The Voynich Ninja

My open-access paper describing the Naibbe cipher is now out in Cryptologia: You are not allowed to view links. Register or Login to view.

Thank you all very much for letting me present the cipher at VMD and for all your insightful feedback on the forum. It greatly improved the final paper.

That's very nice! As you say, it looks a promising road to tread. My compliments.

Thanks for the paper. I would have thought that articles on CRYPTOLOGIA were always behind a paywall. That's obviously not the case.

(27-11-2025, 03:19 PM)bi3mw Wrote: You are not allowed to view links. Register or Login to view.Thanks for the paper. I would have thought that articles on CRYPTOLOGIA were always behind a paywall. That's obviously not the case.

There is an extra expense that authors must pay to make a study open-access. On the balance, I decided it would be worth it in the long term to pay that up-front cost.

Congratulations on getting a paper in Cryptologia      i am so jelly
Very nice, you worked hard and well deserve it.
+1 for making it open-access.

For me a key point is where you say:

"do not reliably reproduce the exact observed diversity of VMS word types, especially rare ones."

Thanks for stating that. I appreciate you mentioning that as rarer words are often overlooked by Voynich researchers. I have nearly finished a long document on the rarer and more distinctive words.

Such intriguing work! Congratulations!

Well done!
And great that you made it open access.

Congratulations on publishing probably the most coherent paper I have seen from the Voynich community in a long time. I would just like to clarify a couple points based on my understand:

1. The tables that you use cannot produce standalone EVA 's', 'l' 'o' 'r' or 'd'? I saw that 'y' is a unigram option but none of the other common standalone single glyphs. Or is this accounted for by the type-1 and type-2 affixes?

2. You focus only on Voynich B, did you do many test on Voynich A as well? My personal opinion is that Voynich A (written by Scribe 1) is the most coherent and shows that Scribe 1 was the most famililar and/or professional of all the scribes with Voynichese. I tend to think that Voynich B would have more 'mistakes' or general inaccuracies compared to A. 

3. Did you try languages other than latin? I'm not sure if the statistics of the results of a German text would be that different than a latin text but it would be interesting to see.

I'm glad that you have a section on the flaws of the cipher itself. I think that any statements of the Naibbe Cipher being too complex, are just being dismissive. Granted that if you were to overfit certain parameters just to meet the statistical properties you want then that would be too far. But if we are to achieve a cipher solution, it has to be significantly more complex than those that we have good historical evidence of from the time, otherwise we would have solved it already.

(29-11-2025, 12:35 PM)Skoove Wrote: You are not allowed to view links. Register or Login to view.Congratulations on publishing probably the most coherent paper I have seen from the Voynich community in a long time.

Thanks! I'll take your questions one by one:

Quote:1. The tables that you use cannot produce standalone EVA 's', 'l' 'o' 'r' or 'd'? I saw that 'y' is a unigram option but none of the other common standalone single glyphs. Or is this accounted for by the type-1 and type-2 affixes?

Correct. This decision was in part a product of wanting as much structural similarity as possible among the unigram word types, as a proof of concept. But at least some of these standalone glyphs are valid bigram prefixes or suffixes. I don't talk about it explicitly in the paper, but a scribe could write a standalone bigram prefix or suffix in lieu of a unigram word type. While it would disobey the formal rules of the cipher, this solo prefix/suffix would also be decipherable (albeit potentially ambiguous). For example, <s> is one of the bigram prefixes within the Naibbe cipher that stands for M.

Quote:2. You focus only on Voynich B, did you do many test on Voynich A as well? My personal opinion is that Voynich A (written by Scribe 1) is the most coherent and shows that Scribe 1 was the most famililar and/or professional of all the scribes with Voynichese. I tend to think that Voynich B would have more 'mistakes' or general inaccuracies compared to A. 

I focused on Voynich B for two reasons. For one, there's simply more of Voynich B than Voynich A, and I wanted as much data as possible. Second, the unusual proportional frequency-rank distribution of Voynich B, You are not allowed to view links. Register or Login to view., is what specifically inspired some of the modeling that resulted in the Naibbe cipher. I agree that Scribe 1 knew their way around the script; they're the one who's most likely to do a complex ligature that connects two adjacent tokens, for example.

Quote:3. Did you try languages other than latin? I'm not sure if the statistics of the results of a German text would be that different than a latin text but it would be interesting to see.

Yes, I did. In some initial modeling work (described in Supplementary Material 3), I found that the cipher happened to work especially well with Latin and Italian. But the cipher totally can be applied to other languages, as well. I know of at least one use of the cipher to encrypt German, as a demo of the (very cool!) Python library pykeedy: You are not allowed to view links. Register or Login to view.

Quote:I'm glad that you have a section on the flaws of the cipher itself. I think that any statements of the Naibbe Cipher being too complex, are just being dismissive. Granted that if you were to overfit certain parameters just to meet the statistical properties you want then that would be too far. But if we are to achieve a cipher solution, it has to be significantly more complex than those that we have good historical evidence of from the time, otherwise we would have solved it already.

I agree: Statistically, the VMS is inconsistent with well-characterized historical ciphers from the period. So if it's a cipher—and it may not be!—then we should be open to a cipher structure that's unusual or undocumented. That said, any such cipher has to be historically plausible: doable by hand with 15th-century materials and broadly in line with the era's cryptographic principles. The Naibbe cipher can be seen as an extreme elaboration of some of the design principles of the diplomatic homophonic ciphers used in 15th-century northern Italy. Again, that doesn't mean the Naibbe cipher literally existed at the time, but it's not as if I need some anachronistic clockwork gizmo to make the cipher work.