Whatever happened to the UK Government's Open Address Register programme?

This is the latest in an occasional series of blog posts about the campaign for open address data in the UK. Previous posts are listed at the bottom.

In this post I examine what little we know about the UK Government's short-lived Open Address Register programme, and question the lack of transparency around ownership of intellectual property rights in UK address data.

Background to the Open Address Register programme

The Open Address Register (OAR) programme was announced in the UK Government's March 2016 Budget:

The government will provide up to £5 million to develop options for an authoritative address register that is open and freely available – making wider use of more precise address data and ensuring it is frequently updated will unlock opportunities for innovation.

In a blog post, Paul Maltby – then Director of Data at the Government Digital Service (GDS) – situated the address register within GDS's wider programme to create "canonical lists of core reference data":

Registers, canonical lists of core reference data, are at the forefront of the government's effort. They enable a standardised way of storing and accessing data, independently of the technology platforms and digital services that use them. Government's expectation is that over time they will form the basis of government's data infrastructure, helping people put government data to work.

and indicated that GDS and Government would work with a range of other stakeholders to explore the benefits of open address data:

GDS Data Group and the Department for Business, Innovation and Skills (BIS) are working in conjunction with a range of other stakeholders to explore how to fully exploit the benefits of open and freely available address data.

We know that we have a lot of work to do, and a lot of people to engage, as we move forward. We are aware of the fundamental role played by local authorities in the creation and maintenance of addresses, and the important contributions of Geoplace, Ordnance Survey and Royal Mail. We look forward to continuing these conversations over the coming months.

But after that … tumbleweeds.

The March 2016 Budget was delivered by the second Cameron government. Theresa May replaced David Cameron as Prime Minister in July 2016, following the Brexit referendum. Cabinet Office minister Matt Hancock, who was mildly supportive of open data and had presumably signed off on the Open Address Register programme, was moved to a more junior post at DCMS.

Subsequent Conservative governments published no updates on the progress or outcomes of work on options for an open address register. GDS's discovery page for a UK address register disappeared in 2017. The wider registers programme faded away and was completely abandoned by early 2021.

(In October 2016, Cabinet Office minister Chris Skidmore promised to update the House of Commons once work on an open address register was concluded – but I can find no evidence that he did so.)

FOI responses, 2017 to 2022

We have a small amount of public information about what happened to the Open Address Register programme, from responses to Freedom of Information requests submitted between 2017 and 2019. (Peter Wells has written previously about some of those requests.)

The £5 million notionally allocated to the programme in the 2016 Budget was apparently not spent. According to two similar FOI responses from BEIS in March 2019 and June 2019, work on the "first phase" of the programme was funded from existing departmental budgets, and the total spent was only £517,000:

We can confirm that expenditure on the first phase of policy development on an Open Address Register (OAR) was not funded by new money but was met from existing Departmental budgets. The total money spent by the Department was £517,000. The following represents a breakdown of the expenditure incurred up to the point policy development on an OAR paused, following the EU Referendum and the General Election and prior to the transfer of addressing policy to Cabinet Office in 2017.

£6,567 Legal Counsel
£487,965 Ordnance Survey Ltd
£22,375 Crown Commercial Service (Cabinet Office Complex Transactions Team)

There was no expenditure in 2017/18 recorded against the OAR programme and the residual Departmental budget cover was therefore removed at year-end because of the formal transfer of geospatial policy. Ongoing addressing policy, within scope of the UK geospatial strategy, is funded by the Cabinet Office.

FOI requests for outputs from the OAR programme were refused by BEIS in 2017 and by both BEIS and Ordnance Survey in 2019.

BEIS's 2019 response attributes the demise of the OAR programme to the May Government's decision to set up the Geospatial Commission:

A cross-government team worked on developing options throughout 2016 however work on the policy development paused following the EU referendum and election.

Since the 2016 announcement, the 2017 Budget included a commitment to establish a new Geospatial Commission. The Commission is a committee within Cabinet Office that, on its launch, was given £80m of funding over two years to support the more effective use of public and private geospatial data. Its partners include government agencies with large amounts of location data, including the British Geological Survey, Coal Authority, HM Land Registry, Ordnance Survey, UK Hydrographic Office and Valuation Office Agency.

In August 2018, Cabinet Office launched a Call for Evidence on its first geospatial strategy, seeking views on how to make best use of the location data held by public, private and voluntary sector organisations. The Call for Evidence, which closed on 24 October 2018, is available online at https://www.gov.uk/government/news/government-launch-call-for-evidenceto-be-geospatial-world-leader. Addressing remains an area of interest for the Government. Cabinet Office anticipates that addressing will form a part of the UK Geospatial Strategy that the Government will be looking to publish in the next financial year.

