The Voynich Ninja

I was just thinking that it would require a great amount of creativity to convert any paragraph of Voynichese into something (anything) that makes sense using any repeatable method. This alone would be quite the feat, and I don't think it is possible.

For the method to be repeatable, it should be possible to express it as a set of rules for someone else to reach the same results. This means, and this is the most important condition, that the method may not include an "interpretative step" which basically turns it into a one-way cipher. 

All "solutions" I've seen so far are either unextendable (only keywords, no paragraphs) or unrepeatable (one-way cipher with too much freedom). Therefore I believe it is impossible to translate a paragraph in a viable way. At least using the many methods that have been revealed to us so far

Koen, this is a fair question and comment, but I will reply with a fair comment of my own:

I am sure you regard my various methods as having "interpretative steps" that you would consider "one-way ciphers" with too much freedom.

The problem is that most actual writing systems would also properly be considered one-way ciphers with too much freedom, which require interpretive steps on the part of the reader in order to understand them properly. For example, would you like to express this paragraph that I am writing right now with a set of rules to convert it into even a rough approximation of its phonetic form? 

Or how about the paragraph from the West Vlaams Wikipedia page that I posted in the Germanic thread -- would you be able to state an explicit, repeatable method that expresses written West Vlaams as a set of rules to convert a paragraph of it into something that makes sense in standard Dutch, or better yet, in a different language?

In my view, the process of reading and understanding written language, any written language, requires "interpretative steps" on the part of the reader, and what goes on in the reader's brain in order to make sense of the meaning of the written language cannot be boiled down into any explicit, repeatable set of rules. In short, actual writing systems of actual natural languages are indeed one-way ciphers, with a large degree of freedom.

Repeatable is the key word.

There have been several proposed translations of very long parts of the Voynich MS text, e.g. John Stojko or Leo Levitov. These cover many paragraphs, and are much, much longer than the outputs of e.g. J. Michael Hermann, or (here) Geoffrey.

However, they *all* share the property that they can be done only by the author of the proposed solution, and are not repeatable.

(12-04-2021, 01:06 PM)Koen G Wrote: You are not allowed to view links. Register or Login to view.This alone would be quite the feat, and I don't think it is possible.

Therefore I believe it is impossible to translate a paragraph in a viable way. At least using the many methods that have been revealed to us so far

I expect it to be perfectly possible to decipher Voynichese in a repeatable way, but I would suspect far from easy to find the precise procedure for so doing. It would probably be quick and straightforward when you know the rules of the procedure well. It would indeed be quite a feat to discover the method which is why it hasn't been done yet.

The methods used so far have often tended to be similar, but they don't constitute the only possible methods.

(12-04-2021, 01:06 PM)Koen G Wrote: You are not allowed to view links. Register or Login to view.I was just thinking that it would require a great amount of creativity to convert any paragraph of Voynichese into something (anything) that makes sense using any repeatable method. This alone would be quite the feat, and I don't think it is possible.

In my opinion, a repeatable method that generates something that makes sense would be THE solution. Of course this has not been found yet, it could be possible that it will be found in the future, who knows?

(12-04-2021, 01:06 PM)Koen G Wrote: You are not allowed to view links. Register or Login to view.For the method to be repeatable, it should be possible to express it as a set of rules for someone else to reach the same results. This means, and this is the most important condition, that the method may not include an "interpretative step" which basically turns it into a one-way cipher. 

Some levels of language rules are quite simple, I am thinking in particular of grammatical rules. E.g. the fact that subject and verb must agree in number. Even more basically, the fact that words belong to different part-of-speech categories and that the order in which words appear next to each other depends on grammar rules. Sadly, grammar is widely overlooked by would-be solvers: they tend to generate word-salads in their target language and jump from word-salads to perfectly grammatical English. But their target language has a grammar itself, though they choose to ignore it.

Also, Voynichese has a lot of structure, e.g. word morphology, repeating words, consecutive similar words, line-effects, paragraph effects, correlation between the last character of a word and the first character of the following word, differences between paragraph-text and labels, the different "dialects" in the different sections.... These phenomena must be explained, for instance they could be an effect of the structure of the underlying language or an effect of the writing system. These features are our best hope to understand more of Voynichese, but, again, they are largely ignored by would-be solvers.

I agree that repeatability is based on rules. Languages have rules and Voynichese has structures that could well be the result of some unidentified rule-systems. In my opinion, only translation attempts that are solidly grounded in both these sides of the problem have some hope to result in something that can be repeated.

