Fwd: what3words support in OpenStreetMap

[Robert Sugar]

Forwarding this from a personal conversation with Tom, who suggested to bounce this to the list to see how the other devs feel about it. Original mail (please read this first) with Tom's response to follow:

cheers,
Robert

----- Original mail -----

> From: Robert Sugar
> Subject: what3words support in OpenStreetMap
> To: tom at compton.nu, john.firebaugh at gmail.com
> 
> Hi Tom/John,
> 
> I am Robert Sugar, a PhD student at the University of Cambridge.
> 
> what3words
> 
> I recently played around with the website http://what3words.com/ , a startup set out to map every 3m x 3m square on Earth with three unique words. For example the Eiffel Tower would be "oiled.vehicle.marriage" (http://w3w.cm/oiled.vehicle.marriage). The combination is easier to remember/text/phone than a lat/long, while being shorter and more precise than a full address. It could be especially useful in less populated areas of Africa or South America, where postcodes and addresses are only sporadically implemented. Some more details you can find in the NorthSouthGIS article: http://www.northsouthgis.co.nz/?p=4398
> 
> why implement?
> 
> 1. having an extra location provider would increase the mapping capabilities of OSM with quite minimal effort, especially for the mobile user.
> 2. OSM-based websites could benefit as well. A lot of interest has already been generated from emergency services, taxi companies, among others.
> 3. this is a feature that currently no major mapping website offers
> 
> 
> implementation details
> 
> as I have seen from the code the implementation could be quite straightforward. One way to do this would be to add a new location provider similarly to how GB, US and Canadian postcodes are handled. The what3words API is quite simple to use (see http://what3words.com/api/reference for details), a simple POST request containing the "three.word.combination" or the "*OneWord" (which are even shorter custom location pins) would return the lat/long in JSON format. The API is currently free to use for small volume, and according to Chris Sheldrick (CEO, cc-d) there is a possibility to provide free access to OpenStreetMap for high-volume requests as well.
> 
> Q: what do you think is the most reasonable way to implement the feature?
> 
> cheers,
> Robert

--- Tom's response ----

> *what3words*
> *
> *
> I recently played around with the website http://what3words.com/ , a
> startup set out to map every 3m x 3m square on Earth with three unique
> words. For example the Eiffel Tower would be "oiled.vehicle.marriage"
> (http://w3w.cm/oiled.vehicle.marriage). The combination is easier to
> remember/text/phone than a lat/long, while being shorter and more
> precise than a full address. It could be especially useful in less
> populated areas of Africa or South America, where postcodes and
> addresses are only sporadically implemented. Some more details you can
> find in the NorthSouthGIS article: http://www.northsouthgis.co.nz/?p=4398

I'm afraid that for now at least I see this (which I actually came across the other day) as much the same as many other such schemes which have been proposed in recent years.

Given that none of those have taken off in any convincing way I see no reason to suppose that this effort will be any different.

> *why implement?*
> *
> *
> 1. having an extra location provider would increase the mapping
> capabilities of OSM with quite minimal effort, especially for the mobile
> user.
> 2. OSM-based websites could benefit as well. A lot of interest has
> already been generated from emergency services, taxi companies, among
> others.
> 3. this is a feature that currently no major mapping website offers
> *
> *

I see no significant benefit to us I'm afraid, until such time as this system has demonstrated that it will actually gain traction.

There is a cost to us however, in both maintenance terms and in terms of committing to supporting something which may not gain traction but which may be hard for us to remove. There are also political issues around the question of supporting what appears essentially to be a commercial operation which may not sit well with many of our users.

Feel free to discuss this on the openstreetmap-rails-dev list if you want and see if other people disagree with me, but for now this is not something I would be inclined to add.

Tom

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.openstreetmap.org/pipermail/rails-dev/attachments/20131112/64d0c624/attachment.html>