Notwithstanding expectations given to BEIS by Cabinet Office – and much evidence of support for open address data in public responses to the Geospatial Commission's Call for Evidence – addressing did not form part of the UK Geospatial Strategy when it was published in 2020.

The Geospatial Commission confirmed in an FOI response in 2022 that development of the UK Geospatial Strategy did not involve any assessment related to UK address data.

2024: open addresses debated in the House of Lords

Earlier this year, several members of the UK House of Lords sponsored an amendment to the Data Protection and Digital Information Bill that would have required the Government to regularly publish a list of UK addresses as open data to an approved data standard.

The amendment and the Bill itself subsequently fell by the wayside, but the proposal was debated in the Lords in late March.

Viscount Camrose, who was then a minister for the Department for Science, Innovation and Technology (DSIT), set out the Government's opposition to the amendment, citing the ostensible outcome of the 2016 OAR programme:

The Government explored opening address data in 2016. At that time, it became clear that the Government would have to pay to make this data available openly or to recreate it. That was previously attempted, and the resulting dataset had, I am afraid, critical quality issues. As such, it was determined at that time that the changes would result in significant additional cost to taxpayers and represent low value for money, given the current widespread accessibility of the data. For the reasons I have set out, I hope that the noble Lords will withdraw their amendments.

Following the Lords debate, freelance journalist James O'Malley asked DSIT for a copy of the 2016 "study" referenced in Viscount Camrose's remarks. DSIT disclosed that output – an Ordnance Survey report on technical work to pilot an Open Address Register using "non Royal Mail IP" – in May, and O'Malley published it last month under cover of his own blog post. I've discussed the OS report further down in this post.

Also following the Lords debate, the peers who sponsored the amendment sent a letter to DSIT requesting a review of the Government's 2016 decision on an Open Address Register, on the basis that "the digital economy has changed significantly since 2016, and the economic case now strongly favours opening up." I have discussed that letter, and related material disclosed to me by the Department for Business and Trade, in my previous post.

New FOI responses from DSIT

In August, I sent two new FOI requests to DSIT.

One request asked for a copy of any reply DSIT had made to the Lords letter, along with any new material held by DSIT (received or produced in 2024) that related to the OAR programme and/or the case for making the Postcode Address File (PAF) or other national address data available to the public as open data.

The other request asked for some additional information underpinning the OS report, as well as any information (such as advice or analysis) related to the basis or extent of Royal Mail claims of ownership of intellectual property rights in data contained in Ordnance Survey's AddressBase.

You can download DSIT's responses to those requests, the second of which I received last week: FOI2024_00211.zip (6.7 MB zipped) and FOI2024_00304.zip (4.7 MB zipped).

DSIT letter: Viscount Camrose's reply to Lord Clement-Jones

DSIT has disclosed a letter dated 10 May 2024 in which Viscount Camrose replied to the letter that Tim Clement-Jones and other peers sent to him after the Lords debate.

Camrose's letter says the following about the Government's 2016 exploration of an Open Address Register:

The work that was undertaken in 2016 considered the creation of an Open Address Register, was a joint project undertaken by HM Treasury, The Department for Business, Energy and Industrial Strategy and Government Digital Service in 2016. The project consisted of four workstreams:

• Policy development
• Legal Advice
• Commercial considerations including negotiations with owners of relevant intellectual property owners in relation to potential inclusion within an Open Address Register
• Evaluation of options to build an Open Address Register that did not include privately owned Intellectual property

As I mentioned during the consideration of the amendment in the House, there was analysis undertaken in 2016 on the viability of creating an address database which did not rely on private sector data, including Royal Mail.

The approach that was taken achieved a greater level of completeness than any other previous attempt, but still resulted in 4.2 million (13%) addresses that are present within the National Address Gazetteer which could not be accurately captured through this methodology. This 4.2m addresses included blocks of flats and halls of residence where there are multiple addresses within a single building. This is a significant proportion of the circa 32m addresses of homes and businesses that we have across the UK and would impair the application of any open dataset within the digital economy as it would exclude significant sections of the population from engaging with them.

