Open source, open data, open options?

I’ve been looking at open source in government and e-participation in particular for a while now, and this seems a good time to try to get into shape my thoughts on some of the reasons why open source is still struggling to gain traction in government in the UK.

Many people (I guess including myself too) start from the assumption that open source is a good thing. At the same time it’s there’s an understanding it’s not obviously a good thing to the decision makers. If it was obvious, there would no need for evangelism as for instance at the recent OpenGovSummit.

Andy Williamson blogged in response to the summit on the ongoing difficulties in fitting open source into digital government. He noted how in the UK there may be attempts at cultural change towards open gov and open data in the Home Office (of all places), but also noted the lack of concrete progress.

In summary: open source (OS) has signally failed to make an impact on government services. There are some exceptions: Government CMS and blogging platforms often using OS platforms such as WordPress and Drupal, some use is made of OpenOffice/LibreOffice and of course infrastructure packages like Apache and Linux underpin some core services. At the same time, the Government Digital Service (GDS) has clarified the rules and requirements around procurement of open source software, so there is movement.

Background concepts

There are plenty of definitions of free/libre open source software (known as FLOSS to the afficionados) and I won’t repeat them here. But there are a couple of characteristics that might be useful to highlight.

First, and most importantly, at their best, open source projects are more than a simple posting of the source code to the web. A ‘proper’ OS application is a part of a project characterised by a community of developers, funders/sponsors and users: network effects mean that together, they are more than a simple sum of their components.

Generally, this is supported by a platform such as Sourceforge or Github which provide an implementation space (code repositories and project management) a discussion space for interaction with the user community (bug reports, ideas, roadmapping) and an archive. However, it is perfectly permissible for a company to host their own OS community.

Second observation: OS has been most successful in developing tools for and by geeks (MP3 players through to operating systems and web-servers) and for consultants (Drupal, WordPress). It has not proven so successful in the delivery of sustainable specialist e-government applications – things that are boring to geeks, and that people should be paid to develop.

Successful OS projects are often (generally even?) dominated by a corporate sponsor (eg the original MySql, OpenOffice or a strong, well known personality (Linux, Drupal) – or both; or if not, there is a risk of the code will not have a clear roadmap or a long term support model.

Third observation: civil society is (rightly) more interested in transparency and participation – which are properties of open data and open standards, and are well delivered through online services using a known API but closed code (eg: Google docs and Twitter)

What is holding things back?

I think that most readers would agree that is naïve to think that open code means that it is cost effective for anyone to take up development of anything but the simplest application.

The reality is that open source projects that will deliver for government’s specific requirements will be controlled by a single organisation, often a commercial commercial company – this model is called Single Vendor or Commercial Open Source. MySociety is an example from the area of eparticipation: their code is freely available, but I am fairly confident that serious code development is still controlled by the MySociety team, simply because they have the knowledge of how it works and what the users want.

In my experience, from the point of view of most eparticipation developers, the main issue has been the lack of a revenue model that supports ongoing open source development. MySociety seems to be an exception in that it has found a financially sustainable solution to OS development. Are there any others?

The fundamental issue is that there has been a lack of demand for open source from the customers – the central and local government departments who are needed to fund the work. Quite legitimately, their focus has been on the completeness, stability and value delivered of the solution they are paying for.

Government departments are more interested in buying a service or a packaged software application which can be tested. Software works as a package because it means they don’t have to invest in recruiting and retaining developers. This is why the idea of an inhouse team carrying out OS development is likely to be a non-starter, despite Tariq Rashid’s claim that “Some [departments] … are benefiting from creating open source to share, consuming open source other colleagues have created and becoming centres of excellence”.

Lining up the incentives is still very difficult. In the end, it can’t a question of good guidance, or evangelists.

This leaves OS as a cost saving (or value delivering) approach. This in turn means long discussions about TCO and the risks associated with security or the proliferation of OS licences.

I saw Andy tweeted that the “Cost of transitioning to open source is an exit cost not an entry cost. It should be factored into TCO of old system”.
…Nice try at confusing sunk cost and marginal/prospective costs!

I think most businesses will conclude that it is much better to wrap an application, open source or not, within a clearly costed service package accompanied by an SLA.

Conclusion and a plea for help

It’s not all negative. There are at least two things that can be learned from open source development practices:

• The well understood mechanism for stakeholder engagement and creating an ecosystem around an application. Probably a good idea in itself when developing e-participation application
• Openness of code is not necessarily a problem – so long as the it is possible to remain in control of the development path. (In fact, the openness can be a selling point)

In terms of transparency and participation in government, open data and open standards are probably more important. So long as there are applications and web services that allow data to be extracted and manipulated, most stakeholders are likely to conclude that is probably enough.

Andy said that: “We need evangelists in government and those who are outside and see the importance of the issue have to back them and support them, or they will fail” and I agree with him that this leaves the promotion of OS dependent on transient people and policies. A permanent acceptance of OS will need organisations to change so that it is possible to construct successful business cases for OS solutions. What that change is, is another question.

Now for the plea for help:

• Do you know of examples where a government agency has found a way to commission and maintain open source solutions? (I mean more than simply rolling out OpenOffice to its employees or implementing WordPress for its public website).
• What do you think is holding back the use of open source in government? Is it an issue?

Further reading

There has been a fair bit of research in this area, many of the projects funded by the EU for example COSPA, CALIBRE (presentation) and FLOSS, FLOSSPOLS and FLOSSWORLD.  names to look for on this subject include Rishab Ghosh and Eric von Hippel.

There was a significant conference in the USA on the subject in 2010 – the proceedings can be downloaded here (JITP 2010 Proceedings: Politics and Open Source).

I have blogged on open source before here and over at PEP-NET. I would also recommend Dirk Riehle’s blog for some thoughtful perspectives.

Update: Adding this line to highlight one previous blog post in particular: Open source projects are male dominated and elite driven, so not a great model for eparticipation generally.