nume: number name scheme

v0.9.2

For too long have we had to communicate base-58 or hexadecimal numbers! And the auction-based name systems and first-come-first-serve name systems are not ideal either. This repo contains a new class of encoding scheme: pronounceable (more-or-less) encodings.

The basic idea is the inversion of the usual approach to account names: force names upon the users, much like how they have gotten their own names.

The scheme implemented in only a relatively-few lines of python is a simple special case: two character sets of equal order. One set contains vowels and the other set contains consonants. Simply take a character from either set in an alternating fashion. With our character set selections, we have essentially base-48.

Example

>>> nume = to_nume(0x18b676bae1e0d99c03c63114fb116ef9d976c13e)
>>> print nume
Jàňèdűŗů Lãbôťĩqůwąqê Þŏķĩwįbec
>>> print hex(to_number(nume))
0x18b676bae1e0d99c03c63114fb116ef9d976c13e

Possible Use for Accounts:

Since numes can be split up into three segments (first, middle, last), an account could be displayed initially with only the first segment without any accent. The third segment (and then the second segment, if necessary) can be displayed as a disambiguator. Finally, all segments could be displayed with complete accents as the ultimate disambiguator.

Forms

>>> nume = Nume(0x18b676bae1e0d99c03c63114fb116ef9d976c13e)
>>> print nume.get_form(0)
Janeduru
>>> print nume.get_form(1)
Janeduru Pokiwibec
>>> print nume.get_form(2)
Janeduru Labotiquwaqe Pokiwibec
>>> print nume.get_form(3)
Jàňèdűŗů Lãbôťĩqůwąqê Þŏķĩwįbec

Minimally unique forms

Given a set of numes, it is possible to find forms such that every represented nume is unique. These are referred to as 'minimally unique forms,' or 'MUFs.' Refer to the code for a definition of test(...) to better understand the following example:

>>> mufs = test(2**15) # use 32768 random numbers (generate MUFs for 32K Ethereum addresses)
>>> for muf in mufs[-12:]: # collisions are resolved last, so we can peek at the end of the list
>>>   print muf
Ijotobom
Jiharulu
Pebokese
Galofoja
Asatohiz Itonihere
Asatohiz Ozadalata
Ewigirih Etoceloda
Ewigirih Odalafaza
Lesacate Rumonopog
Lesacate Jowaziruc
Edegutel Owodilore
Edegutel Alaholeta

(A much larger set is required to end up with some MUFs of a higher form.)

Other uses

This class of scheme could eventually make accounts more 'searchable.' A system could likely be implemented to result in an experience similar to how, in real life, you can tell someone your account name on a legacy social media platform, they search for it and show you the results,and you can point at the correct one.

TODOs:

• Python3 support
• Implement as node module (in a separate repo once finalized)

Non-normative notes:

• My earlier attempts at pronounceable encoding schemes envisioned 'asymmetrical' encoding, where the two character sets used were not isomorphic to eachother. While an interesting algorithm/formula for determining the number of elements per 'digit' was discovered, it was not that useful and no efficient conversion algorithm was discovered. Had such an algorithm been discovered, the next progression would have been dynamic character sets that would permit consecutive consonants and vowels if they were part of common sequences (such as 'tr', 'th', and so on). Despair took over and I implemented the special case on my flight back from Prague. — {;

Changelog

• v0.9.2 (Nov 17, 2018):
  • Update README.
  • Add accent descriptions to better support future experiments in improving 'searchability.'
  • Fix order of 'R' characters.
• v0.9.1-b (Nov 16, 2018):
  • Add MUFs.
• v0.9.1 (Nov 14, 2018):
  • Implement conversion of nume back to number.
• v0.9.0 (Nov 4, 2018):
  • Implement conversion of number to 'nume.'