Location, location, location: precision tracking with What3Words

In the early part of last year, a young woman in the Humberside area of the UK was abducted and taken to a locked room where she had no idea of her location. However, she could still dial the emergency services on her mobile phone. The operator was able to send her a text message containing a hyperlink, which enabled the phone to communicate with the worldwide Global Positioning System (GPS) to identify her location to within just a few feet and display it to her as a unique sequence of three words. The woman read that sequence out to the operator, who was then able to send a police car to the exact location.

That is because that emergency service now exploits a capability developed by a London-based company with the eponymous name What3Words. The crucial role it played was confirmed in an official statement by the Humberside Police Force: “A victim of sexual assault was being held hostage, not knowing where she was. A call-handler talked the victim through What3Words and the three-word address was passed to dispatchers, resulting in the recovery of the victim and the capture of the offender.”

The company has covered the whole surface of the Earth with squares measuring 3x3 metres and assigned each of those squares a unique identity that is represented by a sequence of just three words chosen from a vocabulary of 40,000 English words. By combining GPS signalling with a simple app on a device, such as a smartphone, a user can identify their own location and communicate it to someone else or identify another location anywhere else in the world with equal precision.

There are, in fact, 57 trillion such squares, though the other 35 languages that the system currently serves manage to get by with vocabularies of just 25,000 words since they only cover landmasses and territorial waters. They include Arabic, Chinese, Russian, Tamil, Thai and Turkish as well as all major western European languages.

The company’s chief marketing manager, Giles Rhys Jones, confirms what is involved: “We have a very simple algorithm that turns long, complex GPS coordinates comprising 18 digits into three-word addresses, or the other way around, and that is it.”

The underlying premise, says Rhys Jones, is that the world is still “badly addressed”. Even in developed countries, locations such as industrial estates may not have a street and numbering system, while elsewhere such a system may simply not exist at all. Nor will it exist in temporary areas of habitation such as refugee camps created as a result of war or natural disaster. However, the What3Words system can remedy all such deficiencies immediately.

The company was set up just six years ago when CEO Chris Sheldrick found that he could not ensure accurate delivery of equipment for pop groups he was booking to exactly the right unloading point at venues. Collaborating with Mohan Ganesalingam, now the company’s chief research officer, and a third co-founder Jack Waley-Cohen, now head of corporate development, the idea flourished and was turned into a reality.

The developmental task was not so much to do with writing software than with compiling an appropriate vocabulary from which to form the three-word identifiers. Basic rules included “no rude words”, but also “no homophones”. For instance, both ‘sail’ and ‘sale’ have been excluded. Rhys Jones indicates that every language requires the expertise of around 30 language consultants to compile the necessary word lists, which then become the raw data from which the system’s basic algorithm generates the three-word addresses.

The system is strictly “non-hierarchical” so that similar-sounding words are never used to identify locations in close proximity. It is therefore impossible for a user to be guided to a nearby but erroneous location by inputting a slightly incorrect word sequence. Hence ‘table.chair.damp’ is in Connecticut, US, and ‘table.chair.lamp’ is near Sydney, Australia. The system will show the specified location but suggest alternatives closer to the user’s current location. “If users get it wrong we want them to get it horribly wrong,” says Rhys Jones. The business model for the company involves it being paid a fee by other organisations that want to use it in services they provide. The app is free to download.

It is now included as a standard feature in the navigation system of all new Mercedes-Benz cars and the German car maker has become a shareholder in the company. The recent enhancement to accept voice input allows appropriate hands-free interaction with the system. That upgrade required the system to be trained through exposure to hundreds of thousands of recordings of different voices with different accents.

The system has also been adopted by the postal system of the country most commonly associated with the notion of address-free nomadism – Mongolia. Other applications range from hotel directories to UN-Asign, a crowd-sourcing app that allows individuals to collect and disseminate information about danger points such as flooding or damaged buildings in disaster situations.

Future development initiatives include adding new languages, increasing the number that accept voice input and integrating an optical character-recognition capability. The company also indicates that a substantial increase in the number of UK emergency services that use the system from the current 18 out of 44 is in the offing.

The system’s ability to guide rescuers to people in real danger is still a cardinal feature. When another woman in the Avon area of the UK drove a car containing herself and a young child off a road and into a ditch, she followed the same procedure as the kidnap victim. The official statement from Avon and Somerset Police confirms that “the victim was unable to describe where she was, so we used What3Words to get her to share her location and effectively deployed resources to the scene”.