Supplementary memorandum by Christopher
Roper, Landmark Information Group (OS 01(a))
At the very end of the evidence given by myself
and Michael Nicholson, we were asked about the National Land and
Property Gazetteer (NLPG). This is a subject on which we hold
very different views. This is understandable as Michael heads
the company, Intelligent Addressing that has taken on the task
of creating the NLPG, whereas I have believed and argued for three
years that the concept was fatally flawed.
The idea for an NLPG goes back at least to the
1980s, when Ordnance Survey was a long way from completing its
digital large-scale map of the United Kingdom, when HM Land Registry
(HMLR) was just thinking about a digital registry and when few
local authorities had substantial holdings of digital mapping.
At about the same time, Royal Mail was successfully promoting
the Postcode Address File (another gazetteer) as a useful set
of geographical identifiers.
The first big idea was to establish a standard
(BS7666) that would allow all these bodies to hold and exchange
spatial information in the same way. Since Local Authorities are
charged with naming streets, allocating numbers and passing new
address information to the Post Office, it seemed logical that
they should also produce local gazetteers that could be aggregated
into a national gazetteer.
The production of the standard was driven by
local government interest. Tony Black who had been seconded to
the Local Government Management Board from Ordnance Survey, was
a major influence in shepherding the BS7666 through many layers
of consultation, to adoption. The standard came in four parts.
Specification for a Street Gazetteer
Specification for a Land and Property
Specification for addresses (1994).
Specification of a data set for recording
public rights of way (1996).
We are interested here only in the first three.
So much work and energy, by so many different bodies, went into
agreeing the Standard that problems of actually implementing it
were left to one side. Indeed, this is a problem with any standard
that seeks to meet a number of potentially conflicting needs.
It is either based towards the need of one set of potential users,
or is so loose as to preclude consistent implementation across
The work of assembling a National Street Gazetteer
(NSG) that conformed to the standard did not begin until 1997
and no independent expert believes it is complete, accurate or
robust. The NSG is no better than the data that was supplied by
individual authorities. Worse still, legal tangles over intellectual
property and the powers of local authorities to publish such information
means that access to the NSG, which is held by Ordnance Survey
on behalf of the Local Government Information House (a limited
company controlled by the Improvement and Development Agency for
local government), is tightly restricted and it is therefore of
limited use to the community at large. It is also impossible to
apply any independent check on its quality.
Although data sets have been created that claim
BS7666 compliance, this merely refers to the format of file and
field structures. The standard refers to structure and not to
content; so it doesn't say anything about the quality of data
they contain, completeness, accuracy or consistency. >From
around the country, practitioners are experiencing serious difficulties
in producing Land and Property Gazetteers that both conform to
the standard and are fit for purpose.
Nevertheless, the concept of the BS7666-compliant
NLPG is absolutely central to the concept of the National Land
Information Service (NLIS), which was the subject of a competitive
tender in 1999 to supply "an integrated search system to
support the conveyancing process". The invitation to tender
said that "The NLPG, developed by the Improvement and Development
Agency (IDeA) will be made available as the means of access to
local and central government spatially related information".
According to a notice in the Official Journal of the European
Communities in September 1999, "The cornerstone of (NLIS)
will be a NLPG containing a unique identifier, for each property
or land parcel in the country." More than a year after the
launch of the NLIS service (February 2001), the NLPG is still
not being used for the purpose for which it was designed.
