Anton,
"Mathematically that's correct, but from what I've seen in practice, generally when people consider it from the angle of 'plaintext vs ciphertext' they don't think of maths, but rather of the underlying cultural case study....Culturally, a cipher is a technique used to conceal. In this view, natural language written in some script is not a 'cipher' text, and it has not been 'encrypted', for cryptography is secret writing, and there is no secret in plain words put on paper."
I get and agree with the distinction you're making here, and it's certainly the case that things like Benedek Lang's work on the types of content that appears in cipher texts address those kinds of cultural context issues. That's absolutely a useful kind of research question to pursue. However...
You've accidentally made my point for me when you say, "generally when people consider it from the angle of 'plaintext vs ciphertext' they don't think of maths," because part of the point I am making is that considering it from the angle of 'plaintext vs. ciphertext' does not justify in any way, shape, or form not thinking about the statistics of the text.
If someone thinks spaces in the mss. are word separators in an underlying text (a debatable proposition, but one a lot of work on the text makes), then just the word length distribution combined with the 1st and 2nd order entropy statistics massively constrains the space of viable hypotheses about the underlying text (if there is one). The problem is that you get people who make handwavey appeals to "natural language" thinking that somehow changes that fact or allows them to ignore the implications of the statistics. It doesn't. The characteristics of the text are what they are, and any hypothesis has to engage with that.
"Putting aside some most extravagant natural language theories, generally people are considering a forgotten language or (another case) an unknown script designed specifically for ethnic group having no their script of their own. Researchers pursuing or inclining to this strand may be linguists or people with some background in linguistics."
...or a script designed by a non-native speaker for their own use in recording a language whose relevant characteristics (phonology, tones, etc.) they don't fully understand or hear to record. Or a (possibly lossy or ambiguous) abbreviated text. I am completely onboard with considering those kinds of possibilities.
I want to make clear that I am not suggesting that linguistic approaches or methods have no utility here. For example, off the top of my head I can think of several cases where people have applied techniques from linguistics to look at the morphological structure of "words" in the VMS. That's certainly worth doing. What I am suggesting is that hypothesizing something like the text being "a forgotten language [in] an unknown script designed specifically for ethnic group having no their script of their own" is all to often used as an excuse to ignore, dismiss, or otherwise fail/refuse to engage with the actual, concrete statistical properties of the text. It isn't.
"On the contrary, when people try to find substitution cipher solutions, that's mostly because they are novices in the VMS, and its perceived 'simplicity' blinds them at first. A substitution cipher is the simplest cipher, so it's entertaining to seek for possible solutions, especially if they are limited to suggesting several keywords."
I didn't express myself clearly, so I'll try again. Consider the following two hypotheses:
1) The text is a phonetic recording by a non-native speaker using a constructed or non-Latin script (a "natural language" hypothesis), and
2) The text is a phonetic recording by a non-native speaker using the Latin script that has then been enciphered using a constructed cipher alphabet and a simple substitution cipher (a "ciphertext" hypothesis).
From a hypothesis testing point of view, those are not distinguishable cases (assuming the same non-native speaker, and given the ~mid-20ish number of common glyphs in most transcription schemes). If the statistics of the text agree with or rule out Hypothesis #2, they also agree with or rule out Hypothesis #1. It buys nothing to make handwavey appeals to "natural language" -- Hypothesis #1 is just a special case of Hypothesis #2. Applying cryptanalyitic methods is not somehow ignoring the possibility that it isn't a ciphertext, it's just recognizing that equivalence.
"Just a passing note (I have not read the blog post above discussed)."
Not a problem in the context of what we're talking about here, although I do want to say that Robert's posts on his Goodreads page (You are not allowed to view links. Register or Login to view.) regarding the papers from the Malta conference are thoughtful and well worth reading...
Apologies if this shows up with excessive unnecessary white space between paragraphs -- not sure why that's happening...
"Mathematically that's correct, but from what I've seen in practice, generally when people consider it from the angle of 'plaintext vs ciphertext' they don't think of maths, but rather of the underlying cultural case study....Culturally, a cipher is a technique used to conceal. In this view, natural language written in some script is not a 'cipher' text, and it has not been 'encrypted', for cryptography is secret writing, and there is no secret in plain words put on paper."
I get and agree with the distinction you're making here, and it's certainly the case that things like Benedek Lang's work on the types of content that appears in cipher texts address those kinds of cultural context issues. That's absolutely a useful kind of research question to pursue. However...
You've accidentally made my point for me when you say, "generally when people consider it from the angle of 'plaintext vs ciphertext' they don't think of maths," because part of the point I am making is that considering it from the angle of 'plaintext vs. ciphertext' does not justify in any way, shape, or form not thinking about the statistics of the text.
If someone thinks spaces in the mss. are word separators in an underlying text (a debatable proposition, but one a lot of work on the text makes), then just the word length distribution combined with the 1st and 2nd order entropy statistics massively constrains the space of viable hypotheses about the underlying text (if there is one). The problem is that you get people who make handwavey appeals to "natural language" thinking that somehow changes that fact or allows them to ignore the implications of the statistics. It doesn't. The characteristics of the text are what they are, and any hypothesis has to engage with that.
"Putting aside some most extravagant natural language theories, generally people are considering a forgotten language or (another case) an unknown script designed specifically for ethnic group having no their script of their own. Researchers pursuing or inclining to this strand may be linguists or people with some background in linguistics."
...or a script designed by a non-native speaker for their own use in recording a language whose relevant characteristics (phonology, tones, etc.) they don't fully understand or hear to record. Or a (possibly lossy or ambiguous) abbreviated text. I am completely onboard with considering those kinds of possibilities.
I want to make clear that I am not suggesting that linguistic approaches or methods have no utility here. For example, off the top of my head I can think of several cases where people have applied techniques from linguistics to look at the morphological structure of "words" in the VMS. That's certainly worth doing. What I am suggesting is that hypothesizing something like the text being "a forgotten language [in] an unknown script designed specifically for ethnic group having no their script of their own" is all to often used as an excuse to ignore, dismiss, or otherwise fail/refuse to engage with the actual, concrete statistical properties of the text. It isn't.
"On the contrary, when people try to find substitution cipher solutions, that's mostly because they are novices in the VMS, and its perceived 'simplicity' blinds them at first. A substitution cipher is the simplest cipher, so it's entertaining to seek for possible solutions, especially if they are limited to suggesting several keywords."
I didn't express myself clearly, so I'll try again. Consider the following two hypotheses:
1) The text is a phonetic recording by a non-native speaker using a constructed or non-Latin script (a "natural language" hypothesis), and
2) The text is a phonetic recording by a non-native speaker using the Latin script that has then been enciphered using a constructed cipher alphabet and a simple substitution cipher (a "ciphertext" hypothesis).
From a hypothesis testing point of view, those are not distinguishable cases (assuming the same non-native speaker, and given the ~mid-20ish number of common glyphs in most transcription schemes). If the statistics of the text agree with or rule out Hypothesis #2, they also agree with or rule out Hypothesis #1. It buys nothing to make handwavey appeals to "natural language" -- Hypothesis #1 is just a special case of Hypothesis #2. Applying cryptanalyitic methods is not somehow ignoring the possibility that it isn't a ciphertext, it's just recognizing that equivalence.
"Just a passing note (I have not read the blog post above discussed)."
Not a problem in the context of what we're talking about here, although I do want to say that Robert's posts on his Goodreads page (You are not allowed to view links. Register or Login to view.) regarding the papers from the Malta conference are thoughtful and well worth reading...
Apologies if this shows up with excessive unnecessary white space between paragraphs -- not sure why that's happening...