The legal and commercial aspects of the Open Address Register Programme considered options for alternative access routes to Royal Mail data. As the owners of the PAF Royal Mail government would be required to seek a licence from them in order to make the data available openly. The commercial activity identified indicative costs of such a licence would not have represented value for money for the taxpayer in the event that Royal Mail were willing to consider offering such a licence, which as a private company who owned the data in question they were under no obligation to do.

As a result, the conclusion of the 2016 work was that without the privately owned data the proportion of missed or inaccurate addresses created significant risks for the use of any OAR, and that a licence, should Royal Mail have been willing to grant one, did not represent value for money.

Other recent DSIT correspondence related to open address data

In its response to the first of my recent FOI requests, DSIT (which includes the Geospatial Commission) confirmed that no further assessment or evaluation of the case for an Open Address Register, and/or the case for making the Postcode Address File (PAF) or other national address data available to the public as open data, has been made since the work that was undertaken in 2016, "as the barriers identified through that work remain the same."

DSIT also maintained that, with the exception of Royal Mail, it had not received recent representations from any organisation in relation to making the PAF or other national address data available as open data.

DSIT also disclosed several chains of email correspondence related to formulation of the Government's response to the Lords amendment: between Geospatial Commission staff, with the Department for Business and Trade (DBT) – which leads on the Government's relationship with Royal Mail – and with Royal Mail itself. This correspondence is not very enlightening, as DSIT seems to have redacted any contributions that do not support the line that the Government took in the Lords debate.

Contributions from Chris Chambers and Jamie Clark, Deputy Directors at the Geospatial Commission, indicate the Government did have options for delivering the open addresses proposal in the Lords amendment. But none of those options were "appealing" and DSIT doesn't want us to know what they are:

There's also a second-hand report that somebody at DBT actually set out arguments in favour of open address data. A voice in the wilderness:

Ordnance Survey's Non Royal Mail IP Open Address Register: Pilot Final Report

The most significant output from the OAR programme that we have available is a technical paper, the Non Royal Mail IP Open Address Register: Pilot Final Report, produced by Ordnance Survey in September 2016 and disclosed to the freelance journalist James O'Malley in May this year.

This is one of the OAR outputs that BEIS and Ordnance Survey previously refused to release under FOI. Presumably, DSIT decided it was no longer tenable to withhold the report due to the passage of time and/or the fact that the Government had publicly relied on OS's 2016 findings to knock back the recent Lords proposal.

It is very difficult to assess this OS report, even on its own terms – because we don't know exactly what those terms were. Key information has been redacted from the report, including most of the assumptions under which the Pilot was undertaken:

According to the executive summary, Ordnance Survey was commissioned to produce a Pilot OAR dataset by prototyping a methodology outlined in a "Non RM-IP Address Register Solution paper" that had previously been submitted to the OAR Steering Board.

In its response to the second of my recent FOI requests, DSIT refused to disclose the methodology paper that underpinned the Pilot or the assumptions that have been redacted from the Pilot Final Report.

DSIT also refused to disclose internal and external legal advice that it holds related to the basis or extent of Royal Mail claims of ownership of intellectual property rights in data contained in Ordnance Survey's AddressBase.

Contrary to the analysis in James O'Malley's blog post, the OS report does not describe an attempt to recreate Royal Mail's Postcode Address File (PAF).

According to the report, the Pilot focussed on a "substantial sample of properties covering a range of address types in a variety of geographies" across Great Britain (rather than the UK). The starting point was an address candidate list from the National Address Gazetteer which contained the UPRN (a property identifier), USRN (a street reference), and X/Y coordinate.

We can see from this that the OAR Pilot specification is very different from the PAF specification. Although both datasets contain address fields, PAF does not include UPRNs, USRNs, or point coordinates. The Pilot also covered a wider range of addressable properties than PAF, which only contains postal delivery addresses.

The most surprising aspect of the OS report is the apparent decision to treat all address field data in the NAG and AddressBase – not just PAF data itself – as "Royal Mail IP" and therefore unavailable for purposes of compiling the OAR Pilot sample. AddressBase data was used only to assess the quality of the OAR Pilot data.

Apparently, as of 2016, GeoPlace (which manages the NAG on behalf of OS and local authorities) was unable to reliably distinguish Royal Mail IP from local authority IP in updates submitted to the NAG:

It is important to note that in order to be able to extract new Royal Mail IP free addresses from Authority Updates provided by Local Authorities, authorities would be required to identify these new addresses primarily through the inclusion of street naming and numbering cross-references. The flagging of these records would denote the source as being entirely local authority IPR rather than having any contamination from 3rd party data. To implement this attribution would be a significant acceleration of a set of non-mandatory business processes GeoPlace have been promoting to authorities over the last 10 years. Currently GeoPlace receives Street Naming and Numbering cross references from approximately 70% of authorities but it is unknown whether these represent a complete picture of all Street Naming and Numbering records from these authorities. The remaining 30% of authorities would need to implement significant business changes to their address data maintenance processes to fully link Street Naming and Numbering with local gazetteer maintenance. This carries some significant cost burden which local government would expect to be covered under this proposal.

