Re: Hexlish Test :-)

Byrl Raze Buckbriar wrote:

On Thu, 23 Jan 2025 19:03:04 +0100
Stefan Claas <fgrsna.pynnf@vagrearg.eh> wrote:

My encoder/decoder works only from hexadecimal to hexlish
and back, which is simple, but can then be used to encode
binary data to hex first and then hexlish.

 
 
You can already do that by setting custom digraphs or doublet
digraphs for the words that would collide on reverse decoding.
 
For example:
 
BACK ==> PAC || PACK ==> PAC || both collide to PAC.
 
Resolution:
 
BACK ==> PPAC || PACK ==> PAC || BACK uses double P.
 
Or the converse:
 
BACK ==> PAC || PACK ==> PPAC || PACK uses double P.
 
Another example:
 
BAG ==> PAC || BACK ==> PAC || both collide to PAC.
 
Resolution:
 
BAG ==> PACC | BACK ==> PAC || BAG uses double C.
 
This may also be done with vowels where useful.
 
Or use a custom digraph of letter combos not used in English:
 
BACK ==> PJAC || PACK ==> JPAC || PJ & JP not whole words.
 
If separation of words is required for clarity, insert any letter that
does not belong with either of the twain letters it divides.
 
You can make your own rules as you see fit for your case.
 
See the link for notes about double digraphs and custom digraphs:
 
https://soc.octade.net/octade/p/1736830573.829713

Thanks for the information, much appreciated! I discussed this but
come to the conclusion that this is for me a major task, involving
a lot of work. Therefore I guess a reference implemenation, either
from you or someone else, who is a native English speaker should
handle this task. This seems to be a complex task, in order to work
properly, with the right English dictionary and mapping dictionary etc.
 

For creating ciphers, map each English letter to a pseudorandom,
secretly keyed set of trigraphs, then cycle through the trigraph per
letter before re-using it. This will not create unbreakable ciphers if
the trigraphs are re-used, but for short messages without re-use of
trigraphs it will be reasonably secure. Using trigraphs gives 157
graphs per English letter, so as long as no English letters occur more
than 157 times, the cipher should be a tough nut to crack.

I already thought about using the Diana Cryptosystem, or the German
'Dein Star' OTP Cryptosystem, with the Hexlish output. Well, ...

Regards
Stefan

--
Onion Courier Home Server Mon-Fri 15:00-21:00 UTC Sat-Sun 11:00-21:00 UTC
ohpmsq5ypuw5nagt2jidfyq72jvgw3fdvq37txhnm5rfbhwuosftzuyd.onion:8080 inbox
 age1yubikey1qv5z678j0apqhd4ng7p22g4da8vxy3q5uvthg6su76yj0y8v7wp5kvhstum