Could do better…

This was the title of presentation given by Mary Spence, the outgoing president of the British Cartographic Society, at the Institute for Archaeologists annual conference last week. The presentation was given in the same session in which I was presenting, revolving around new technologies, particularly mobile spatial technologies, in archaeology. I was looking forward to the talk by Mary Spence as I had previously heard her on the radio talking about issues relating to making maps in today’s techno-centric world; she was, at that time, pushing for improvements in the quality of maps and the media had picked up on her criticisms of the kinds of maps based on eg Google Maps and Yahoo Maps which are often cartographically quite poor.

I must admit, the main reason I was interested in seeing her talk was because I share many of her concerns. There are indeed many maps out there which are bordering on unintelligible and the quality of many is not up to the exacting standards set down by professional cartographers. The lack of contextual features on a Google Map is indeed often problematic and the example Spence gave on a GeoCommons map (where there is an obvious rift in a terrain model of India used as the basemap) is simply very poor. It is easy to find examples of poor quality maps and numerous examples were shown, exhibiting a mixture of poor quality cartography (i.e. the design and layout of the map) and poor quality data (i.e. inaccurate or erroneous content); I see these as two separate but related issues, the effects of which are exaggerated by the ease with which anyone can now quickly and easily produce a map without necessarily knowing how best to do this or how to find out how to.

I was, however, rather disappointed in the presentation as whilst being critical, the bigger issue of access to good quality data from which to make maps was firmly ignored whilst some unfair criticisms of map production were levelled.

Breaking news: GIS specialists can’t make maps…

The idea Spence put forward that GIS practitioners in general are ignorant of cartographic convention and practise and are the ones producing poor quality maps is simply wrong in most cases. I also found it a bit of an odd thing to say as there are GIS specialists within the BCS. It is true that there are some people working as GIS specialists who are wholly unsuited to the job due to a lack of appropriate skills or the job having been advertised well below the going rate for a suitably skilled person: You pay peanuts, you get monkeys, as exemplified by some of the maps Spence showed as examples which did not even include basic map furniture such as legends (and which is a particular problem in the low-paid cultural heritage sector). Having said this, the vast majority of qualified GIS professionals are fully aware of the history of map-making, cartographic conventions and how to make legible maps which are fit for purpose. It is often the case, however, that it is not GIS professionals who are producing maps these days: in many workplaces, desktop GIS is provided to users who may have no formal GIS or cartographic background with only rudimentary training provided, whilst the GIS professionals role is restricted to systems development and data maintenance, it being seen as primarily an IT role similar to database or website maintenance. More broadly speaking, map production has become truly democratised often with a resultant drop in standards but to blame GIS professionals for this is unfair. GIS itself is an invaluable tool for producing the highest quality cartography and in the hands of skilled users provided with good quality data, does just that. There is definitely a case for more GIS/cartography training for users expected to produce maps but given the costs associated with training, many organisations simply cannot afford this.

Making maps

This leads on to and is inextricably linked to my other major concern with what was presented, or in this case, what was not: access to and quality of data and associated processes of map production and how these affect the quality of cartographic output. Traditionally, cartographers would redraw maps for particular purposes. Time would be spent graphically tweaking information to make it as legible as it could possibly be. Treating maps as a pure representation of data rather than as simply a drawing loosely based on but separate from data makes it possible to systematise the production of making maps. It is often simply not feasible to spend time redrawing individual features for each map on which they appear: This is more akin to the production of a piece of art than an acceptable way of producing functional maps in a modern business context.

In GIS-based map production, features are recorded as geometry, attributed appropriately and then displayed according to a symbology based on this attribution. Data can be re-used and recycled without the need for constant manual intervention (other than to keep the features up to date). This makes the cartographic output entirely dependent on the quality of the underlying data and where this is of poorer quality, the quality of the resultant map will suffer. There is nothing inherently wrong with this approach, it simply requires input data to be well structured and of the highest integrity, arguably a good thing.

Of course, treating the information as purely graphical and undertaking extensive redrawing to make the data fit for each individual application is one approach and this seemed to be what was being advocated: Indeed, Spence seemed to be suggesting that this approach is unproblematic, simply part of the process of cartography, it is after all what cartographers have always done. But this is to miss the point: This conflates the processes of data maintenance and map production. Why should people who want to create maps in the modern world using modern tools have to constantly revise and amend to make up for poor quality source data? Why can we not have access to good quality datasets produced and maintained by dedicated providers? Of course, there are always choices regarding what to include/exclude on maps and how features should be presented but the use of GIS makes the division between data and presentation explicit and to argue that successful map making must always include graphical fudging of datasets goes against the current trend of democratising map making through platforms such as GeoCommons and Google Maps which take very systematic approaches to geographic information and are dependent on accurate data.

