Ruby's Greek Thread

[tavie]

That's not what I'm saying.  I'm saying entropy is a symptom, along with line patterns, that tells us that whatever we're dealing with, it is not a simple substitution cypher.  And that this has consequences for our ability to exclude candidate languages.

By symptom, I mean that the entropy levels and the Voynichese line patterns are characteristics that we do not see in normal, meaningful text, including one that has been through a simple substitution cypher.  These symptoms tell us that if there is meaningful text here, a lot of things have been done to this plaintext to generate these effects.

Individually, these "things" may be simple or complex.  Together, they may constitute an extremely complex cypher system that was advanced for the era, as some here have suggested.  Or they may constitute a system that has  a simple design at its core with added bells and whistles, including some aspects that weren't intended for concealment purposes but which have the effect of it being more complex, e.g. parsing issues.   Or whatever.   We don't know.

My point is that whatever those things are, their complexity as a whole makes it hard to exclude a specific language.  Let's say someone points out that the relative frequencies of Voynichese initial and final characters don't come anywhere close to matching the relative proportions we see in the contemporary Greek.  That would exclude Greek if the system were a simple substitution cypher.  But since it's not, we can't exclude the possibility that the complexities of the system have altered the plaintext Greek words in a way that generates artificial, unGreeklike letter distributions.  Indeed, if we did exclude Greek on that basis, then I think we'd end up excluding all other candidate languages on similar bases.  

If it were that easy to exclude a language, we probably would have found the right language by now.   So while people could come up with characteristics of Greek that are inconsistent with your theory to give you a better sense of where it might need modification or areas for further investigation...I can't see how we could exclude Greek.  

Moreover, if I understood the Wikipedia article correctly, the notion of entropy is quite recent, much later than our manuscript and the large number of coded letters. How did decoders proceed without calculating entropy?

I was answering your question about excluding Greek; I didn't say that calculating entropy is necessary for decipherments.  It isn't.  But as a "symptom", it is useful, along with other Voynichese behaviours, in getting a sense of what could - or could not - have been done to a plaintext to generate these symptoms or effects.