find out where this gang of professional programmers meet OSS. This is a personal view of part of the meeting, so don’t expect a considered OSS Watch view as you read this!

The conference took place at the Randolph Hotel in Oxford (familiar to Inspector Morse lovers) and looked to have an audience of over 100. As you might expect, mostly men, mostly normal enough but a definite sprinkling of middle-aged ponytails and creatures of the night.

How to fund open source

Paul Everitt, Zope Europe, chaired a panel session on how to get funding for developing software.

Steve Alexander (http://www.SchoolTool.org) described the ideas of Mark Shuttleworth (the South African space-travelling millionaire) who funds software for school administration, partly using a bounty system; in this model, someone offers cash for the first person to solve a problem. Could this work in other systems, ie offering $200 to fix a bug in a bit of software? Shuttleworth himself has an open bounty for a fix to Bugzilla; see http://www.markshuttleworth.com/bounty.html for details of his schemes. This model is quite nice for when you want to keep a foundation doing the overview and design separate from the funding, and for when users really want to solve a particular problem now but don’t have their own resources. What if two people apply at the same time? It needs a brokering system to get people registering an interest or working together.

Peter Hollands (Cisco foundation) discussed general ideas of funding; were we interested in sustainability of projects or of programmers? — not necessarily the same thing. There is money coming from foundations for the public good; the big deal is giving IPR to humanity/society, modelling the fact that most people make small increments to existing knowledge. Open source is doing for society the same job that public libraries and other services did.

Aljosa Pasic is an evaluator for European Commission projects, as part of his day job working for a Schlumberger IP service company, employing 50,000 people. He talked about how to get the EC to pay for software development. First off, the EC does not fund OSS, it funds good ideas; but open source is a plus. Evaluators have one hour to read a proposal, so bear that in mind when writing an application. Don’t worry about budgets or workplans, get a good idea, in line with the commission’s eGovernent standards and plans. Get a lobbyist or join up with someone else, or you won’t survive.

In questions, the following points were made:

• The EC does not fund existing projects; but philanthropists might. Either of them might also pay for open-sourcing existing closed-source software.
• All these funding ideas assume that there is a good story and a good implementation plan behind it. Get a professional help!
• What’s the point of working on EC funding, as most of their projects never deliver? Are software companies actually capable of producing software anyway? The industry does not have a good track record compared to other engineering areas.
• Don’t forget that raising money is a competetitive and hard business. Why not join with universities, who do lots of this?
• Spend money on writing good specifications. Don’t expect people to fund unplanned coding.

Open Source Licensing 101

David Ascher, of ActiveState (http://www.activestate.com/), recently bought by Sophos, gave an introduction to licensing for developers and for business folks.

Businesses compete on core competencies, and don’t do things they don’t have to, because writing software is expensive; businesses need to control risk. The motivation of OSS programmers has to do with pride and making a difference (perhaps this is a narrow traditional view of the OSS programmer?) Business people, on the other hand fear anything which threatens the plan. However, business and OSS are not incompatible; there is plenty of scope for partnerships and shared interests.

David introduced the idea of four licence groups:

• Copyleft: GPL
• Liberal: BSD, Mozilla, Apache
• Corporate: Sun, Apple, Netscape
• Library copyleft: LGPL as well as non-OSI licences like that of Microsoft. He noted the danger of GPL use stifling development, and thinks people often choose it by default, incorrectly. What people probably intend is that people must simply put back changes into the common pool (reciprocity). How far does the idea of derivative go in software? There are complications of combining licences when GPL is involved.

IBM’s take on Linux

Richard Moore talked about what IBM’s involvement in Linux is about.

Linux is a huge success; it has arrived in all sorts of areas, especially related to web services. IBM has seen a new growth of virtual Linux services on 390 mainframes, and also in physical clusters. IBM do lots of work on Linux to make both these types of solutions work properly, and care about device drivers. They also deploy Linux on desktops very happily. IBM has its plans to drive Linux forward, and see plenty of virtues in having a standardized open system which does not force people in particular directions. This lets them offer optional support for serious money. Radically, the cost of support now relates directly to its value for the client business, not with hidden amortizing of development cost (since the development effort is effectively free for IBM). Some of their priorities for 2004 are (unsurprisingly):

• their hardware (especially high-level). Linux is well-designed for this
• security
• virtualization, for clustering situations
• Windows interoperability
• kernel 2.7
• their On Demand system, by which companies get IT stuff as and when they need it, not in advance just in case

More details on http://www-124.ibm.com/developerworks/oss/linux/.

Hewlett Packard and Open Source

Mark Butler, of HP Labs Bristol, outlined what HP are up to.

HP likes open source, coming from their programmers at the grass roots. They like it because their work gets used by real people out there. They don’t do corporate licences, they use mainly BSD and LGPL. They concentrate mostly on Linux-related stuff — Samba, hardware support in distributions, Apache, Ant, Cocoon. At the Bristol labs specifically, they do Java work, related to W3C standards. Some of their projects:

• DSpace: digital repository system
• Jena: semantic web framework, stuff for doing OWL, RDF, etc
• DELI: delivery of web content for multiple devices
• FOA / WH2FO: software for dealing with XSL FO. Word to XSLFO, and a graphical editor for XSL FO

HP use OSS to encourage uptake of new technology, both inside and outside the company. They see openings for consultancy by small and large companies. They like the idea of sharing development costs with other companies. There are social advantages internally because of the emphasis on getting things done, and breaking down organizational barriers. There is also a danger, however, of macho competitions between programmers. Mark noted that organizations like Apache are important because they sort out some of the problems associated with collaboration.

More details on http://opensource.hp.com.

