The Voynich Ninja

(25-09-2020, 08:15 AM)Mark Knowles Wrote: You are not allowed to view links. Register or Login to view.What about a manuscript where the same label can be found on different pages with quite different images?

What about a manuscript where the repeated labels have very similar spellings from one to the other?

And what about a manuscript that has all the properties of labels combined that we have discussed?

How about giving an example of a late medieval document with plant illustrations that has meaningless text?

I am aware of two older studies into long-range text correlations.

Both are mentioned at this web page (which still requires a lot of work):
You are not allowed to view links. Register or Login to view.

There was the Letter Serial Correlation (LSC) analysis by McKay and Perakh. The latter was quite interested in the Voynich MS, and had found a sponsor for funding the digitisation of the MS well before the time that the Beinecke did this. However, this was never agreed with the library.

Anyway, the conclusion was that the MS text exhibits a typical curve for meaningful texts, that disappears as soon as one scrambles the words. However, since the underlying process is not completely understood, it was not real 'proof' of meaning in the MS.

The other was a 2001 paper by Gabriel Landini. He noted a peak in the spectral analysis that corresponds approximately with the average line length of the MS text. This seemed to be a manifestation of Currier's 'Line as a functional Unit' observation.

In the paper by Andreas Schinner of 2007, the anomalous random walk behaviour could tentatively be explained by 'something unexpected happening' at each line transition, i.e. another manifestation of Currier's "LAAFU", but this is very far from clear. It is also beyond my state of knowledge of such processes.

(24-09-2020, 12:15 AM)Voynichgibberish Wrote: You are not allowed to view links. Register or Login to view.I have run dozens of frequency analysis on several languages and compared it to the Voynich Manuscript ... I even compared it to classical and modern ciphers ...

Under which name did you publish your results, please?

@MarcoP

Thanks so much for doing those experiments.  Since your results are comparable to each other (internally consistent, unlike Schinner’s which found the VMS to be an outlier) yet not completely comparable to Schinner’s results as a whole - do you think variations in method of preparation between data sets could be the explanation for all of it?

I agree with you that the switch to fiction prose for English rather than sticking with the Bible feels like cherry picking.

Do you have an opinion of how the VMS being a transliteration rather than a direct expression might impact data like this?  Or do you think your results are consistent enough to show that this is not an issue, at least if you treat the input the way that you did?

Thanks for providing further thoughts you may have.  I also appreciate you pointing out where l may be misinterpreting - this is not simple stuff, you must admit!

(25-09-2020, 12:29 PM)ReneZ Wrote: You are not allowed to view links. Register or Login to view.In the paper by Andreas Schinner of 2007, the anomalous random walk behaviour could tentatively be explained by 'something unexpected happening' at each line transition, i.e. another manifestation of Currier's "LAAFU", but this is very far from clear. It is also beyond my state of knowledge of such processes.

To be a bit more specific about this, when using Eva, the average line length is (approx) 160,000 / 5700 = 28 characters. This point can be indicated in the graph from Schinner, as follows (blue lines):

[attachment=4800]

This shows that there is reasonable compatibility with known plain texts before this point, and gradual divergence happens after this point.

@ReneZ @MarcoP

Would there be a similar internal structure that could be paralleled for the Bible in English? I know there are discussions of its repetitiveness.  But why would that be special for English compared with translations of the same text in other languages?

(25-09-2020, 12:44 AM)-JKP- Wrote: You are not allowed to view links. Register or Login to view.I've posted examples of labels that have been distilled down to three characters.
I've posted examples of labels that have been broken into small pieces that cross across illustrations.
I've posted examples on my blog of hot/cold wet/dry distilled down to 2 characters.
And way back somewhere on the forum I posted a link to a manuscript that has all the good days bad days info.
I've also made numerous references to Llullian-style diagrams where there are cryptic-seeming short labels (they are mnemonics), but if you know what the diagram is about, you can interpret them.
I've also made references to Greek labels that consist of two characters throughout the book.
I've posted examples on my blog of cryptic-looking but very logical two- and three-character indexing systems.

Your comments about 2-character and 3-character systems remind me of another cipher, the so-called You are not allowed to view links. Register or Login to view. used by the German Army in World War I. While the complete encryption method with the columnar transposition step would be anachronistic to the 15th century, the simple first step of writing 25 letters in a 5x5 grid and converting each letter into its 2-letter "ADFGX coordinates" would have been well within the imagination of a creative 15th century author. "FAXDF ADDDG DGFFF AFAX AFAFX" doesn't look at first glance like it could represent meaningful natural language text, but it can and it did.

