In Memory of Man Podcast - Robot Crime Blog

The Voynich Manuscript: Cracked With a Brute-Force Attack in a Weekend"

robotcrimeblog.com Season 1 Episode 67

Use Left/Right to seek, Home/End to jump to start or end. Hold shift to jump forward or backward.

0:00 | 20:06

600 years of cryptographers couldn't read the Voynich Manuscript. A weekend brute-force attack cracked it. It's a medieval drug manual. Six plants identified, including Cannabis on folio 16r. Full proof, free, public domain →


  https://voynich-production-8b6c.up.railway.app/

SPEAKER_01

Imagine inventing this incredibly complex uh twenty-two character secret alphabet. Oh yeah you spend years, I mean literally years perfecting the cryptography. You write hundreds of pages of beautifully illustrated, completely impenetrable code.

SPEAKER_00

Right.

SPEAKER_01

And what is the grand universe altering secret you're trying to hide? It's uh the recipe for a medieval cilantro hangover cure.

SPEAKER_00

Which is just so incredibly human, honestly.

SPEAKER_01

It really is. Now, for those of you who have been obsessively following the saga of Yale Benecki MS408, well, better known to the world as the Voinage manuscript, you already know the stakes here.

SPEAKER_00

Oh, absolutely.

SPEAKER_01

We aren't going to rehash all the centuries of bizarre theories about, you know, aliens or time travelers.

SPEAKER_00

Thankfully, let's skip the aliens.

SPEAKER_01

Yeah, exactly. Instead, our mission for this deep dive is to explore a brand new document. It's dated today, actually, June 20, 2026. And this paper claims to present the most complete computational decipherment of the Voynich manuscript ever produced in a single session.

SPEAKER_00

Aaron Powell And honestly, this document truly reshapes our entire understanding of the text. Yeah, I mean, what we are looking at isn't just another fringe theory. It is this massive multidisciplinary synthesis. You have computational linguistics, advanced paleography, and vision AI, and they've all like finally converged.

SPEAKER_01

Which is wild.

SPEAKER_00

It is. The researchers have provided a functional 22-glyph cipher key, and it actually allows us to read real 15th-century botanical remedies that have been locked away for what 600 years.

SPEAKER_01

Okay, so let's unpack this because anytime someone claims to have, quote, cracked the voyage, the entire cryptography community just collectively rolls its eyes.

SPEAKER_00

Oh, yeah. They groan loudly. Right.

SPEAKER_01

We've all seen the graveyard of failed translations, and the biggest trap cryptographers fall into is, well, it's apaphenia.

SPEAKER_00

The human brain's desperate desire to see patterns and random noise.

SPEAKER_01

Exactly. If I stare at a TV scream full of static long enough, I mean eventually I'm going to see a face. Yeah, you will. So before we even talk about the cipher key itself, we have to look at how this new 2026 computational analysis mathematically guarantees that the manuscript isn't just some elaborate empty hoax, like a locked box with nothing inside.

SPEAKER_00

Well, this raises a really important question. How do we identify the ghost of a real language using nothing but mathematics? Like completely stripping away human bias. Right. The researchers tackled this by feeding just an unprecedented data set into their models. We are talking about um 37,919 word tokens. Wow. That's a lot. Yeah, comprising over 8,000 unique words. And all of them constructed from this bizarre 22-character alphabet that they call the EVA glyphs.

SPEAKER_01

But you know, just having a lot of words doesn't mean those words actually mean anything. I could generate 30,000 random strings of letters on my keyboard right now, and it would just be complete gibberish.

SPEAKER_00

Precisely. Which is why they utilized a core cryptographic metric. It's known as the index of coincidence or IC.

SPEAKER_01

Okay, break that down for us.

SPEAKER_00

So to understand the IC, you have to think about the mechanical structure of natural language. If you throw, say, a thousand scrabble tiles on the floor, the distribution of those letters is essentially flat and random.

SPEAKER_01

Right. It's just a mess. Right.

SPEAKER_00

The IC of completely random gibberish sits at about 0.038. But human languages are clumpy.

SPEAKER_01

Clumpy.

SPEAKER_00

Yeah. Clumpy. We have syntax. In English, the letter E appears constantly. Q is almost always followed by U, and certain suffixes just repeat endlessly. Aaron Ross Powell Right.

