[okfn-labs] opening up what3words

Friedrich Lindenberg friedrich.lindenberg at okfn.org
Fri May 15 10:05:03 UTC 2015

Ok, great to hear that it's not a very big technical challenge. Would you
be able to whip up an algorithm that, given a word list of length N and
with a steady sorting, would transform a lat/lon into three of these words
and back? I like the idea of basing it off the mechanism Maidenhead seems
to use, but extending that with the word list.


- Friedrich

On Fri, May 15, 2015 at 1:15 AM, willi uebelherr <willi.uebelherr at gmail.com>

> Dear Friedrich,
> in the FAQ i have read:
> http://what3words.com/faq/#toggle-id-15
> Q: What about latitude & longitude (GPS coordinats)?
> A: We love latitude & longitude and have based what3words on it. We take
> those long strings of numbers & letters and convert them to three simple
> words.
> You need 16 digits, 2 characters (+/-/N/S/E/W), 2 decimal points, and a
> space/comma/new line to specify a location to an accuracy of 3 metres using
> GPS co-ordinates. That’s great for computers and devices when humans don’t
> get involved, but humans actually do get involved. what3words is a human
> interface for latitude & longitude.
> Q: Why did you use 3 words rather than 2 words or 4 words?
> A: Dividing the world into 3m x 3m squares (which is small enough to be
> useful for almost all practical purposes) gives you 57 trillion squares so…
> – With 1 word you can uniquely name 40,000 3m x 3m squares, so that’s
> every 3m x 3m square in a small village. But the world is bigger than this
> so it would be useless system.
> – With a sequence of 2 words you can uniquely name 1,600,000,000 3m x 3m
> squares, so that’s every 3m x 3m square in an area the size of Hawaii. But
> the world is bigger than this so it would be a useless system.
> – With a sequence of 3 words you can uniquely name 64,000,000,000,000 3m x
> 3m squares, so that’s every 3m x 3m square in the world (actually we have a
> few combinations spare).
> – With a sequence of 4 words, we could uniquely identify every 3m x 3m
> square in an area in 40,000 times the size of the world, but there would be
> no point. If mankind inhabits the sun anytime soon, we’ll be all over it,
> literally, with a minor branding adjustment.
> ---------------
> With that, we see, its not a new algorithm. Its only a mapping for a
> linguistic projection system. It is helpful, of course. And based on the
> WC-numvers we can use it directly in any world map system.
> My question is different. The WC84 system is asymmetrically. On the
> equator, we have big distances based on degrees. On the poles, the
> distances goes to zero. We have triangles. Technicaly not a real problem.
> But not elegant.
> many greetings, willi
> Cordoba
> Am 14-May-15 um 18:47 schrieb willi uebelherr:
>> Dear Friedrich,
>> yes, the idea is good and we need the free algorithm for that.
>> Therefore, we have to make some inside looks. I am very interested for
>> that and for this cooperation.
>> For me, i am working for the real interNet, the interconnection of local
>> networks. It is based on local autonomous networks with the connection
>> to neighbor local network. The result is a meshnet on all continents.
>> The global part of the IP address is derived from the geographical
>> position from the local network. And this have always his singularity.
>> For that, we can use the WC84 (world coordinate system 1984) or any
>> other coordinate system like this.
>> With this method of creating IP addresses we don't need any "Internet
>> Governance" and all this stupid organisations like IANA, ICANN, ISOC or
>> IGF. Also not any NIC organisation. The people can organize it all self.
>> I have included some persons with english (not member on this list):
>> Quiliro Ordonez Baca, Software Libre y FSFLA, Ecuador
>> Diego Saravia, Software Libre, Argentina
>> Reanata Aquino, proyecto de investigación IGF en LA, Brasil
>> Jose Felix Arias Ynche, IGF, Peru
>> many greetings, willi
>> Cordoba, Argentina
>> Am 14-May-15 um 15:01 schrieb Friedrich Lindenberg:
>>> Hey all,
>>> during the OKFNlabs hangout I just mentioned what3words [1], a hip new
>>> startup that has developed a geographic grid system which turns the
>>> entire
>>> globe into 3x3m boxes that can be addressed through sets of three words
>>> (e.g. what.the.fuck).
>>> It's a nifty idea, and they're pitching it as a solution for developing
>>> countries where many settlements don't have well-defined addresses.
>>> Unfortunately, it's a closed platform: instead of publishing the
>>> mechanism
>>> so that it can be widely adopted, all conversion to and from w3w must
>>> happen via their API. While they are offering some NGO pricing, this
>>> leaves
>>> me concerned that development organisations might actually adopt this and
>>> produce datasets with copyrighted location names in it. Basically,
>>> it's Dun
>>> & Bradstreet D-U-N-S for places [2], and it will create a licensing
>>> disaster if left untreated.
>>> Since the idea is good, though, and there is not yet an "installed
>>> base", I
>>> would like to propose we pre-empt this disaster by creating an openly
>>> licensed clone of the mechanism. (I also hope this might exert enough
>>> pressure on them to reconsider their approach). Unfortunately, I don't
>>> know
>>> very much about GIS and geospatial reference systems, so this is a
>>> call for
>>> help:
>>> ## Who can help come up with a word-based geospatial grid system?
>>> A lot of the necessary components seem to be available in places: making
>>> word lists of 3-5 character words should be easy, and there's already
>>> some
>>> existing grid systems like MGRS [3] - and making an API to wrap this once
>>> an implementation of the core mechanism is there shouldn't be too hard,
>>> either [4].
>>> Who could help this? Post your ideas at
>>> https://github.com/pudo/open3words/issues :)
>>> Cheers,
>>> - Friedrich
>>> [1] http://what3words.com/
>>> [2] https://twitter.com/pudo/status/598884980182413312
>>> [3] https://en.wikipedia.org/wiki/Military_grid_reference_system
>>> [4] http://developer.what3words.com/api/
>>> _______________________________________________
>>> okfn-labs mailing list
>>> okfn-labs at lists.okfn.org
>>> https://lists.okfn.org/mailman/listinfo/okfn-labs
>>> Unsubscribe: https://lists.okfn.org/mailman/options/okfn-labs
>>>  _______________________________________________
> okfn-labs mailing list
> okfn-labs at lists.okfn.org
> https://lists.okfn.org/mailman/listinfo/okfn-labs
> Unsubscribe: https://lists.okfn.org/mailman/options/okfn-labs
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.okfn.org/pipermail/okfn-labs/attachments/20150515/e959fa82/attachment-0004.html>

More information about the okfn-labs mailing list