(05-05-2026, 10:56 AM)oshfdk Wrote: You are not allowed to view links. Register or Login to view. (05-05-2026, 10:07 AM)Labyrinthinesecurity Wrote: You are not allowed to view links. Register or Login to view.To me, the most intriguing finding is that these two signals are completely orthogonal. Within the Herbal section, the correlation between the cho/che ratio and the d/l ratio is essentially zero. Knowing the switch state of a folio tells you nothing about its d/l ratio, and vice versa. They are independent dimensions of the manuscript's structure, driven by separate mechanisms that happen to partially align in Currier's coarse classification.
I don't think I understand most of this, could you maybe explain the implications a bit more? If the "signals" do not map to A/B then are we just talking about some arbitrary split of folios into groups and then identifying some metric orthogonal to this division? What's the use of this? I think you mentioned some "structure" that this approach unveils, what is this structure?
The proposed structure to sustain the A/B split is a 3-layers construct: a template system, a switch modulating the use of these templates (per folio), and an independent d/l dimension.
(05-05-2026, 11:45 AM)Labyrinthinesecurity Wrote: You are not allowed to view links. Register or Login to view.The proposed structure to sustain the A/B split is a 3-layers construct: a template system, a switch modulating the use of these templates (per folio), and an independent d/l dimension.
Thank you for the answer, it didn't clear up things for me, but maybe this will help the others.
Are there any implications for potential deciphering attempts or Voynichese generators?
I have difficulty understanding this also. You need to remember that most people visiting this forum are unfamiliar with the technical, analytical and statistical jargon that you have used. For their benefit, please, could you state in more simple terms what your conclusion is.
(05-05-2026, 11:55 AM)oshfdk Wrote: You are not allowed to view links. Register or Login to view. (05-05-2026, 11:45 AM)Labyrinthinesecurity Wrote: You are not allowed to view links. Register or Login to view.The proposed structure to sustain the A/B split is a 3-layers construct: a template system, a switch modulating the use of these templates (per folio), and an independent d/l dimension.
Thank you for the answer, it didn't clear up things for me, but maybe this will help the others.
Are there any implications for potential deciphering attempts or Voynichese generators?
Thanks for your honest feedback
I am updating the paper and I'm uploading the support material on github right now, I hope it will clarify things.
Yes, there are some implications for potential generators (generators are not meant to decipher, but I get your point):
1) Discrete folio-level switching between two states with well-separated rates
2) Compatibility with a template system with fixed and switchable classes, where fixed templates are invariant to the switch and switchable templates respond strongly with no reversals.
3) A gradient, orthogonal to the bimodal switch
4) At least three functional regimes among character pairs (categorical, intermediate, free-variation)
(05-05-2026, 11:59 AM)dashstofsk Wrote: You are not allowed to view links. Register or Login to view.I have difficulty understanding this also. You need to remember that most people visiting this forum are unfamiliar with the technical, analytical and statistical jargon that you have used. For their benefit, please, could you state in more simple terms what your conclusion is.
We propose that the Currier languages A and B, which are assigned to each folio of the manuscript, are the consequence of a mechanism dictating which kind of words are permissible and which are not in each folio. Hope this helps?
Do you still hold the claim that there are regular differences in plant images on Currier A and Currier B pages?
(05-05-2026, 12:06 PM)Labyrinthinesecurity Wrote: You are not allowed to view links. Register or Login to view.Yes, there are some implications for potential generators (generators are not meant to decipher, but I get your point):
1) Discrete folio-level switching between two states with well-separated rates
2) Compatibility with a template system with fixed and switchable classes, where fixed templates are invariant to the switch and switchable templates respond strongly with no reversals.
3) A gradient, orthogonal to the bimodal switch
4) At least three functional regimes among character pairs (categorical, intermediate, free-variation)
So, from your answer I assume there are some implications for the generators but not really for direct deciphering attempts?
(05-05-2026, 12:24 PM)Rafal Wrote: You are not allowed to view links. Register or Login to view.Do you still hold the claim that there are regular differences in plant images on Currier A and Currier B pages?
Nice question! Let me clarify, I believe there are 2 questions :
Question 1) I claimed that the switch sided with the features of the Herbal morphometrics and against Currier in some folia of the manuscript. I'm still investigating this, but I think I was wrong. The switch is pretty much aligned with Currier and the Herbal morphometrics.
Question 2) there are differences in plant images on Currier A and Currier B pages, but I am not part of the team who figured this out; its Gheuens, K. and Rapaport, C.
I have verified that their classification aligns with Currier A and B, and I can tell you that, if the features are correctly assigned to each plant, the align is quite robust, so their claim sounds good. But it is a bit circular, since the mapping between features and languages was made "after the fact". There are no held-out folios in that story. Still, their finding is remarkable (just my personal opinion)
(05-05-2026, 12:34 PM)oshfdk Wrote: You are not allowed to view links. Register or Login to view. (05-05-2026, 12:06 PM)Labyrinthinesecurity Wrote: You are not allowed to view links. Register or Login to view.Yes, there are some implications for potential generators (generators are not meant to decipher, but I get your point):
1) Discrete folio-level switching between two states with well-separated rates
2) Compatibility with a template system with fixed and switchable classes, where fixed templates are invariant to the switch and switchable templates respond strongly with no reversals.
3) A gradient, orthogonal to the bimodal switch
4) At least three functional regimes among character pairs (categorical, intermediate, free-variation)
So, from your answer I assume there are some implications for the generators but not really for direct deciphering attempts?
Exactly! As far as I'm concerned, I'm not working on cracking/deciphering the VMS, but on finding hints on its underlying structure through machine learning, which I think is more likely to bear fruits at the moment.
I have been too much of a hater recently, so I will say something nice about machine learning: It has certainly allowed people to post results to this forum