SPEAKER_01

So a higher index of coincidence means the text is more rigidly structured. It's like clumped together in predictable human ways.

SPEAKER_00

Exactly. So for context, standard Italian has an IC of.00738.

SPEAKER_01

Okay.

SPEAKER_00

The Voinish manuscript has an IC of.0771.

SPEAKER_01

Wait, wait. So the voinish script is actually more structured than natural Italian.

SPEAKER_00

Yes. And that is a massive clue about the underlying dialect and also the nature of the cipher itself.

SPEAKER_01

That's fascinating.

SPEAKER_00

It really is. But the true nail in the coffin for the whole hoax theory is what happens when you test the periodic index of coincidence.

SPEAKER_01

The periodic IC.

SPEAKER_00

Yeah. If a Renaissance trickster was using a polyalphabetic cipher-like, say a visioneer cipher, where the translation alphabet constantly shifts using a hidden keyword just to confuse code breakers, the periodic IC graph would show massive spikes and valleys.

SPEAKER_01

It would be all over the place.

SPEAKER_00

Exactly. But the voinish graph is completely flat. It stays right at.077. This proves mathematically that we are dealing with a direct substitution cipher.

SPEAKER_01

A substitution cipher where one voinage symbol essentially equals one Latin letter. Right. But wait, if it's just a simple substitution, why hasn't frequency analysis cracked it before? I mean, people have been trying for decades.

SPEAKER_00

Because the grammar rules are highly specialized. Look at the phonetic mapping they discovered. In the manuscript, 42.7% of the words end in a specific glyph that kind of looks like the letter Y. Okay. If you map that mathematically, it perfectly matches the frequency of Italian words ending in the consonant N, like con or non.

SPEAKER_01

But wait, standard Italian words almost always end in vowels. You know, pizza, roma, spaghetti. Why would nearly half the words end in a consonant?

SPEAKER_00

Uh, because we aren't looking at modern standardized Italian.

SPEAKER_01

Oh, I see.

SPEAKER_00

We are looking at a 15th century northern Italian dialect. So perhaps Venetian or Lombard. And in those dialects, dropping the final vowel, which is called apicope, was incredibly common.

SPEAKER_01

Okay, that makes sense.

SPEAKER_00

Yeah. And when you combine that dialectical quirk with highly specialized botanical abbreviations, standard frequency analysis models just break down completely.

SPEAKER_01

That is absolutely brilliant. I mean, the math proves the room exists and it even tells us the shape of the room.

SPEAKER_00

It does.

SPEAKER_01

But math alone doesn't turn on the lights. You still need a wedge to physically pry the cipher open.

SPEAKER_00

Right. You need a way in, a starting point that anchors all the rest of the 22 glyphs.

SPEAKER_01

And here's where it gets really interesting for you listening.

SPEAKER_00

Yeah.

SPEAKER_01

Because the wedge that cracked this entire 600-year-old mystery came down to a single two-letter substitution.

SPEAKER_00

Yeah, just two letters.

SPEAKER_01

The researchers hypothesized that the vointage digraph, Cho, which by the way, appears everywhere in the text, translates to the letter V or U. And they didn't just guess this. It was a collision of totally independent disciplines, all pointing at the exact same conclusion.

SPEAKER_00

It's the ultimate convergence of evidence. I mean, computationally, when they fed the rule chi equals V or U into the database, it immediately produced double the amount of direct 15th century Italian word matches compared to any other vowel or consonant substitution.

SPEAKER_01

Wow.

SPEAKER_00

Yeah, it was a statistical runaway winner.

SPEAKER_01

But they didn't just rely on the supercomputers, right? They actually look at the physical pen strokes on the vellum.

SPEAKER_00

Right, the paleographic evidence. Years before this computational model even existed, an independent researcher named Pelling published a visual shape analysis. Okay. By studying how the medieval quill actually moved across the page, he demonstrated that the Chocoliff was a stylized evolutionary mutation of the Latin letters U and V.

SPEAKER_01

That is so cool.

SPEAKER_00

And mathematically, this digraph makes up about 6.8% of all manuscript characters, which beautifully, perfectly mirrors the combined 5.1% frequency of V and U in standard Italian botanical texts.

SPEAKER_01

I just love when handwriting analysis and supercomputers arrive at the exact same answer.

SPEAKER_00

It's so satisfying.

