Tobias Escher at the OII

Today I gave a presentation to the Advisory Panel on Public Sector Information (APPSI), the independent body that advises government and its departments on how to make best use of the information they produce. This public sector information (PSI) includes data on legislation, geographical maps, weather data, financial spending – all kinds of information the state does collect and produce.

The presentation was meant to give some overview of the Web 2.0 phenomenon and what implications it has for how government should make its information available. I argued that out of the two main distinctive features of Web 2.0 – collaboration (user-generated content, social web, etc.) and sharing (XML, APIs) – arise a number of challenges to the way governments provide their information:

1. visibility – People do not any longer turn to the government or indeed, any particular trusted information provider or brand. Instead, they use a search engine and in the UK that is usually Google.
   Data from our experiments in which we invite people to our computer lab to find pieces of government information on the Internet shows that 90% of them would go to Google to find the answer to a question. Only 1 or 2 out of 10 people would eventually end up on the government portal direct.gov.uk.
2. competition – What is more, due to this heavy reliance on search there is no guarantee that people will obtain their information from government. Instead, Web 2.0 technologies have been used for a number of non-governmental projects that mix and mash the information available on public sites to add real informational value and enable easier access. Some of these projects have been highly successful, such as UpMyStreet.com for local area information and TheyWorkForYou.com about parliamentary affairs.
   As a matter of fact, in our experiments we tend to find that some questions are answered without any government sites at all despite the fact that all the information we were asking for was essentially produced and provided by the state.
3. control – Now, whether or not you think this is a problem is a slightly different question but evidently there are issues around accuracy and control when public sector information is used and re-presented by third parties (be they social enterprises or businesses). Obtaining school league tables from the BBC website might seem less critical than for example travel advice.
4. license – Last but not least, the rapidly evolving technology has enabled a variety of applications current licensing models for public sector information need yet to find an answer to. A case in point is the recent controversy about the use of UCL CASA Virtual London model in Google Earth that was stopped by Ordance Survey because as Andy Hudson-Smith notes “The OS currently does not have the ability to license models for public usage and this is from a government-funded and approved agency.”

While I am not an expert in this arena, from my point of you there are some basic steps government should take in order to make PSI 2.0:

1. provide structured formats – this increases the value of the information both for the private sector as well as for volunteers who want to play with the data. I am not talking about some laboriously defined standard – anything that provides some unique identifiers and a basic structure will do
2. leverage the wisdom of the crowds – Following on from that I think government should not spend too much money on information visualization themselves. There is simply no way departments can come up with some of the more ingenious ideas of data representation and cross-referencing that the Web community has generated. Instead it makes sense to offer the data, let others make interesting stuff with it and if it works, incorporate it into the state site. There have been some interesting examples of that for the BBC (see BBC Backstage) and the UK parliament (who now gets its post code – MP lookup from UpMyStreet).
3. government certified data stream – one way to leverage the creativity of the Web community while at the same time providing some form of guarantee for the accuracy of the data is a kind of “government certificate” for sites that basically states the project is using the official government data stream and incorporates changes to the official stream in real-time.
4. sort out the licensing – Yes, this is difficult. Does not necessarily mean everything should be available free of charge.

Some of the things that came out of the subsequent discussion will be available via the APPSI meeting notes soon. For now, let me know your opinions on the question of how government should deal with PSI in a Web 2.0 world.