Geotagged at home (51.427051,-0.333344)

Push Pins, Dots, Customisation, Brands And Services; The Three Waves Of Making Digital Maps

The year 2013 has been a great year for maps and a greater year for maps in the United Kingdom, culminating in events that huddled together under the Maptember banner; OpenStreetMap's State Of The Map, the AGI's GeoCommunity and FOSS4G. But there was another event in 2013 that was map related and that was the 50th. anniversary of the British Cartographic Society's Cartographic Journal.

First published in 1964 and edited under the watchful eye of fellow map geek and cartography nerd Ken Field, the Cartographic Journal has been around for longer than I have. Just. This is something that makes me feel slightly less old than I usually do. In February of this year, Ken got in touch with me and asked me if I'd be willing to contribute an article to the 50th anniversary edition of the journal by writing something that attempts to answer the question what does cartography mean to you? Naturally I had to think long and hard about this and after some 30 seconds emailed Ken back saying I'd be privileged and delighted to. So I started writing. As is so often the way, what finally transpired and was published in May, bore little resemblance to my initial thoughts, but thanks to a permissive licensing approach on the part of the publishers, I'm able to reproduce the article below.



This is not the article that I set out to write. The working title for that article was going to be something along the lines of cartography is subjective; my favourite map probably isn't your favourite map. But every blog post, article or conference talk I write has to start somewhere and armed with this working title I set about trying to find my favourite map.

I was spoilt for choice as I had a vast array of sources to choose from. On Flickr there's the photostream of the Norman B. Leventhal Map Center at the Boston Public Library with over 3,000 maps to choose from. Also on Flickr are all the maps I've come across and and have favourited. Then there are my tagged social bookmarks on Delicious plus a plethora of mapping and cartography related articles that sit within my browser's bookmarks. That's a lot of maps and that's not even counting those that are offline in my reasonably large collection of map books.

Figure 1

I soon found out that reviewing all of these maps was by no means as simple a task as I'd first thought. I'd expected a favourite to leap out of the browser's window or book's page at me. Maps were certainly getting my attention but for the wrong reasons. I was able to discount maps I didn't like or maps I was ambivalent about, but even getting the beginnings of a short list of favourites was proving a thankless task. There were just too many good maps. Time was passing and I wasn't anywhere near finding a favourite map, let alone writing an article about it.

But as I continued browsing maps I noticed there was something else on my laptop's screen that was vying for my attention. Actually there were two other things. One was a terminal window that was open and logged into a remote server somewhere on the Internet where I keep a large stash of geographic and mapping data. The other was the icon for the TextMate text editor, sitting in the dock of my MacBook Pro, which I use for coding things, usually maps related code. Realisation slowly dawned on me that my favourite map was yet to come into being; it was the next one that I would make, and the one after that and so on. What I would use to make my next map is what this article should be about.

Despite working with geographical data for almost 30 years, it was not until 2007 that I made the first map for myself. Now I'm not a cartographer or a professional map-maker; if I had to describe myself as anything I'd term myself as a hybrid of a geotechnologist, in the literal sense of someone who works with geography and technology, and a neogeographer.

The combination of geotechnology, of neogeography and more traditional cartographical disciplines has given us repeated phases of acceleration and disruption. In my opinion the uses of digital maps are in the tail end of the second wave of innovation and we're starting to see the beginnings of a third wave. Each wave overlaps, there’s no clear border or delineation between them, but each wave has distinct characteristics.

When talking about making and using digital maps, the focus is on the way in which consumers, developers and the web use resources to create map based experiences, rather than the process of gathering the spatial data that underpins a digital map.

