Three unique words can address any place in the world | Hacker News Reader

Three unique words can address any place in the world

what3words.com

2 pointsnefitty14 comments

Yes, three words. But combined with their proprietary database...

Postal addresses (agreed, not well organized in many countries) and gps coordinates are at least open. And gps coordinates can be shortened by using letters.

We would actually only need a universal algorithm that can translate any gps coordination into something readable and back. No database needed. No dependence on any company.

Let's say an earthquake just occurred in a town named "El Pueblo de Nuestra Señora la Reina de los Ángeles de Porciúncula". Emergency personnel can try to lug this big name around, people can try to input into map apps or they can check out the three word address for the town and pass that around much more easily. The example I've given is an exaggeration (https://en.wikipedia.org/wiki/Wikipedia:Unusual_place_names) but this use is obviated when a person is traveling in a foreign country with unusual names for streets, stores, etc. I think this is the most interesting everyday use for the technology.

Who says your smart phone is going to work in an area that just had a big earthquake? The power might be out, the cell towers could have fallen down, or everyone could be trying to use the system simultaneously to tell their families that they're still alive. (In some parts of the world, they may never have had cellular service at all! Or they may not be able to afford phones that can run this app.)

For example, it was nearly impossible to use a cellphone in downtown NYC in the aftermath of the 9/11 attack.

The backup plan would be to have a local copy of the database cached on everyone's phone, but that would take up a lot of space.

Also, who is going to translate the database into every language used in the world? Three English words would be as hard for a Chinese speaker to remember as a pair of geographical coordinates.

Does it work with a database of word triples to locations, or is it a list of words, and an algo to convert between word triples and locations?

I would think the latter would be much smaller to keep on one's phone.

the surface area of the earth divided by 9 square meters is around 57 trillion (source: wolfram|alpha ) , so for word triples, you would only need around 4000 words stored in the phone. But, that wouldn't have the redundancy (any of the 3 words wrong would get the wrong result), so if you have any two of the words be enough to encode the result (so, getting 1 word wrong would result in the area you mean, and maybe two other areas you didn't mean), then you need uh, about 8 million words?

I'm not sure english has that many words.

oh wait maybe its only on land.

That would uh, actually only help a little...

Ok I don't understand their claim that it is redundant, but I'm pretty sure they don't /need/ to make it use a central lookup table when you can just convert coords into lists of words.

IDK if they have their algo being open though. If they don't I guess I should make an open clone of it, because not having it be open is dumb. A standard to refer to a location should be open if it is to be useful.

The use cases are clear. Also for less dramatic situations :-)

But my problem is the reliance on a database, especially a proprietary one. Feels like solving one problem by creating another.

Therefore an algorithm for converting gps coordinates into a few common words would be ideal (and to convert back).

Oh but there is, and has been for a long time now:

https://en.wikipedia.org/wiki/Geohash

Yes, I am familiar with that kind of approach. However, something like 'geohash.org/eyckqv01zy2v' is not what I mean because it is similar to '38.7436214,-9.1952231': both can only be communicated by copy-and-paste.

You missed the word 'readable' in my post ;-)

Dang! I was working on something like this, but for data in general.

I was thinking of using it for things with I guess a large space, because I thought I needed like 5 words, not 3. Or, no, it looks like I was using a smaller word list. Yeah.

I was thinking of using it to, for one thing, refer to people in a unique way.

I also had some other ideas for uses.

As far as implementation went, I had a python 2.7 bytes <-> words converter, but I hadn't quite fixed a bug leading to extra 0s at the end.

I work for what3words. It is an algorithm, rather than a database and as such is less than 10mb. It also works offline with no cellular connection and uses GPS. All 57 trillion squares in all the languages we support are on your phone. We are currently in 9 languages including French, Spanish, Portuguese, Russian, Swhili and we have Arabic and 14 others on the way. Thanks for your interest.

After a quick glance at the technical docs, it seems there is no way to get 3 words <-> position conversion without hitting your API or use your API. Why don't you distribute the list of words along with the algorithm, so anyone can work with it ? Because that's the number one thing people want.

Some people who are using the system:

The British Museum - archeology - w3w.co/British-museum Pollinate Energy - solar lighting the slums - w3w.co/pollinate-energy In2care - mosquito traps in Africa - w3w.co/in2care Remind me where - navigation for the visually impaired - w3w.co/remind-me-where Cartero Amigo - deliveries in the favelas - w3w.co/cartero-amigo

Not again!

Why do you keep flogging this dead horse?

It's fundamentally flawed because there is no relationship between neighbouring / nearby addresses. Thus totally and utterly worthless as a location / navigation system.

Resubmitting it to HN every few months won't change that fact.

As written, overkill. One unique word can address any place in the world.

"Here"

Reply on news.ycombinator.com