[messaging] "Pseudoword" base32 fingerprints

[Brian Warner]

On 2/5/14 5:35 PM, Trevor Perrin wrote:

> But it would be interesting to see what a better wordlist looks like.

Incidentally (and diving into the geek deep end), you might look at the
list of "gismu"[1] created for the synthetic language Lojban. It was
created (back in the 70s, with only minimal computer assistance) to meet
the criteria of maximal phonetic distance between words, and semantic
overlap with existing languages (weighted by number of speakers). So you
wind up with some words that actually sound kinda like words you already
know, but are still reliably distinguished from each other. All are 5
letters long.

 http://www.lojban.org/publications/wordlists/gismu.txt

(pipe into "| cut -c2-6|grep -v ' '" to get just the main words.. did I
mention the 70s already? :-)

There are 1342 words on the list (that text file includes some non-gismu
too), so you'd need one word per ten bits.

cheers,
 -Brian

[1]: pronounced "geese-mu", these are root words, that get combined to
     express more complex concepts