The method to read the Voynich ms text would be repeatable in the strict sense that you mean if it were a precisely rule-based cipher based on a known standard written language, such as Latin. (Actually, it's hard to think of any other written language in early 15th century Europe with such a well-established standard form that it could be the basis of a recoverable repeatable strictly rule-based cipher. Any other written language before the invention of the printing press would have had too much and too broad variation in its underlying written forms to begin with. Maybe Italian would also have qualified, I'm not sure.)

For example, the recently deciphered final Zodiac cipher was an example of such a recoverable, repeatable process. That's because the underlying text was more or less standard English. 

But is the Voynich ms text even based on an underlying known standard form of an existing written language, which was then encrypted by some method? If this was not the case, then no strictly repeatable method can be expected to decipher it.

Brief comment on a cryptographic aspect of the question: Typically, cryptographic encryption techniques increase the apparent entropy (randomness) of the resulting cipher text, as compared with the underlying plain text message before encryption. But if the Voynich ms text was indeed the result of a strictly repeatable encryption process applied to a text in an underlying known standard language, why then did this process apparently decrease the entropy of the resulting cipher text rather than increase it?

Marco: I even forgot to mention the gramaticallity of the output. But I wonder if, as a challenge of sorts, if it would be possible to produce a long word-salad without using an interpretative step.

I fear that even this might be very difficult.

One issue with looking for rules is, I think, that there seems to be relatively little interest in understanding Voynichese in terms of its historical context. Looking for rules without a knowledge of broadly some of the kind of rules used at the time seems to limit oneself.

For one who believes that the Voynich is written in cipher understanding the cipher context of the time is crucial. Evidence clearly shows that the period from which the Voynich is carbon dated was a period of rapid development and innovation in the cipher techniques used in diplomacy, the area where the most advanced ciphers of the period are to be found and also where they were most commonly used. Comparing diplomatic ciphers used in 1404 with those used in 1445 the increase in complexity is profound. Whereas comparing the ciphers used in diplomacy between 1445 and 1495 the differences appear to be negligible. The cipher used by Giovanni Fontana is very very simple compared to diplomatic ciphers from 1424 in the State Archives of Florence, indicating that knowledge of these techniques were most likely confined to a small number of people in the chancelleries of the major Italy city states. This was a period of experimentation with different cipher techniques some which became obsolete and the standard techniques only really crystallised by the middle of the century. Unfortunately so much is lost from the historical record for this time that forming a more complete picture of the evolution of cipher techniques over this period is very hard.

(12-04-2021, 04:36 PM)Koen G Wrote: You are not allowed to view links. Register or Login to view.Marco: I even forgot to mention the gramaticallity of the output. But I wonder if, as a challenge of sorts, if it would be possible to produce a long word-salad without using an interpretative step.

I fear that even this might be very difficult.

It probably depends upon your taste. Do you fancy Hawaiian word-salad?

It also probably depends upon what you count as a word. For example, since I have been looking at Middle English recently, here is a list of various attested forms of the accusative form of the 2nd person plural pronoun: 
yow, ou, yowe, you, youe, yo, yoe, yogh, yaw, yew, yhu, yu, yw, yhow, ȝou, ȝow, ȝouȝ, ȝowȝ, ȝo, ȝowe, ȝu, ȝw, ȝue, ȝiow, ȝeu, ȝew, ȝewe, ȝaw, ȝhow, ȝhowe, ȝiu, ȝeow, ȝehw, ȝuw, gow, gu, giu, geu, geau, eou, eow, eo, eowe, eu, euwȝ, ou, ow, æu, owe, hou, heou, heu

And likewise for the nominative form of the 3rd person plural pronoun:
þei, þey, þeie, þeye, þeiȝ, þeyȝ, þeih, þeyh, þhei, þhey, þai, þay, þeȝ, þe, þee, þi, þie, þy, þye, ai, ay, yei, yey, yai, yay, þeȝȝ; þæȝe ; tai, tay, dei, dey, dai, day ; tei, tey, teȝȝ

With such a variety of word forms to choose from for every word, I might be able to produce a long Middle English word-salad without using any interpretative step. I have used an interpretative step in my own methods precisely because I do care about producing grammatical text that makes sense (which is a different matter than whether or not my efforts and results have succeeded or failed).

Please note: I do consider that the ideas about verbose cipher units discussed on your blog last year (e.g., EVA [ok] as a single unit) will be quite conducive to the process.

(12-04-2021, 03:44 PM)geoffreycaveney Wrote: You are not allowed to view links. Register or Login to view.why then did this process apparently decrease the entropy of the resulting cipher text rather than increase it?

I don't believe that there is any fundamental law that an encryption technique must increase entropy. It is perfectly possible for a cipher to decrease entropy.