I have been a bit of a skeptic over the industry golden child that is What3Words (W3W). When it first came out, as a GIS nerd, I questioned how this was any better than an address, simple coordinates or “the pub under the weird dog sign in town”.
The argument that What3Words is better for getting you to a place faster is also relatively invalid. People would like you to think that saying “covers.pill.trades” is more safe and reliable than saying “the Strawberry Zebra”, but let me tell you a little secret — national mapping networks and Google invest a lot of money into recording localisms. I know this through experience, having spent a few months adding alternative names into Ordnance Surveys systems many decades ago. I have no doubt that if I phoned for an ambulance and said I was at the Strawberry Zebra, they would know EXACTLY where I was (and yes, it is a real object worth a streetview).
That said, I had started to grow a little fondness for it. Quirky little things like being able to find your tent at Glastonbury or finding the area of the beach that you were meeting your friends was nice, even though we could just have shared our coordinates normally.
Also Read: How a network-based solution could be the key to combatting GNSS outages
Testing for rescuers
On June 1, 2021, BBC News released a story from Mark Lewis, who works at Mountain Rescue England and Wales (MREW), which said that the use of the W3W app had been “testing” for rescue teams. He gave the BBC a database from the last 12 months which listed 45 locations across England and Wales that rescuers received from lost or injured walkers and climbers, which turned out to be incorrect.
Examples included:
jump.legend.warblers which was in Vietnam
duties.factory.person was located in China
dignitary.fake.view turned out to be in India
refuse.housework.housebound was in Australia
middle.plugged.nourished was in the US
demand.heave.surprise was actually in Canada
flesh.unzip.whirlwind was in Russia
The article also relates that many of the W3W addresses can be easily confused where there are similar addresses a letter different not too far away. An example of this being circle.goal.leader and circle.goals.leader, both within 2km of each other.
“This is not an algorithm issue. Something like 73% of What3Words addresses contain a word that can be changed just by adding or removing a letter,” says Security Consultant Andrew Tierney from Pen Test Partners.
When you consider the extent of words in the English dictionary, take out any offensive or misconstrued word, and you have around 150,000 words. You can re-use these words in combinations but cannot have them near each other, which means that you need to cover the Earth’s 13003 million hectares of land with an insufficient number of words. In this case, W3W has used plurals to further increase the number of potential words, though this also leads to issues, as identified by Tierney, when words sound too familiar in close proximity.
Also Read: Tropical cyclone risk assessment using GIS mapping
End for W3W?
The idea behind W3W is sound and is great for recreational use, though in a world where we share our location through mobile devices, have GPS on tap and have grown with addresses and street names all around us, it should not be used for emergency services — the risk is far too high.
It is true that you could provide an address that is very similar to one nearby, I live on “Newtown Road” and a couple of miles away there is a “Newton Road”, but the area and postcode are different, so hard to get wrong from an emergency service point of view. Furthermore, there isn’t a risk of mis-locating an address outside of the country.
Is this the end for W3W? No, with a tweak of the algorithm and a bit more country bias, I think that this could still be a powerful system as it has grown to be so popular. The main question will be how to address and overcome the trust that has been lost.