Software patents

A collection of folks talked about the situation with patents.

Puay Tang, Research Fellow at the University of Sussex. Software patents are a high-profile business at the moment, with strong opinions for and against. The for argument says that patents: - incentivize innovation - good returns on R&D, thus good for business - patents don’t conceal information - SME’s have no way to survive without some protection, and patents help them attract venture capital - helps competitiveness of European software industry against USA (so believes the EC); actually, this argument was used by the US against Asia in the 1980s (though the empirical evidence is rather that this did not have effect) The against people says that patents: - impose a barrier for small firms who cannot afford to play; big companies use it to threaten small competitors - locks companies into negative wars - inadvertent infringement of patents which are anyway written to maximize the chance of claims by the owners; whole process (registering takes 18 months or so) is too slow anyway - software is cumulative; it does not make sense to patent snapshots in the process - the volatile and vague nature of software development makes it hard to pin down what is happening; hard to pin down prior art, and to decide what is non obvious; lots of bad patents get into the wild

Steve Probert of the UK Patent Office said that his normal life was people being angry about patents not being issued, so it was odd being with people who were against the system. He described how the patent system works. The process is in four steps:

• search for relevant prior art by an expert examiner in the area for a day or two; this is reported to the applicant, who often withdraws at this stage. That stage usually takes about 6 months.
• after up to 18 months, the application is published for the first time (A publication); this has no relevance to how likely it is to be approved, but is always done to allow other parties to examine what is claimed
• substantive examination by an expert who compares claim against search results, and assesses whether the claim is properly written (ie it discloses enough to be shareable by others, its scope is clear etc)
• publication if patent is granted, after examiner is satisfied (B publication). This can include changes by the claimant after questions from the examiner What gets given a patent is ultimately determined by the court system; the patent office cannot override a judicial decision on whether something is patentable. This can be an expensive procedure! The law is not really clear enough at present, so there is a need to formalize the last 20 years of case law. This is what is driving the EC discussions, to get a clear, harmonized, law across Europe.

James Heald of UCL, a key figure in the campaign against software patents (see eg http://www.ffii.org.uk/), talked for the FFII. This (http://www.ffii.org.uk/) is a grass-roots organization of small business, users and programmers which exists to lobby against software patents. The European Commission is currently trying to find an agreement between governments by May so that they can go back to the European Parliament and try to dump the Parliament’s September 2003 rejection of software patents. The UK is behind the Commission, other countries are more ambiguous; it now seems to be down to political machinations whereby countries horse-trade deals. Other players are Nokia, who strongly back patents, and organisations like FFII who had a walking demonstration in Brussels on April 14th and have been promoting a net strike. Still plenty of technical argument about what computer program as such means! It seems to be a matter of whether the program has an essential facilitating step in doing something—people agree this is patentable. Beyond this it gets vague. Do we patent algorithms? data structures? file formats? processes? FFII believe that software patents will do huge damage to open source. The problem is much worse than that relating to copyright (cf SCO), because patents can catch inadvertent infringement. They also don’t think patents work in the software industry anyway, as that works by competition, copyright etc.

Eric Raymond had a quick rant. He said he was not against patents per se, even in some cases for software (eg RSA algorithm); but in practice the experience in the USA had been that software patents have been a disaster:

• it is impossible to find sufficient experts to properly assess the claims
• patents have been used as anti-competitive lockdown devices by large corporations holding war chests of claims

Sarah Weir of Cancer Research Technology provided a perspective on the use of patents in other industries. In biotech fields patents have been a necessary tool to get inventions out into production, and it works well to get rapid publication working (papers are published immediately a patent is claimed, so other researchers do get the information quickly). In these fields the system has been honed so that it works in practice. Technology transfer offices in eg universities are well set up to help academic inventors.

In questions, we heard from an SME which owns a software patent. Their company is developing an invention which a university had patented, a claimed process. Unclear whether it really makes any difference to the success or not of the company; it helps give some credibility to the business. The legal protection is not worth much, as a big company could easily kill them.

The patent office man stressed that US patents do not apply in the UK; but European ones do. He also said that many of the European Parliament amendments are against other European law; James Heald says other lawyers disagree with that assessment. It was claimed that the UK law will not change at all; software patents are allowed now and will be allowed in the same way in future if the Commission gets its way. The Patent Office gets 30,000 applications a year, of which 8,000 are granted; they don’t imagine a huge increase as they say the rules won’t change. Others were doubtful, saying that the number of patents being filed was going up hugely; and that the searching process will be be very hard.

If the law is not really changing, why do Nokia care so much about the European happenings? Who knows.

In general the audience were highly sceptical of the patent process for software. Personally, I was not so sure this is really the watershed issue for open source software. The law has a tendency not to upset society much, so if open source fails, it will probably not be 100% the fault of patent law.

Panel on open source in education

After some wine and snacks courtesy of Blackwells, the day ended with a panel-based BOF about open source in education. Andy Robinson showed Report Lab, which is about dynamical creation of nicely-formatted PDF from resources on web servers. He is hoping that universities and such like will find this attractive for documents like prospectuses. Paul Browning pointed out that universities will be obliged to issue combined transcripts and personal records in the future, for which this sort of technology would be appropriate. The most important parts of Report Lab are open source (http://www.reportlab.org/rl_toolkit.html), so interested parties can build their own systems.

Steve Alexander described SchoolTalk (see above), which is designed to provide an open source suite for managing school records. It is in early stages and it is written in Python, so maybe not for the mass target today, but a definitely interesting initiative.

Paul Browning talked about open source software in use in Bristol, and I talked about OSS Watch. And that was probably enough for the day.