I don't see any versioning on this, and that seems like it is something that needs to be implemented ASAP if you actually want any of these addresses to start to be used in the wild.
If you make any changes to your word lists (in response to, say, all the comments here that say you should be omitting rude words), your old addresses won't be valid.
If someone gives me the address "Afraid Anus" (to pick an example someone found below), but I'm using a new version of your word list that omits "Anus," how am I supposed to know which version of the word lists I should be looking at?
It's all well and good to say that it will be forever backwards-compatible, but what if people find real flaws with the system?
(Another potential flaw: are homophones excluded? What if I'm giving the address over the phone?)
That's totally true (and I missed that), but it seems to me that the Country/State-specific look-up table is still the unversioned weak point.
What if state or national borders change?
Or how about disputed areas? Which table should I go to to look up words for Kashmir? And is the choice of table an (unintended) political statement?
Let's say I am creating an address in Kashmir. I have the integer coordinates 32,76, so I look up the number in the Indian table and I get 04.
I then hand my address to someone else, who believes Kashmir is Pakistani. He looks up the Pakistani table, which includes the table for Azad Kashmir, and looks up the coordinate for 04, which is 33,73. Oops.
True, "Azad Kashmir" is currently considered to be a distinct region, but it is certainly within the bounds of credible that Pakistan officially renames that area to "Kashmir" alone, in an attempt to bolster their claim on the region.
Then you'd have two distinct areas named Kashmir in your tables, but two different interpretations of the geographical bounds of those regions in the minds of the people in Pakistan and India.
The advantage of some of the other systems is that they are dependent on nothing other than latitude and longitude.
----
(Edit: to be clear, I think this is a cool project, and any edge-cases that may be problematic here are much tinier than my original, incorrect concern about word-list versioning.)
Yes, Xaddress creates 2 kinds of codes, the short one is inmutable and does not depend from geopolitical boundaries, you can check http://ubicate.me for the short code implementation.
If you make any changes to your word lists (in response to, say, all the comments here that say you should be omitting rude words), your old addresses won't be valid.
If someone gives me the address "Afraid Anus" (to pick an example someone found below), but I'm using a new version of your word list that omits "Anus," how am I supposed to know which version of the word lists I should be looking at?
It's all well and good to say that it will be forever backwards-compatible, but what if people find real flaws with the system?
(Another potential flaw: are homophones excluded? What if I'm giving the address over the phone?)