Until Dr Geoffrey Robinson arrived as Director
General of Ordnance Survey in October 1998, it had been assumed
that Ordnance Survey would produce the first cut NLPG on a similar
basis to the exercise that had produced the NSG, through a contract
between Ordnance Survey and LGIH. Indeed, Tony Black, a senior
director of Ordnance Survey, had committed the Agency to producing
the NLPG and was absolutely furious when Geoff Robinson questioned
both the specification and Ordnance Survey's ability to deliver
Instead, he distanced Ordnance Survey from this
process, and left the NLIS implementation team with a serious
problem. This was a serious embarrassment for government as a
bid for Invest to Save funds had already been approved for creating
the National Land Information Service, and of the £3 million
involved, OS had been allocated £500,000 to undertake the
work on the NLPG. Ordnance Survey never published its reasons
for withdrawing from the process, but early in 1999, it became
clear that the NLIS conveyancing support project would be taken
forward by a partnership between IDeA and HMLR, without OS.
The joint venture between LGIH and Property
Intelligence plc to produce the NLPG was announced on 16 July.
Intelligent Addressing is simply a subsidiary of Property Intelligence
set up for this specific project. There was no public procurement
process and the details of the agreement remain obscure, though
both parties insist that they are acting together for the public
good. First details of what this partnership would deliver were
contained in a set of Briefing Notes sent to Local Authorities
in the Autumn of 1999. At that time, we were promised a complete
NLPG within three years.
The first steps of the process outlined in the
Briefing Notes is to assist Local Authorities in sorting out their
address-based data into BS7666-compliant form, attaching a Unique
Property Reference Number to each record. This work will be undertaken
by Intelligent Addressing, "a team of experts gathered from
(Property Intelligence) and elsewhere specifically to co-ordinate
work on address management and property gazetteers and help to
create and maintain the NLPG."
Reading the Briefing Notes it is extremely hard
to understand what all the fuss is about. It all seems extremely
sensible and workable, though the promised output seems to be
more like a National Address Gazetteer than the NLPG. However,
there are two central questions, to which it is currently difficult
to get clear answers: Why did its creation create such serious
difficulties between Ordnance Survey and IDeA that they led Dr
Robinson to resign in October 1999; and (directly or indirectly)
to the dismissal of a senior director of Ordnance Survey, Tony
It is also worth noting that Ordnance Survey
was engaged in its own process of indexation. In the NLPG Briefing
document, this is described as OS embarking "on a process
which will result in the creation of a set of closed polygons
for each geographical object on it Land-Line series maps. It is
intended that these polygons will be cross-referenced to the NLPG.
As the NLPG is created, OS will take a feed of change intelligence
to assist in the process of keeping the map base up to date and
accurate". This was written prior to the launch of OS Mastermap.
Is this simply a battle to see who assigns Unique
Reference Numbers (URN) to addressable objects on the ground or
on the map? There could be important issues of control, intellectual
property rights and work flow wrapped up in the assignment of
URN. I have a rather simple view expressed in the following pair
Either: OS assigns a URN to features (spatial
objects) that appear on the OS map, and the Local Authority links
its gazetteer of named and addressable objects to these URNs;
Or: The Local Authority assigns a URN to each
of the named and addressable objects appearing in its gazetteer
and (through some as yet unexplained process) tells the world
where they are on the OS map (point or polygon co-ordinates).
Neither system is perfect, and both would require
close co-operation between the parties, but it seems to us that
this is not a technical issue, but a commonsense issue. The problem
with leaving the task to local authorities, each with its own
ideas about how to go about the task, what resources to apply
and what priority to assign it, is that we are most unlikely to
end up with a fit-for-purpose national gazetteer that is consistent,
current or complete.
If we want a single geo-referencing system,
but recognise that there will be many gazetteers, held by different
authorities for different purposes, then we want it tied to what
is probably the best large scale digital map base in the world,
Ordnance Survey's Master Map.
This implies that there should be one set of
URNs assigned, initially, by Ordnance Survey. As new buildings
appear or buildings are demolished or re-structured, Local Authorities
can collaborate with OS to flag URNs as changed or obsolete and
to allocate new URNs.
Despite frequent, optimistic assertions to contrary,
most informed outsiders believe the effort launched by Local Government
Information House (LGIH) and Intelligent Addressing (IA) almost
three years ago (July 1999) has not yielded a reliable or consistent
National Land and Property Gazetteer (NLPG).
