On September 19, 2012, Apple Inc. released a new app for its iOS 6 operating system that was meant to unleash the company from the shackles of an increasingly competitive relationship with Google.
The app in question was Apple Maps, an entirely new mobile mapping resource that displaced Google’s long-revered Google Maps app that had gained ubiquity across the iOS and Android platforms. With all the requisite pomp and circumstance, Apple hailed its new mapping capabilities as the most advanced to date, and promised to deliver a whole new way of viewing the world. And yet within a mere few hours, media outlets, users and businesses quickly discovered that Apple’s maps were fraught with spatial data errors, such as misplaced streets and points of interest, incorrect labels and abysmal driving directions. What could have gone wrong?
Let’s take a quick look at an overly simplified history of mobile information resources. For most people, the most basic mobile device is a watch for telling the time; this has actually been the case for a couple of centuries. Eventually, as technology advanced, we had more mobile tools that were job-specific, but most people didn’t carry things like slide rules or surveying transits on a daily basis. But along came small portable calculators that became all the rage, and then cell phones and now smartphones emerged as “must have” pieces of personal technology. And because many individuals are now walking around with a functioning GPS unit (not just for outdoor enthusiasts anymore), location has suddenly become a critical aspect of the fundamental knowledge set we all expect to have at any given time — which we also have through our smartphone, maintained automatically for us.
From a content standpoint, we are in a renaissance of all things spatial as the need for location-based information has become high demand. This is the driving force behind all these map apps on our phones and tablets. For the longest time, Google maintained dominance (and still does) because it was one of the first to actualize mapping technologies as a common need for every consumer. Apple built on this with its integration of Google Maps in the iPhone, which then tied a robust spatial data foundation with a slick mobile platform to which all kinds of locale-specific advertising, directions, deals, offers and sales pitches could be seamlessly attached for on-the-fly consumption. Along with this came new ways to have fun with your phone’s GPS capabilities, such as the Foursquare app that lets you virtually “check in” to locations and score points for more check-ins, as well as become “mayor” of a place to which you check in frequently.
Spatial data has a bright future on the mobile platform, so what could have happened with Apple’s Maps? Let’s first think about the importance of spatial data quality and our faith in such data. Long before the more recent advances in cartographic and mobile technology, maps had been ascribed throughout history with a high degree of objective accuracy by map consumers. However, all well-trained cartographers realize that maps and spatial data are in actuality the by-products of a long process of empirical data acquisition, both on the ground and remotely sensed, and this data passes through several stages of subjective human interpretations based on the technology available, specific purpose, the intended audience and the organizational and/or cultural context in which the whole process takes place. Despite this, most map viewers persistently believe that maps are an honest reflection of “ground truth” — that the map shows exactly the way things are.
The idea of maps being the result of a complex process of generalization from real-world observations is very relevant to spatial data. The traditional paper maps already contain a certain degree of inaccuracy, but many spatial data sets of streets and highways were initially captured from the paper maps. The task of digitization, whether performed manually or digitally, is tedious and rife with various types of spatial data errors. Fortunately, more and more spatial data is being obtained digitally and without the interim generalization steps — such as being derived purely from digital sources such as satellite imagery and on-the-ground GPS capture of street data — or a combination of both, such as Google’s sometimes controversial Street View cars that roam cities and photograph, well, everything (Figure 1). The pure digital approach can still introduce a variety of errors, based on the technology being used to capture the data. But for the GPS units, the problem really doesn’t lie in the technology and devices, but rather in the data being deployed on such devices.
From my own estimation as a cartographer, Apple grossly underestimated the monumental task of managing spatial data and delivering reliable spatial data quality. Regardless of how the data was captured, or from which data vendor it was purchased, there is a content integration task involved that requires an enormous amount of effort. Digital map producers such as Google and Microsoft have been at it for years (in fact, I was involved in the latter’s mapping efforts, which date all the way back to 1993), and these entrenched firms are still working to get it right.
Given the plethora of data sets in existence, how many do you think have been thoroughly checked against the real-world location for accuracy? This isn’t a new issue for spatial data, as it’s always been an issue with cartographers. During the Age of Exploration, California was infamously displayed for decades as an island on the basis of maps from one voyage, which sailed around the Baja peninsula and concluded that it must be an island. Years ago when I was working as a cartographer at Thomas Brothers Maps (now part of Rand McNally), I was responsible for updating several areas in the Seattle area street atlas. I did the best update possible with sources on hand, but when I moved to Seattle for graduate school, I field checked my own work and discovered many errors on the paper map.
Similarly, spatial data accuracy has a lot of aspects; it’s not just about the proper location of a street or other object, but also the proper attribution, making sure a name is correct or that an area is given the right state or country identity. Even worse, in areas of geopolitical dispute, if the data is incorrect, it could potentially exacerbate an already tense situation. Looking at the Apple Maps example (Figure 2), the Senkaku/Diaoyu Islands, which are hotly contested between China and Japan, look to be rubber-stamped onto the map with duplicate islands. Needless to say, this is not how the real islands appear.
And this raises a larger issue pertaining to maps and localization. Generally speaking, North America, Europe and parts of Asia are known for having very good spatial data and there are a few key companies which have dominated the street data market and work diligently to ever-improve their data. But as location-based technology becomes more widespread globally, interesting challenges lie ahead. Data capture in developing areas of the world is improving but still lags behind. Some companies like Google are finding ways to improve the situation through crowdsourcing (as in the Google Map Maker program) while others invest in new data capture and technologies for satellite image processing. Still, there will be local roadblocks in some cases. For example, China remains sensitive about its spatial data resources which can make it difficult for foreign data providers to verify location accuracy or attribution within China.
Companies will undoubtedly arise to compete with Apple, Google and others in the mobile mapping space. But such competitors need to fully understand the content implications; it’s not just about map geometry and delivering a new app to a mobile device, it’s about all the other content that is associated with the map. They might believe that localization of such apps is only limited to the user interface, but what about the place names themselves? What about local cartographic conventions when it comes to symbology? Yes, maps contain a cartographic language of their own that must sometimes be “translated” between locales in order to ensure user comprehension.
The issue of spatial data quality across locales is only going to increase as new and better ways are developed for incorporating spatial data in our lives, and as more and more people travel extensively.