The Voynich Ninja

(13-04-2026, 10:29 AM)oshfdk Wrote: You are not allowed to view links. Register or Login to view.

(12-04-2026, 10:16 PM)CorwinFr Wrote: You are not allowed to view links. Register or Login to view.Hi oshfdk, I have redo eveything you have the clear analysis.

The attached PDF seems to contain a list of claims, not the analysis or any supporting evidence. The claims by themselves seem quite arbitrary to me. I'm not saying your approach is wrong, but it would be nice to have at least one piece of specific evidence that supports the claims. What's the strongest argument that supports your hypothesis?

I have already retracted one claim, I'm sorry I was lost in the Voynich during night experiment and don't respect fondamental bas I will stop for a while and I'm sorry for quite eveyrhing.

Anyway here was the reasonning :

On claim that anyone can verify in five minutes with the ZL transcription. Hopefully, I'm not sure of anything.

The gallows block-initial rule

Pick any paragraph (starred block) in the VMS. Look at its first word. In 79% of cases (620 out of 784 blocks), that word starts with a gallows character (p, t, k, or f in EVA).

Now look at every other word in the manuscript, the ones that are NOT first in their block. Only 6% of those start with a gallows.

That's a 12.8× enrichment. Gallows characters are almost exclusive to block-initial position. This is not subtle, it's a hard positional constraint on how the writing system works.

Two more, equally easy to verify:

The character n appears in 6,116 words. In 6,030 of those (98%), it is the last character. It almost never appears word-internally.

The character q appears in 5,420 words. In 5,389 of those (99%), it is the first character. It almost never appears word-internally.

What thispossibly means

These are not statistical tendencies, they are near-absolute positional rules. The characters don't encode phonemes (no natural language restricts a letter to one position with 98-99% consistency). They encode structural roles:

- Gallows = "this word starts a new paragraph/recipe"
- n = word-final marker
- q = word-initial marker

This is how abbreviation systems work: special markers for word boundaries and structural positions, separate from the content characters. That's one of the reasons the Tironian shorthand model fits — Tironian notes use dedicated strokes for positional modification of base signs.

These three facts are verifiable by anyone with a spreadsheet and the ZL transcription. No botanical identification needed, no corpus matching, no statistics beyond counting.

On top of that I build some method to encore properly the Voynich into JSON, compare with a corpus, etc. etc. etc.

Anyways as I said I was in a infernal spiral on this Woynich it's so... fascinating. Surelly all of this is wrong. I stop everything on this and I'm sorry.

(13-04-2026, 10:39 AM)CorwinFr Wrote: You are not allowed to view links. Register or Login to view.This is how abbreviation systems work: special markers for word boundaries and structural positions, separate from the content characters. That's one of the reasons the Tironian shorthand model fits — Tironian notes use dedicated strokes for positional modification of base signs.

But this is not a particularly strong argument? If two things have a common feature, this doesn't mean they are the same thing. It's a bit like saying "a giraffe is yellow and has spots, so giraffes are leopards". The statistics regarding paragraph initial characters have been well known for decades, Tironian notes have been researched in the context of the Voynich Manuscript many times, I think successfully linking both requires a valid model that would explain how exactly this notation works.

(13-04-2026, 11:00 AM)oshfdk Wrote: You are not allowed to view links. Register or Login to view.

(13-04-2026, 10:39 AM)CorwinFr Wrote: You are not allowed to view links. Register or Login to view.This is how abbreviation systems work: special markers for word boundaries and structural positions, separate from the content characters. That's one of the reasons the Tironian shorthand model fits — Tironian notes use dedicated strokes for positional modification of base signs.

But this is not a particularly strong argument? If two things have a common feature, this doesn't mean they are the same thing. It's a bit like saying "a giraffe is yellow and has spots, so giraffes are leopards". The statistics regarding paragraph initial characters have been well known for decades, Tironian notes have been researched in the context of the Voynich Manuscript many times, I think successfully linking both requires a valid model that would explain how exactly this notation works.

You're right that shared features don't prove identity, that's a fair objection. I'm not claiming the VMS is Tironian.

I'm claiming personal mnemonic shorthand is a serious hypothesis that hasn't been adequately explored.

The reasoning is simple: if this were a straightforward cipher or a known language, someone would have solved it by now. 600 years of brilliant minds have failed.

That failure itself is a constraint, it tells us the search space is probably not what we think it is.