To understand why I believe it cannot succeed,
one needs to understand the process that was going to be used
to build it. The NLPG depended on aggregating Local Land and Property
Gazetteers (LLPG) that were to be constructed and maintained by
those local authorities (District Councils and Unitary Authorities)
that have the responsibility for naming streets, allocating house
numbers and registering house names.
Each property was to be assigned a Unique Property
Reference Number (UPRN). The UPRN would be the key to maintaining
consistent records across government and to supporting the National
Land Information Service (NLIS).
So why hasn't it worked? The first reason is
that most local authorities have still not signed up to deliver
and maintain an LLPG to those charged with compiling the NLPG.
In the absence of local co-operation, the compilers have used
already available data from Ordnance Survey (Addresspoint), the
Valuation Office Agency and Consignia (the Postcode Address File).
However, since none of these is complete, there is no way of identifying
errors and omissions.
However, even if every local authority had agreed
to compile an LLPG, the resulting NLPG would have failed to meet
the requirements of its compilers. This is because, we are never
going to have a consistent address gazetteer unless we first have
a consistently maintained National Street Gazetteer. At present,
although Local Authorities are empowered to name streets, you
find inconsistent spellings in different parts of each council,
on different map bases maintained by Ordnance Survey and even
on the street name plates themselves.
Over the past two decades, some evangelists
have claimed that the problem of consistency in these matters
could be achieved through the application of a standard (BS7666).
All kinds of data (including the NLPG) claim BS7666 compliance,
but it is gradually dawning that this standard can be applied
in a variety of different ways, and says nothing about the consistency,
currency or completeness of the dataset. All it guarantees is
the structure of each record.
So what is to be done? For over a year now,
Ordnance Survey has been working with HM Land Registry, the compilers
of the NLPG (IA & LGIH) and Consignia, to rescue and deliver
a fit-for-purpose NLPG. They were effectively ordered to do so
by ministers. Their deliberations have been shielded from public
view, but we are constantly told to expect an announcement shortly.
For a variety of reasons, although all the participants
were committed to find a way forward, their deliberations have
failed to produce a positive outcome. The NLPG compilers are loath
to admit failure. They want the solution to build on their efforts.
Equally firmly, Consignia and Ordnance Survey are unwilling to
surrender control over the specification and maintenance of their
There is, however, a commonsense solution that
has been urged by outsiders ever since the NLPG project was first
mooted. This does involve throwing away the NLPG as it stands.
The only dataset that could be verifiably complete,
current and consistent is Ordnance Survey large-scale mapping.
This can be compared against what is on the ground at any point
in time. It never will be 100 per cent complete or current, and
it certainly isn't at present, but it is the definitive record
of the built environment. It is used by HM Land Registry to record
property boundaries and titles. The word definitive is essential
to what follows.
"Definitive: Having the function of finally
deciding; determinative, final." (Shorter Oxford English
In other words, in the absence of an accepted
correction from the appropriate naming authority, the Ordnance
Survey map should be regarded as the primary source of street
names. As it happens, Ordnance Survey hasn't taken this role particularly
seriously and streets may be spelled in one way in one OS data
base and in another way in another OS data base. Nevertheless,
OS remains the only logical repository at present for a definitive
National Street Gazetteer.
Once we have an agreed National Street Gazetteer
(NSG), then it is a comparatively easy task to link each and every
building in the country to a named (or numbered) segment of street
or road. Once again, Ordnance Survey is the only logical repository
of the definitive National Buildings Data Set (NBDS). Whether
Ordnance Survey continues as a Trading Fund, becomes a Government
Owned PLC, or is fully privatised, the national interest demands
that these two definitive datasets are created and maintained.
