At 01:24 PM 9/4/2000 -0700, Ray Dillinger wrote: ...
An interesting point: There are ancient inscriptions in Wales that no one has been able to read in modern times. Deciphering an unknown langauge, not related to known languages, when it is written in an unknown script is a feat of linguistics that transcends mere cryptanalysis and has, so far, rarely or never been done.
And, as language, doubtless it has regular structure, patterns, grammar, and the flexibility of use that people in everyday lives need in speaking - and presumably they're not even encrypted.
"Poor Man's Crypto", possibly even better than digital crypto, may consist in creating an artificial language together, and then using it whenever you don't want to be eavesdropped on.
That sound like the Navajo codetalkers. I can see two easy problems with this. A secret shared is no secret. If even one person versed in the language were to side with the opposing front, all records written in that cypher would become open. A new language would have to have new words for practically everything. Any borrowed word would open the language up to analysis. If you didn't get around to inventing a word for digital recording. You had digital, but you forgot recording, then saying <digital> recording in a sentance, would give someone a clue to grammatical structure. Unfortunately, to get a sufficient vocabulary to be flexible, would require a larger population using the language. If the language is sufficiently difficult to learn, it might be useful as a code but it would be hard to extend the population who could use it. If I remember my history, which is not to say that I do, the Codetalkers method worked because there was a small population who knew the language already, none of them were acquired by the Japaneese, learning the language was difficult, (the missionary who suggested it had managed to learn it some, if memory serves), and the language had existed, and been used, enough to be sufficiently complex. Still not complex enough. They had to spell some things out, like placenames. If just two people contrived it, then what they might have to say to one another might be secure, but would be limited to topics they had discussed in detail before, or related topics. If a population of 1,000 spoke it with fluency, and had for several years, the language may be able to deal with just about any current concept or object, but the opposition would almost certainly have access to the language as well. This would seem to limit the language to making disparaging comments about the person ahead of you at the checkout stand, confident that she didn't know what you were saying about him or her. Or discussing the shoplifting of luxeries with your schoolmates, relatively confident that the store clerk wouldn't know what you were planning, or even that you might not be casually discussing last nights game. Both examples I've suspected I might have witnessed. Good luck, Sean