A cipher has a finite key space. A language has phonetic structure. Both are attackable. But a personal mnemonic system, abbreviations invented by one person for his own use, memorized through daily practice, never shared, creates a search space that is essentially infinite. There is no key to find, no grammar to reverse-engineer. That would explain the resistance to decryption better than any cipher or language model.

Tironian notes are the historical precedent for exactly this kind of system: arbitrary base signs, personal modifications, positional markers. I'm not saying "the VMS is Tironian", I'm saying the structural properties I observe (positional constraints, prefix+root+suffix compounds, logograms for function words) are consistent with that tradition. It's a hypothesis, not a conclusion.

If someone has a better model that explains both the structural regularities AND 600 years of failed decryption, I'm genuinely interested.

(13-04-2026, 11:16 AM)CorwinFr Wrote: You are not allowed to view links. Register or Login to view.I'm claiming personal mnemonic shorthand is a serious hypothesis that hasn't been adequately explored.

I think it has been explored many times including on this forum and Tironian notes have been discussed and researched, but of course it won't hurt exploring it once more. So, if and when you have any good evidence, I'd love to read about it.

Both shorthand and Tironian notes are mentioned in 1978 D'Imperio classic on the Voynich MS, so I think these are by no means new or under-explored ideas.

(13-04-2026, 12:10 PM)oshfdk Wrote: You are not allowed to view links. Register or Login to view.

(13-04-2026, 11:16 AM)CorwinFr Wrote: You are not allowed to view links. Register or Login to view.I'm claiming personal mnemonic shorthand is a serious hypothesis that hasn't been adequately explored.

I think it has been explored many times including on this forum and Tironian notes have been discussed and researched, but of course it won't hurt exploring it once more. So, if and when you have any good evidence, I'd love to read about it.



Both shorthand and Tironian notes are mentioned in 1978 D'Imperio classic on the Voynich MS, so I think these are by no means new or under-explored ideas.

There is a fascinating idea to model the Voynich and a medical corpus as heterogeneous graphs, learn node embeddings, then align the graphs to detect recurrent patterns consistent with a mnemonic shorthand system, but there is already an enormous amount of work by brilliant researchers in every direction and I definitely don’t have all that knowledge.

Before trying I was just testing if it could be a good option.

(13-04-2026, 12:10 PM)oshfdk Wrote: You are not allowed to view links. Register or Login to view.

(13-04-2026, 11:16 AM)CorwinFr Wrote: You are not allowed to view links. Register or Login to view.I'm claiming personal mnemonic shorthand is a serious hypothesis that hasn't been adequately explored.

I think it has been explored many times including on this forum and Tironian notes have been discussed and researched, but of course it won't hurt exploring it once more. So, if and when you have any good evidence, I'd love to read about it.

Both shorthand and Tironian notes are mentioned in 1978 D'Imperio classic on the Voynich MS, so I think these are by no means new or under-explored ideas.

What would be your best advice or best direction for a computational analysis by Machine Learning (not hallucinating LLM) ?

(13-04-2026, 12:52 PM)CorwinFr Wrote: You are not allowed to view links. Register or Login to view.What would be your best advice or best direction for a computational analysis by Machine Learning (not hallucinating LLM) ?

I think it makes sense first to spend a few weeks or more studying what has been attempted so far. For some recent ML attempts I think there have been a few threads by @quimqu.

Personally, I don't think ML is of much use without some good underlying hypothesis that can explain many puzzling features of the manuscript. I believe most attempts of just throwing some computational power at the manuscript and expecting a breakthrough are doomed.

(13-04-2026, 01:06 PM)oshfdk Wrote: You are not allowed to view links. Register or Login to view.

(13-04-2026, 12:52 PM)CorwinFr Wrote: You are not allowed to view links. Register or Login to view.What would be your best advice or best direction for a computational analysis by Machine Learning (not hallucinating LLM) ?

I think it makes sense first to spend a few weeks or more studying what has been attempted so far. For some recent ML attempts I think there have been a few threads by @quimqu.

Personally, I don't think ML is of much use without some good underlying hypothesis that can explain many puzzling features of the manuscript. I believe most attempts of just throwing some computational power at the manuscript and expecting a breakthrough are doomed.

Yeah I think you are right, I made the same mistake as a lot before me, going full speed against a wall. Now I will study the wall and the landscape...