To my mind, such systematic approaches to geographic information are a good way forward, supported by adequate training in how to use such systems. It is simply not possible to have hordes of dedicated cartographic specialists endlessly redrawing maps using the digital equivalent of pen and ink, keeping the activity as the reserve of these specialists (remember the typing pool anyone…?): the role of cartographic specialists therefore becomes one of system developer and trainer, one aspect being to develop the cartographic aspects of GIS-based map production but also to ensure that users have enough skills to use the systems effectively to produce exemplary cartographic output up to the exacting professional standards laid down by organisations such as the BCS. And given a large number of users are simply members of the public who want to use available tools to get information they want to convey to others out there, the training and information must be provided in a way that is accessible to all: If the BCS see the proliferation of mapping tools amongst non-professionals as reducing the quality of cartographic output, then the way to tackle this is surely to target those users in particular, to allow them to produce better maps, accepting the fact that these tools and platforms will be used to make maps by people who, in many cases, will not have had any formal cartographic training but who are interested in producing maps. This is not deskilling but democratisation and has been happening across the board in the world of IT as formerly specialist activities become accessible to more people; the same thing has happened with all kinds of activities from word-processing to web-design with people developing associated skills in all kinds of related areas from image processing to database management and networking. Of course, there is still a need for specialists, but this democratisation changes the role of the specialist from dedicated practitioner to facilitator/enabler and highly specialised practitioner.

As I have already stated, the use of GIS facilitates high quality cartographic output but only where the source data is well structured and of high quality. Yes, we could just keep painting over lower quality source data to hide inconsistencies and errors but access to high quality source data which does not require such intervention is surely the best way to proceed. During the questions at the end of the presentation, this question of access to good quality source data in the form of OS data was raised and my interpretation of what had been said during the talk was confirmed. Spence responded that there are free sources of information out there for those who cannot afford to use OS data (a two tier geoweb if you will based on ability to pay; not something we should be aiming towards), ignoring the fact that the quality of such datasets is often simply not up to scratch and the use of it as is often results in the kinds of problems criticised during the talk. Not a problem if your workflow is entirely manual and graphically based; you simply reinterpret the data, paint over the cracks and create a new pretty picture. It is a problem if your workflow is more systematised or you do not have the skills or tools to be undertaking this kind of work. How should members of the public with only Google Maps to work with undertake map production in this way? How should my users who use a largely automated process of map production using ArcGIS templates achieve this, and how do they necessarily know where any inaccuracies and problems with content lie? Surely, much better to ensure the map base used is complete and accurate to start with, eliminating the need for manually correcting it, and the best way to do this would be to improve access to high quality digital datasets. Otherwise, digital cartography is little more than a digital analogue of traditional cartography where each map must be hand drawn and the resultant map is solely the intellectual product of an individual rather than a standardised representation of geographic information; this is more art than an effective business workflow with product and efficiency of production in mind (I can just see the guffaws as my time estimates for producing maps leaps up tenfold due to the additional time needed as I artistically redraw source data to make it cartographically acceptable for each and every map).

So, Spence was criticising the quality of maps based on platforms such as Google Maps, but what is the alternative? We have good quality maps in the UK, provided by the Ordnance Survey, but their use is often precluded by the astronomical fees levied by the OS combined with draconian, complex and often unworkable license agreements. She suggested the use of free or open sources of map data such as the Peoples Map and Open Street Map but the use of such data in a GIS workflow can result in exactly the kinds of issues she was criticising: the integrity of the data is often lacking, resulting in poor quality cartographic output if inaccuracies and inconsistencies are not first identified and resolved.

Free our data!

So, we need to free our data. It is, to my mind, unacceptable that the Ordnance Survey are permitted to restrict access to our national map base in the way that they do (there is a good article on this topic here). Furthermore, their vociferous pursuance of anything which they perceive as in breach of their obstructive license agreements should be stopped: Three good examples of the OS acting unreasonably follow.

Firstly, anecdotal tales regarding local walking groups having received cease and desist notices having dared to publish online maps based on OS maps showing interesting walks.

Secondly, threatening local authorities with legal action over incredibly useful ‘what’s in my neighbourhood’ type maps is ridiculous; such maps used Google Map bases in order to avoid using OS mapping online, something that would cost more than the annual budget of an entire council in many cases, but it was the councils own datasets which the OS then claimed ownership of as they had been captured against OS map bases. This notion of derived data as the OS interpret it needs to be challenged.

Thirdly, the most farcical of the lot regards crime maps, much touted by the Home Secretary and which police forces rushed to produce to meet government targets. At which point, the OS threatened legal action for breach of copyright, the data again having been geocoded against OS products. How can such a ridiculous situation be allowed to happen? Surely, the role of the OS, as national mapping agency, should be to facilitate and enable other national bodies fulfil their responsibilities not obstruct them.

If the BCS are serious about improving the quality of maps being produced and used then one invaluable discussion they could throw their weight behind would be to improve access to high quality OS datasets for all, not just big business who can afford to pay the exorbitant license fees. To state, as Spence did, that even BCS members have to pay OS license fees, therefore it is acceptable that everyone should is a non argument: the idea that already overstretched resources in local authorities and the heritage sector should be used to fund the OS in this way is a disgrace. But it emerged why the BCS will not get involved in this debate: Spence indicated that the OS are partners in the BCS, hence the BCS cannot / will not get involved in the debate or be seen to criticise the OS. And hence my response to the BCS on this issue of improving quality of maps: Could do better…