Nigeria has become the seventh country to adopt what3words for mail deliveries
what3words.com
what3words.com
Here's some good reporting on the company: https://www.theatlantic.com/technology/archive/2016/06/the-m...
"doors.aware.secure" gives you no indication of where it could be, especially when the neighbor is "hotel.leans.poetic"
If it's not meant for humans, then why not use GPS coordinates which are even more precise?
Not to mention, I'm sure there will be insensitive randomly generated strings like "muddy.ashes.pardon" https://map.what3words.com/muddy.ashes.pardon
It's a solution searching for a problem, and existing mail and building addressing schemes are far superior to it.
> Street addresses worldwide are inaccurate, unreliable and don’t exist at all in many places. Poor addressing is expensive and frustrating, hampers economic growth and development, restricts social mobility and affects lives.
> Street addresses can usually identify a building, but aren’t accurate enough to help a courier or taxi driver find the correct entrance. This results in delayed or failed deliveries, and numerous ‘where are you’ phone calls.
If you'd like to visit my office its at atomic.pipes.rods.
https://support.what3words.com/hc/en-us/articles/207769875-C...
Apart from that, it's basically "just" geohashes with a different (population-density based) encoding.
There's a pretty good, working parody at http://what3fucks.com
Geohash is patent free btw. I am sure something can be done here as well.
Because anyone can pick whatever 3m square portion of the building they want to, it seems it would make mail sorting harder...a bundle for one specific building.
1. No existing national addressing system exists.
2. Streets are not named (or do not exist).
3. The address space to be described is extremely sparse.
4. Mailing addresses are transient (i.e. buildings move, as tents, vehicles, etc).
A dense apartment building in an urban area is unlikely to be described by any of these constraints.
As mentioned elsewhere, these problems can also be solved by latitude/longitude coordinates. However, this simply provides a friendlier "hostname" convention to a numeric backend.
This company seems a far cry from ICANN, however.
I think it's kind of dumb that it requires a company to run, though. There are around 170,000 words in the oxford english dictionary. If you take any combination of three random words, you would have a grand total of 170,000 x 170,000 x 170,000 possible permutations, or 4,913,000,000,000,000 possible permutations.
If you only took the top 50,000 most commonly used words, then you would have 50,000 x 50,000 x 50,000 = 125,000,000,000,000 possible permutations. That's 125 trillion combinations.
Guess how much disk space a word list of 479,000 words takes up? Oh, only 4.5 megabytes. (see: https://github.com/dwyl/english-words)
It would be extremely easy to build an index of words, number them, then correlate them to GPS coordinates, all with a simple client-side algorithm.
There is literally no reason at all why this needs to be a single company in control of this. It could be a completely open source project, maintained by the community. And it wouldn't need servers, because all you would need is a small client-side app that would convert the numbers of your current GPS coordinates to/from the word list combos (which would only be a couple megabytes in size). This could be easily distributed as an android or ios app that would run completely offline.
Evaluation of Location Encoding Systems (https://github.com/google/open-location-code/wiki/Evaluation...)
Conversely, (as another HN user pointed out), OpenStreetMap's opinion of what3words is quite negative: "what3words is a commercial, non-open, patented location reference schema. Open data advocates (such as the OpenStreetMap community) would generally advise against adopting it at all."[1]
What I like about their approach is that you don't run the risk of your business having a silly sounding address.
I found the habitual part of Earth to be 24,642,757 square miles [0]. A square mile = 2589988.11 square meter [Google].
So that comes close to 7.1 trillion locations. It results in 7000 Gb for each byte stored.
This makes me wonder if there is some kind of algorithm is involved. The location names seem randomly but might not be. Maybe its only random for the superficial observer. Which could mean with a sufficient big enough sample the algorithm can be found. Maybe even a relative small dataset with locations close to each other could work already.
But they do map the entire earth, including oceans.
Or appeal.appear.ledge.
I hope that they get great success.