Effectively, this meant that the OAR Pilot was based on a highly artificial assumption that all local authority address data is "owned" by Royal Mail.

The methodology in the OAR Pilot report is based on a mix of approaches to produce the sample, including use of automated geo-processing to append data from other OS geographic datasets to locations on the address candidate list from the NAG.

The report concludes with OS's recommendation that the OAR Pilot should not be taken forward as a potential solution, for various reasons:

This recommendation seems reasonable if we assume that the objective of the exercise was to develop a "canonical" open address register that could function as a lightweight replacement for AddressBase. But as we have no outputs from policy development, it is unclear whether OS or the Government carried out any analysis to assess the market potential for re-use of a "96% complete" open address product, based on use cases that are not supported by AddressBase or PAF. It is also worth noting that although the report highlights problems with the accuracy of data in the OAR Pilot sample, AddressBase and PAF also have data quality issues.

It is a source of concern that, based on Viscount Camrose's remarks in the recent Lords debate, successive Governments seem to have accepted Ordnance Survey's OAR Pilot report as the final word on open addresses in the UK. Given that OS has a vested interest in maintaining AddressBase as a source of commercial revenue, it would have been more credible to outsource the development of the OAR Pilot to a disinterested party.

At the very least, the Government should have commissioned a review of OS's report and recommendations. The DSIT disclosures include an attachment that it describes as "the ODI review of Ordnance Survey Pilot" – but this turns out to be only a June 2016 proposal from the Open Data Institute to carry out review work. There is no evidence that Government took up ODI's offer.

Comments

As DSIT has refused to disclose its advice on the extent of Royal Mail's ownership of UK address data, we don't know whether the technical methodology in OS's OAR Pilot report reflected Royal Mail claims of ownership or followed from the Government's own understanding of the legal position.

It seems obvious that Royal Mail owns database rights in PAF as a whole, and has a solid claim to copyright in the assignment of postcodes to addresses. But individual addresses are factual information and not subject to copyright – even if the precise form of the address has been tweaked to comply with the PAF specification. It is not clear that address records created in a local authority gazetteer necessarily contain Royal Mail IP, even if the authority uses PAF to create the records.

This may be fact-sensitive, according to local processes. Possibly, the Government is conflating Royal Mail's IP rights with the additional control of public authority data given away to Royal Mail under contract in the PAF Public Sector Licence, which states that any new records created using PAF "may not be supplied or any access to it provided to any third party" except as permitted under the licence. This restriction operates as if Royal Mail is acquiring IP rights in the created data, but is not equivalent to the conditions under which IP rights normally arise.

Ideally, the extent of Royal Mail's ownership of IP rights in UK address data would be clarified in the courts. Unfortunately, as UK address data is effectively managed under a Government-backed monopoly, there is no third party with a strong interest in mounting a legal challenge to whatever theory of ownership the Government has adopted.

In my view, the OS report does not really demonstrate whether it is feasible to produce an authoritative list of UK address data as maintained by local authorities, without using intellectual property owned by Royal Mail – in line with the proposal in the recent Lords amendment. The feasibility of that proposal depends mainly on the development of policy options, including clarification of IP rights – not on technical development of an entirely new data product. It is possible that the 2016 OAR work covered some of that ground but, given the evident lack of wide-ranging consultation and Government's ongoing refusal to release any policy or legal analyses, we should not assume that the case for open addresses was considered in depth or in good faith.

Although the OAR programme was conceived and carried out by previous Conservative governments, recent FOI responses from DSIT and DBT returned under the new Labour government have been no more forthcoming. As in other policy areas, it is likely that departments responsible for data strategy will continue to follow their established agendas unless or until ministers (or their close advisors) start to take an interest.

Previous posts related to the campaign for open address data in the UK

FOI release: Royal Mail's view on a House of Lords proposal for open address data in the UK (15 August 2024)

Open addresses in the House of Lords – what happened? (28 March 2024)

Thoughts on campaigning for open address data in the UK (30 June 2023)

UK address data: a primer and bibliography (last updated 30 November 2022)