The first wave of digital maps was the combination of a various factors slowly coming together. Critical Mass; The First Wave Precursor

If you look back to before 2005, digital maps were complex and costly to produce. If you worked in this field, you probably had specialist knowledge on how to manipulate geographic and spatial data sets. Part of the cost of early digital maps was the length of time it took to produce them. Large amounts of CPU cycles were needed to convert the map data vectors into the raster images to the Web needed. Another limiting factor was the cost of the disk storage that mapping data sets demanded. In 2000 a Gigabyte of disk space would cost you around $15, compared to the $9000 10 years earlier, but the days of cheap and affordable storage in multiples of Terabytes hadn't yet arrived.

It was as difficult to use early digital maps, as it was to produce them. Availability and adoption of always-on broadband, was yet to arrive. The dominant form of Internet and web access was via dial-up modems, which were pitifully slow by today's standards. The early digital map services were characterised by small maps to save on download times. These services were also functional maps with an emphasis on making conventional paper maps available online. There was little time or motivation in making interesting maps that were easy to use. The First Wave; Mash Ups, Push Pins and Brands

The combination of fast CPUs, more bandwidth and cheaper larger storage came together in 2005 and the first wave of making digital maps started. It was in 2005 that Google launched their Maps API, followed in quick succession by similar offerings from Yahoo and from Microsoft. This was the wave of the web map mashup and it then that I made my first map. It was nothing special and nothing now remains of it, not even a screen shot. It was made using the Yahoo! Maps API and put push pins onto the Yahoo! Maps canvas. But it was the first map I made and I was proud of it at the time.

The first wave was also a branding wave. Whenever you made a web mashup you weren't only creating a map customisation that worked with your data and for your intent. You were also helping the companies that produced web map APIs by giving them free advertising on your web site with their brand. The old adage about no such thing as a free lunch was very much in evidence and looking at the maps you find on today's web, it's still the case. With a few exceptions, a large web mapping corporate organisation powers almost every store finder or local product or service finder and their brand sits, sometimes uneasily, with the brand behind the web site.

This trade off between ease of use and availability of web maps and their branding and styling of was the stimulus behind the second wave of digital maps. The Second Wave; Customisation And The Absence Of Brand

The second wave was the wave of customisable maps as well as of 'open' maps. The growth of the crowd sourced OpenStreetMap and the relatively open availability of being able to do interesting things with this data started to produce maps in all shapes, forms and most importantly, styles. OpenStreetMap itself was responsible for some of these styles, but companies such as CloudMade, founded in 2008 and San Francisco's Stamen Design, founded in 2001, started to make digital map tiles that were not only wildly different to the stock corporate theme of the original set of mapping APIs but in some cases were works of art in their own right as well as being maps.

Figure 2

Hand in hand with the proliferation of map styles came mapping APIs which were in marked contrast to the APIs of Google, Microsoft and Yahoo! The functionality available through all of the mapping APIs were roughly on a par. You created a basic HTML document and pointed an API at an element of that document which the API filled with a map on your behalf. The map is what we now term a slippy map; one that you can drag, pan and zoom around with a mouse or your finger. The map appears to slip around on the screen and is a viewport onto the larger one that appears to be hidden behind your web page. You can add controls to the map, you can add custom overlays onto it and in most cases you can even add other map tiles from a different source. But the starting point is always the originating company's branded map and even if you can change the tile source to customise the map, not that many people will choose to do so. This is in much the same way as the web browser that comes preinstalled onto your computer is the one that most people tend to use. Even if you provide a way to customise something, only a small percentage of people will generally take advantage of that facility.

The second wave has also been a wave of opposites; of the brand-less map and of the return of the vector map.

With open sourced APIs such as OpenLayers and more recently the Cloudmade originated Leaflet, there was no branded map to start with. You had to make a choice and select a digital map tile provider in order to make your map. In contrast to the brand-less maps APIs, toolkits such as Mapstraction allowed you to abstract away the specifics of an individual mapping API. With this approach you can preserve your investment by using a single API. You can also move to another provider by changing as little as a single line of JavaScript code.