Furthermore, it is essential that no private
or public entity is allowed to extract a monopoly rent from this
data. The data needs effectively to be in the public domain, with
feedback mechanisms in place to maintain its currency, completeness
and correctness. Once this is done, many of the disputes and technical
arguments that have paralysed the production of an NLPG for almost
a decade will disappear.
Local Authorities will feed corrections into
the definitive data sets (NSG & NBDS); everyone else will
regard these data sets as definitive, and proceed accordingly.
Local Authorities can build LLPGs or not, BS7666 compliant or
not, as the elected representatives see fit; Consignia can continue
to use Postal Towns as a defining element of the postal address;
Ordnance Survey will stop trying to fit Addresspoint to the Postcode
Any address, held by local or central government,
by the private sector (eg utilities), or by Consignia, can be
matched to a unique building listed in the definitive data sets.
This will, over time, lead to a more consistent management of
address-based data across the public and the private sectors.
It provides a structure that can be implemented immediately (delivered
within 12 months) and steadily improved over time.
What are the obstacles to this happening?
1. There has to be an agreement that the
NLPG hasn't met and cannot meet its stated objectives in its present
form. Until there is high-level agreement on that point, resources
will continue to be devoted to shoring up the existing edifice,
hoping against hope that all will be well. Qualified independent
assessors need to assist in this process.
2. There has to be agreement that none of
the contending parties should be allowed to extract a monopoly
rent from this basic building block of a national geospatial infrastructure.
At least three public sector bodies (OS, IDeA, Consignia) currently
contend with one another for exclusive intellectual property rights
in this arena.
3. It has to be agreed that the definitive
national data sets have to be built top down, with the naming
authorities simply verifying and proposing corrections to the
data. Streets and buildings are constantly being added and deleted
as the built environment evolves. Better tools are needed to enable
those involved to keep pace with the changes.
The purpose of the national definitive data
sets is to provide an atomic set of entities that will allow the
unambiguous cross referencing of data across both the public and
the private sector. This is an urgent task that is now recognised
in government and in the industry. It is a source of deep frustration
that those most centrally concerned are marching down a road that
leads nowhere, and that their discussions are shrouded in secrecy.
It is a national scandal.
Supplementary memorandum by Christopher
Roper, Landmark Information Group (OS 01 (b))
The Second Phase of the Ordnance Survey's Quinquennial
Review has been asked to consider if Ordnance Survey should remain
a Trading Fund or become a Government Owned PLC. It is impossible
for anyone to answer this question until we establish exactly
what we (Parliament, Ministers and Taxpayers) expect from Ordnance
The recent enquiry into Ordnance Survey, by
the Urban Affairs Sub-committee of the House of Commons Select
Committee on Transport, Local Government and the Regions, highlighted
a number of concerns: an appropriate form of regulation for Ordnance
Survey; its role as an effective monopoly in some segments of
the market; and the prices it charged for its products.
My own belief, based on many years of working
with Ordnance Survey, is that there is little or no reason for
it to continue as a single organisation. Ordnance Survey currently
does a number of different things, some of which could be done
equally well by the private sector, some of which need to be done
in the national interest free from commercial pressures. These
different functions were not examined by the Quinquennial Review.
A. Collects of authoritative survey data
covering the whole country;
This really does need to be done once and once
only. It could be done by private sector contractors, but there
is an absolute need for a body that guarantees the quality of
that data in the national interest. It may not be sensible to
have the same body both collecting the data and assuring its quality.
Despite Ordnance Survey's reputation for quality, not all of its
output has been fit for purpose in recent years.
B. Provides large-scale digital map data;
Large-scale mapping, until recently Land-line
and in future Mastermap, has provided OS with its principal source
of revenue. It is widely used by engineers, architects and surveyors,
but the national institutions that absolutely require it are HM
Land Registry (HMLR) and Registers of Scotland (RoS). A good case
could be made for handing over the maintenance of this data to
these institutions and then providing the data at marginal cost
to all other users.