I have a general comment about the "gibberish/autocopying" theory of the Voynich ms text and the debate about it:

If we compare the decryption of the Voynich ms text to the proof of the Riemann Hypothesis in mathematics, then the "gibberish/autocopying" theory would be comparable to the claim that the Riemann Hypothesis is false. But the many failed attempts to prove the Riemann Hypothesis do not at all mean that the hypothesis is false, nor do they actually constitute any serious evidence that it is false. They simply show that it is very, very hard to prove. In mathematics it is understood that one needs just as rigorous evidence to prove a hypothesis is false as one needs to prove that it is true. The same standard should apply to the gibberish/autocopying theories.

(25-09-2020, 02:19 PM)ReneZ Wrote: You are not allowed to view links. Register or Login to view.To be a bit more specific about this, when using Eva, the average line length is (approx) 160,000 / 5700 = 28 characters. This point can be indicated in the graph from Schinner, as follows (blue lines):

Hi Rene,
if I understand correctly, the X axis corresponds to positions in the binary path, not in the original character sequence.  Since Schinner (like Schenkel) maps characters to 5 bits patterns, the position on the graph corresponding to 28 characters is 28*5=140. But, since the change in direction is so "slow", I don't think that this shift makes a great difference.

(25-09-2020, 02:34 PM)MichelleL11 Wrote: You are not allowed to view links. Register or Login to view.Would there be a similar internal structure that could be paralleled for the Bible in English? I know there are discussions of its repetitiveness.  But why would that be special for English compared with translations of the same text in other languages?

(25-09-2020, 02:15 PM)MichelleL11 Wrote: You are not allowed to view links. Register or Login to view.Since your results are comparable to each other (internally consistent, unlike Schinner’s which found the VMS to be an outlier) yet not completely comparable to Schinner’s results as a whole - do you think variations in method of preparation between data sets could be the explanation for all of it?

Hi Michelle,
as I said, I don't fully understand the implications of these measures. They seem to me to be very far from anything directly related to the structure of the text. It could also be that my code is bugged, so I wouldn't take it too seriously. If you check Schenkel's paper, you might find some more information about how different texts behave. For instance, in Fig.5 he shows plots of the actual random walks for the texts he discusses: he corrects the plots for the overall linear drift, so that the last position is made to have Y=0 like the first one.
These are the plots I get for the four texts in my plot.

[attachment=4801]

It could be that the larger Y-range in these plot is the "cause" of the greater alpha values, but I am not sure.

As you suggest, data preparation could also be relevant. For instance:

Kokol et al Wrote:Each character is then transformed into a six bit long binary representation according to a fixed code table. It has been shown by Schenkel (Schenkel 1993) that the selection of the code table does not influence the results as long as all possible codes (i.e. we have 64 different codes for the six bit representation – in our case we assigned 56 codes for the letters and the remaining codes for special symbols like period, comma, mathematical operators, etc) are used.

Schinner relaxed Kokol's requirement (emphasis  mine):

Schinner Wrote:As a first step it is necessary to encode the characters of the texts under investigation to bit sequences. It has been shown that the actual definition of this code table has negligible influence on the interesting quantities, as long as all (or at least almost all) possible bit patterns are used

Schinner used 5 bit codes (vs Kokol's 6 bits) with only 32 possible patterns. But EVA only uses about 22 characters, so I don't think that "almost all possible bit patterns" were used. This is also the case for the other texts, since Schinner only processes lower-case characters, with no punctuation.

But this could be irrelevant as well: I don't understand enough to say anything meaningful, I am sorry.

(25-09-2020, 02:15 PM)MichelleL11 Wrote: You are not allowed to view links. Register or Login to view.I agree with you that the switch to fiction prose for English rather than sticking with the Bible feels like cherry picking.

If "cherry picking" implies intentionally twisting data, I don't think this is the case. I believe it is just confirmation bias at work: Schinner's reading of Schenkel and Kokol's works happened through the lenses of the idea that Voynichese is gibberish. He was unable to see those parts that point in a different direction.

Thanks Marco, I didn't remember the part about the bits, but as you say, the factor 5 does not make a major difference.