SPEAKER_01

But you know, to prove it, you have to actually apply the substitution to the text and see if it yields anything functional.

SPEAKER_00

And it did. Once they established that troll operates as V or U, the vocabulary of a 15th-century apothecary just began to emerge from the noise.

SPEAKER_01

Just started pouring out.

SPEAKER_00

Exactly. The manuscript word chol decoded to vowel, pointing toward Balerian root. The string shea translated to vin, indicating wine or vinca. Oh wow. But the absolute confirmation, like the definitive Rosetta Stone of the Voinish manuscript, is found on a very specific page designated as folio F9V.

SPEAKER_01

Yes, the Viola page. I love this part. Walk us through how this specific illustration acted as the key, because the way multiple disciplines collided here is just so incredibly satisfying for anyone who loves problem solving.

SPEAKER_00

So on this folio, there is a vibrant, deeply detailed illustration of a plant. Decades ago, human botanists identified the drawing as viola tricolor, commonly known as heartsees. Right. Recently, advanced vision AI models analyzed the specific botanical morphology. So the blue-violet five-petaled flowers, the loof structure, and the AI independently corroborated that it belonged to the Barago or Antusa family, which firmly aligns with the viola identification.

SPEAKER_01

So the human experts and the AI both totally agree. We're definitely looking at a drawing of a viol.

SPEAKER_00

Exactly. And directly adjacent to that flower, the very first word in the enciphered text reads C-H-E-O-L.

SPEAKER_01

And when you apply the wedge?

SPEAKER_00

When you apply the cryptographic wedge, E fai becomes V. The E maps phonetically to I, the O remains O, and the L remains L. So C E O decodes flawlessly to viol, the Latin and Italian root for viola.

SPEAKER_01

Boom. That one four-letter word just clicks the entire lock open.

SPEAKER_00

It really does.

SPEAKER_01

Because if you know that word is V-O-L, you can reverse engineer the surrounding characters and suddenly the cipher just cascades. Yep.

SPEAKER_00

It all unravels.

SPEAKER_01

Armed with this key, the Voinich manuscript completely sheds its reputation as some esoteric magical puzzle from another dimension. It transforms into an intensely practical pharmaceutical formulary.

SPEAKER_00

An incredibly grounded functional manual for a practicing renaissance physician. And the researchers proved this by running their decoded translations against the gold standard of historical medicine, Diascorides' De Materia Medica, which was basically the absolute authority on pharmacology for over a millennium.

SPEAKER_01

Okay, so let's talk about the actual medicine, because this is where the cipher hits the real world. And I really want to know what this 15th century doctor was prescribing. Let's start with some of the safer everyday remedies.

SPEAKER_00

A perfect example is their translation of the Valerian page. Valerian, which the manuscript decodes as Val, was highly regarded in the medieval world as all heel.

SPEAKER_01

All heal. I like that.

SPEAKER_00

Yeah, it was primarily prescribed for insomnia, severe anxiety, and nerve pain. And what's remarkable is how safe and practical the application is. Yeah. The text aligns beautifully with the teachings of the Salernatin Medical School, which was the first and most prominent medical school in medieval Europe. They advised preparing valerian via cold water maceration. And the dosage instructions even translate to a highly pragmatic standard, basically use as much as fits in three fingers.

SPEAKER_01

I love that. The idea of a medieval doctor writing encrypted, super secret prescriptions that essentially just say, you know, take three fingers of this route and call me in the morning. It's so deeply human.

SPEAKER_00

It really grounds the whole thing, doesn't it?

SPEAKER_01

It does. And you see that same highly specific dosing with the centauri plant, right?

SPEAKER_00

Yes, centauri, which they decoded with about 70% confidence. It was utilized as a powerful stiptic for lacerations and snake bites. Okay. The linguistic decoding here is fascinating because a researcher named Bax had previously translated the raw text as C.H. iron.

SPEAKER_01

Chiron. Wait, like the mythological centaur?

SPEAKER_00

Exactly. The mythological centaur Chiron was said to have healed Hercules with this very plant, which is literally why we call it Centauri today.

SPEAKER_01

That is amazing.

SPEAKER_00

And the Voynage text offers an incredibly strict dosage. Two drams of the root infused in wine. It's actually one of the few instances where Dioscorides himself demanded such explicit measurements.