Digital maps have almost always been vector maps at heart, but this has historically been a back end function in the map-making process. Rendering vectors in a browser or mobile app took processing power that just wasn’t available to the early digital maps, even if vector data is traditionally smaller and more compact than the bitmap images which power the traditional slippy map. But hand in hand with the use of vectors in mobile clients such as Google’s Maps app and Nokia’s HERE Maps app there’s also been a resurgence of the use of vectors in the browser.

Most of today’s modern browsers support both of the two competing implementations for vector graphics; the World Wide Web Consortium’s SVG and most recently, HTML5’s Canvas. As JavaScript APIs and toolkits spring up to take advantage of vector graphics, vector maps have started to appear. The New York Time’s Mike Bostock maintains D3.js and Adobe’s Dmitry Baranovskiy maintains Raphaël. Both of these are JavaScript libraries which allow you to visualise vector data and both of which support some form of maps.

Figure 3

Of course, in order to visualise vector data, you need vector data to work with in the first place and free and open sources of this are increasing rapidly from 2005’s GeoCommons (now owned by ESRI) to 2009’s Natural Earth in addition to proprietary vector data from the mapping corporates. The Third Wave; Maps As A Service

We're now at the tail end of the second wave and beginning to see the emergence of a third wave. This is the wave of maps as a service or MAAS, a specialisation of software as a service, commonly known by its SAAS acronym.

Maps as a service takes away the bother of having to write JavaScript code to exercise a mapping API and instead allows you to simply upload your raw data and the MAAS of your choice will do the heavy lifting for you. An early exemplar of MAAS was Stamen's Dotspotting, which whilst a simplistic 'dots on maps' implementation, allows a relatively sophisticated map visualisation to be produced with virtually no prior experience of working with geographical data sets or with maps. More recently CartoDB has taken the Dotspotting approach several steps further allowing extremely sophisticated mapping visualisations to be created, enabling these maps to be embedded in other web site and allowing access via an API. The latest entrant to making visualisation more accessible is Google Map Engine Lite, which builds on the heritage of Dotspotting and CartoDB.

Figure 4

Maps as a service means more than just simply making the process of creating digital maps easier and more accessible. It also incorporates the process of creating map tiles and vectors and making them accessible to everyone. From open source projects such as TileDrawer to the growth and success of MapBox, you can now control and manage the entire production of a digital map, from raw data, through mapping API to final user experience.

The digital maps waves commenced with an initial critical mass. Fast broadband Internet pipes offer ever-increasing bandwidth. The processing speed of CPUs continues to prove Moore’s Law. The ever-decreasing price of mass storage is now levels that were almost unthinkable a decade ago. Digital maps have proliferated across our desktop and laptop computers, our mobile phones and tablets, across almost anything that is connected to the Internet.

Despite industry commentators predicting the death of the map, the digital map still remains one of the best ways of visualising geographic, spatial and local information. Even if you were never taught to read a map, a map is inherently comforting and familiar and we automatically orientate ourselves to one. Looking Forwards; Repetition And The Unexpected

We’re now in the early stages of the third wave of making and sharing digital maps, be they bitmap based slippy maps, vector maps or 3D maps. Each wave has built on the success of the previous, usually accompanied by challenging the existing status quo. There is much irony here, when one considers that the origin of a wave is often said to be a disturbance in the surrounding medium.

Wherever the third wave takes the map and in whatever shape or form the inevitable fourth wave takes is unclear but the continuing development and innovation around the map is one of the key things that keeps making maps so compelling and such an interesting space to work in. As the return to vector maps shows, the waves of digital maps are a unique intersection between revisiting and learning from past technologies and innovation and disruption. To paraphrase George Bernard Shaw slightly history repeats itself and the unexpected always happens. This article originally appeared in Volume 50 of the Cartographic Journal in May 2013.

Gary Gale

I'm Gary ... a Husband, Father, CTO at Kamma, geotechnologist, map geek, coffee addict, Sci-fi fan, UNIX and Mac user