C. Publishes small/medium scale maps;
There is no good reason, other than custom and
practice, why this is done by the public sector. Certainly the
production of the maps could and should be contracted out to private
printers. The Director General argues that keeping an available
stock of all sheets for the emergency services is a reason for
OS continuing to produce these maps which are of high quality
and popular with consumers, but there are a wide variety of ways
in which supply could be guaranteed. The real point is that there
is no real connection with the production of these maps, a cartographic
and printing task, and the maintenance of the large-scale data.
D. Provides the government with advice on
There is a real need for this, but pressure
on Ordnance Survey to maximise profits from its captive customers
in the public sector has meant that its advice has not always
been objective. The suspicion that it was self-serving led directly
to the resignation of the last Director General. It seems extraordinary
that the Quinquennial Review didn't enquire into the circumstances
that led to Dr Geoffrey Robinson's abrupt resignation after just
one year in the job. Witness after witness at the Parliamentary
enquiry argued for a Geographical Commission.
E. Acts as Custodian of Definitive National
Ordnance Survey's role in this area should be
central to its mission, but it has never been clear. However,
Ordnance Survey is the natural custodian of a definitive National
Street Gazetteer; a National Buildings Dataset; and a set of administrative
boundaries. These are all vitally needed for the purposes of joined-up
government and would not cost vast amounts of money to maintain.
Their maintenance could be funded by the Office of National Statistics.
It would be quite possible to conceive of a
National Mapping Agency, with fewer than 100 employees, that performed
functions D & E above and provided quality assurance for function
A. It should be completely free of any commercial role and be
funded centrally. There is no good reason, except for sentiment
and tradition for keeping the publication of small scale maps
(function C) in the public sector. This should be open to competition
as it is in France where small scale maps published by the Institute
Geographique National compete with maps from Michelin and other
specialised publishers. It is important to maintain a central
archive or large-scale digital mapping (function B), but it would
make much more sense to shift control of this process into HMLR
and RoS, institutions that, unlike Ordnance Survey, actually use
the data as a central part of their work. This database maintenance
could be funded from the cash flow of property registration and
Everything else that Ordnance Survey currently
does could be privatised and open to competitive market forces.
There is a clear contradiction between Ordnance
Survey's role as a dominant commercial producer of geographical
information and its other roles as advisor to government and guarantor
of the quality and fitness for purpose of its information. If
the government wants to push the organisation to be more commercial,
it becomes doubly important to safeguard the national interest.
Some of the issues are quite arcane and technical, which probably
explains the failure of the Quinquennial Review (Phase 1) to probe
Ordnance Survey's role in any depth. The separation of roles described
above would achieve this.
Ordnance Survey's reason for being has been
dominated in recent years but its internal drive to survive as
a single institution: preserving a bloated headcount, resisting
customer demands for change; and focused on a historical mission
that was becoming irrelevant to the needs of government and society
in the late twentieth and early twenty first centuries. Simply
opening the door to market forces, without considering the needs
of government or society, is not helpful. Ordnance Survey, left
to its own devices, will not take these needs into account if
they conflict with what it identifies (rightly or wrongly) as
Identifying Ordnance Survey's core mission in
terms of the public interest, instead of equating it with the
production of particular products, would produce a slimmed down
institution, better qualified technically to assure quality and
act as the custodian of definitive national spatial datasets.
Ordnance Survey is the 800-pound gorilla in
the Geographical Information marketplace. This market has been
historically stunted by Ordnance Survey's desire to control access
to data in order to preserve its dominant role in its core markets.
If specialised information service providers are to flourish in
this arena, Ordnance Survey needs to be broken up. There is little
doubt that the use of large-scale digital mapping across government
and across the private sector at present is sub-optimal thanks
to constraints imposed by Ordnance Survey as it seeks to meet
Treasury-imposed revenue targets.
At the very least, the arguments for breaking
up Ordnance Survey, should be examined.