SPEAKER_01

But the medieval medicine cabinet gets a lot darker and honestly a lot more dangerous when you look at the treatments for things like intestinal parasites.

SPEAKER_00

Oh yes.

SPEAKER_01

Which brings us back to my favorite bizarre hook from this research, the encrypted cilantro.

SPEAKER_00

Ah, yes. Coriandrum setivum or coriander. So that vision AI and the linguist analysis perfectly agreed on the deeply pinnate lacy leaves and compound umbells shown in the manuscript. And the decoded Italian text instructs the physician to utilize the ways of wine of the plant, which indicates a really potent alcoholic extraction.

SPEAKER_01

Okay, but we use coriander in tacos today. Why would a doctor bother encrypting a recipe for cilantro wine?

SPEAKER_00

Because in the 15th century, they weren't just like sprinkling leaves on their food, they were utilizing primitive alembics to distill incredibly concentrated essential oils and decoctions.

SPEAKER_01

They did this to aggressively kill internal parasites and fungal infections. And if you extract coriander at that magnitude, it completely crosses the threshold from medicine to poison.

SPEAKER_00

Yikes.

SPEAKER_01

Yeah. Diascortes explicitly warns that an excess dose of concentrated coriander causes severe amnesia, violent confusion, and even death.

SPEAKER_00

So you aren't just hiding a recipe, you are hiding a highly toxic manufacturing process.

SPEAKER_01

Exactly.

SPEAKER_00

That makes a lot of sense. But the decipherment of these plants isn't always smooth sailing, right? There's some massive academic friction in this document when the Vision AI and the linguists actually start bumping heads.

SPEAKER_01

Yes, and we cannot ignore this conflict because it really highlights how volatile this translation process remains. Right. Take the folio that the Vision AI confidently identified as cannabis sativa. The AI analyzed the palmate compound leaves with seven to nine leaflets, the tuberous root, and the reddish berry clusters, and it flagged it as hemp with 80% confidence.

SPEAKER_00

And what was the prescribed use for that? Pain, inflammation, insomnia, and muscle spasms, treated by crushing the seeds in wine, which perfectly tracks with historical texts warning that excessive use, quote, dulls the senses.

SPEAKER_01

Makes sense.

SPEAKER_00

However, an incredibly respected linguist named Bax heavily contested this. Bax looked at the unenciphered Arabic text on that exact same page, which just plainly reads, ARR. AR. Yeah. And in medieval Arabic medicine, ARR refers to juniper or cypress.

SPEAKER_01

Oh. So the AI is telling us we are looking at a picture of weed, essentially. But the actual text on the page says juniper. Why the massive clash? Is the AI just struggling to parse these highly stylized, weirdly proportioned 15th-century doodles?

SPEAKER_00

Well, that is certainly one plausible explanation. I mean, the Sheen Vision is trained on photographs of real anatomically correct plants, right? Not surrealist Renaissance watercolors. Right, right. But there is another, much more fascinating possibility: intentional steganography.

SPEAKER_01

Wait, you mean the author deliberately mismatched the text and the images as like an intellectual booby trap?

SPEAKER_00

Exactly. Think about it. If you are a physician trying to protect your proprietary parasite cures and pain relievers from rival doctors, you might draw a picture of a climbing vine like Brayonia, which the AI actually spotted on another contentious page. Okay. But then you label it with the Arabic word Kaur, meaning blackseed. If a rival doctor steals your book and follows the picture, they might accidentally kill their patient with toxic braonia instead of healing them with black seed.

SPEAKER_01

Oh wow. If we connect this to the bigger picture, that theory perfectly sets up the broader mystery of the manuscript. Yight because it begs the question, who was this person?

SPEAKER_00

Right.

SPEAKER_01

We know it's not a hoax now, we know it's functional, sometimes highly dangerous medicine. But who goes to the immense trouble of inventing an entirely new cryptographic alphabet just to write down plant recipes?

SPEAKER_00

Well, based on the linguistic and historical data, the researchers have proposed a hypothesized author profile, though they assign it a pretty modest 55% confidence rating.

SPEAKER_01

Fair enough.

SPEAKER_00

They believe the author was an Italian Jewish physician operating in the early 15th century.

SPEAKER_01

Walk us through the reasoning there, because the evidence for that is surprisingly compelling.

SPEAKER_00

It really comes down to what is present in the text and what is glaringly absent.

