Get lost: is Apple Maps on a road to nowhere?

Mapping and navigation is at the heart of how we use smartphones today. By extension, the Apple Maps app is at the heart of iOS 6. And so Apple’s decision to swap Google Maps for Apple Maps in its new operating system (and in the newly launched iPhone 5) was bound to attract some attention.

Misplaced locations, unrecognisable landmarks, and irrelevant search results are surely not the epithets Apple would have wanted to accompany its new mapping application’s release. As “a company that prides itself on not releasing any product until it is perfect” this must be a misstep?

Unfortunately, imperfection is unavoidable and inherent in any map, indeed in any geographic data.

Imperfection begins with the map data. In Apple’s case, the underlying map data are supplied by TomTom, the world-leading supplier of in-car navigation devices.

In light of criticism directed at Apple Maps, TomTom was quick to defend the accuracy of its data. But no mapping company, including TomTom, would claim its data were perfect.

And imperfections are only magnified when data is combined from multiple sources.

In addition to TomTom data, Apple Maps combines data from more than a dozen other suppliers – for instance, geographic data about points of interest is supplied by Yelp.

Imperfection has multiple facets. Map data may be topologically inaccurate (if the map says I can turn right at the next intersection, can I actually turn right?) or positionally inaccurate (do things appear at their correct geographic coordinates?).

Perhaps surprisingly, data that is topologically accurate need not be positionally accurate (nor vice versa).

Using a process called map matching, today’s navigation systems can reliably identify which road my vehicle is driving on and which intersection comes next (high topological accuracy).

This is even the case when the coordinate positions encoded in the underlying map data and generated by my GPS contain quite substantial errors (low positional accuracy).

Another facet of imperfection is the currency of map data. In a constantly changing world (new roads, new buildings, moving businesses, renamed stadiums) map data needs to be maintained.

For a data set with global coverage, such as Nokia’s, this can mean up to 2 million updates a day.

If data sets of different currency are combined, mapped differences show up immediately, such as text labels at locations where the underlying geometric data has no feature (yet).

Even if geographic data is current and topologically and positionally accurate, it is notoriously difficult to perform accurate searches on geographic place names (termed “toponyms”).

Many places share the same names, a feature known as homonymy – Ararat in Armenia and in Australia, London in England and in Canada, and numerous Springfields around the world, both real and fictional.

Even if unique, many place names may be ambiguous when placed in a query (“Melbourne Motors” is a company’s place, not the city; “Street Road” is a road not a street; “Battle” is not a battle).

So imperfections are unavoidable in our maps, our map data, and the procedures we rely on to organise and search that data. Merely making maps digital does not make them correct.

Back in 1998, the Observer gleefully reported on a German motorist who, ignoring road signs, drove his car into the Havel River because his in-car navigation incorrectly showed a bridge instead of a ferry connection.

It seems that a similar mishap befell Apple Maps when classifying a locality in Ireland called “Airfield” as an airport.

Even Google Maps, of course, contains errors, despite leading the field with the highest quality map data. As with other map producers, Google relies heavily on ordinary people to spot and report errors.

Recruiting legions of users in this way helps all the major map producers to achieve much higher levels of map accuracy than they could hope to reach if they had to do all their quality control in-house.

As a result, there is every reason to believe Apple Maps will close the accuracy gap on Google, and most likely will do so quite rapidly.

But we can be sure none of the competitors will ever offer perfect maps: the only certainty is uncertainty.

Matt Duckham receives funding from the ARC for research into mapping and navigation.

Stephan Winter no recibe salario, ni ejerce labores de consultoría, ni posee acciones, ni recibe financiación de ninguna compañía u organización que pueda obtener beneficio de este artículo, y ha declarado carecer de vínculos relevantes más allá del cargo académico citado.