SPEAKER_01

Okay.

SPEAKER_00

First, look at the ingredients of the language itself. The underlying grammar relies heavily on foundational Italian prepositions. Words decoding to dela, dal di.

SPEAKER_01

Italian foundations got it.

SPEAKER_00

Yet the medical terminology is deeply infused with raw Arabic, utilizing terms from towering Islamic medical texts like Abacina's canon of medicine.

SPEAKER_01

So it's basically this melting pot of European and Middle Eastern science.

SPEAKER_00

Exactly. But the most staggering piece of evidence is the visual context. Across 214 pages of an illustrated manuscript produced in early 1400s Europe, there is a total, absolute absence of Christian iconography.

SPEAKER_01

Really? None at all.

SPEAKER_00

None. There are no crosses, no saints, no subtle nods to the Catholic Church whatsoever. For medieval European document, that is almost statistically impossible, unless the author operated completely outside that dominant religious structure.

SPEAKER_01

Right. An Italian Jewish physician would be uniquely positioned to absorb the Greek texts of Dioscortes, the Arabic mastery of Avicenna, and the local Italian herbal traditions, all while remaining culturally distinct from the Catholic hegemony.

SPEAKER_00

Exactly.

SPEAKER_01

And in the Renaissance, medical knowledge wasn't just some open source public good, it was raw economic power. If you had a unique, effective treatment for a wealthy merchant's gout or a nobleman's insomnia, that was a highly lucrative trade secret.

SPEAKER_00

It was intellectual property. And you protected your intellectual property with a cipher.

SPEAKER_01

So what does this all mean? When you pull back the curtain, the Voyage manuscript is simply the incredibly practical, heavily encrypted notebook of a multilingual physician who was just trying to maintain his competitive edge.

SPEAKER_00

Pretty much, yeah.

SPEAKER_01

But before we get too celebratory about cracking the case, we have to inject a pretty heavy dose of reality into this deep dive.

SPEAKER_00

Yes, we do. And to be fair, the researchers themselves are remarkably transparent about this limitation. The reality is no one has fully translated the Voynich manuscript. Not even close.

SPEAKER_01

Yeah, that is the crucial caveat here.

SPEAKER_00

The manuscript contains 5,137 lines of text. Using this newly discovered computational key, the team has only been able to produce readable, coherent, syntactically correct Italian for 272 of those lines.

SPEAKER_01

Wow, so we haven't emptied the save at all. We've just managed to shine a flashlight on the very first stack of papers sitting right by the door.

SPEAKER_00

Exactly. The vast majority of the vault is still completely shrouded in darkness.

SPEAKER_01

And what's fascinating here is a lingering puzzle pulled directly from the data, a bizarre paradox that the researchers highlight, which completely upbands our understanding of why this book was even written.

SPEAKER_00

Oh, this is the best part.

SPEAKER_01

Lay it on us.

SPEAKER_00

So throughout the manuscript, the author writes down specific Arabic and Persian plant names, like AR for juniper or KAR for black seed, but they don't encrypt them. Wait. Those vital exotic medical terms are left entirely unenciphered. They're just written as raw labels directly on the page.

SPEAKER_01

Wait, wait. So the most important active ingredients, the exotic foreign medicines, are just sitting there in plain text for literally anyone to read?

SPEAKER_00

Yes. But the mundane, everyday Italian words, the simple verbs, the prepositions like of or from, those are hidden behind a maddeningly complex, mathematically rigid, 22-glyph computational cipher.

SPEAKER_01

That makes absolutely no logical sense. I mean, if your entire goal is to protect your valuable medical recipes, why would you leave the critical active ingredients completely readable but then go to the exhausting trouble of encrypting the word wine?

SPEAKER_00

It's baffling.

SPEAKER_01

It really forces you to ask a very strange question about the author's true intent.

SPEAKER_00

It does. Was this brilliant cryptographic system actually designed to hide the medicine? Or was the medicine just the unencrypted wrapper? Like a mundane distraction meant to divert our attention from whatever else is hidden in the 4,800 lines of text that we still cannot read.

SPEAKER_01

Talk about a chilling thought. You finally crack the lock safe, you pull open the heavy metal door, and you realize the blueprints inside are for an entirely different building. Well, we will leave you to mull over that paradox on your own. Thanks for joining us, and we'll catch you on the next deep dive.