Has the Voynich Manuscript finally been cracked? Is the following a plausible explanation?

By Joseph D'Agnese
Wired Magazine 9-1-4

 

In 2002, an Englishman named Gordon Rugg slipped back in time. Night after night he spread his papers on the kitchen table once his children had gone to bed. Working on faux parchment with a steel-nibbed calligraphic pen, he scribbled a strange, unidentifiable, vaguely medieval script. Transliterated into the Roman alphabet, some of the words read: "qopchedy qokedydy qokoloky qokeedy qokedy shedy." As he wrote, he struggled to get inside the mind of the person who had first scrawled this incomprehensible text some 400 years ago.

 

By day, Rugg, a 48-year-old psychologist, teaches in the computer science department of Keele University, near Manchester, England. By night, as an intellectual exercise, he has been researching one of the world's great oddities: the Voynich manuscript, a hand-lettered book written in an unknown code that has frustrated cryptographers since its discovery in an Italian villa in 1912. How impregnable is the Voynich? During World War II, US Army code breakers - the guys who blew away Nazi ciphers - grappled with the manuscript in their spare time and came up empty. Since then, decoding the book's contents has become an obsession for geeks and puzzle nuts everywhere.

 

Then came Rugg. In three months, he cooked up the most persuasive explanation yet for the 234-page text: Sorry, folks, there is no code - it's a hoax! Lifelong Voynichologists were impressed with his reasoning and proofs, even if they were a little chagrined. "The Voynich is such a challenge," says Rugg, "such a social activity. But then along comes someone who says 'Oh, it's just a lot of meaningless gibberish.' It's as if we're all surfers, and the sea has dried up."

 

When the news of Rugg's breakthrough was published last winter, everyone missed the bigger story. Rugg cracked the Voynich not because he was smarter, but because he focused on what everyone else had missed. Then again, this came naturally to Rugg: He has made a career out of studying how experts acquire knowledge yet screw up nevertheless. In 1996, he and his colleagues developed a rigorous method for peering over the shoulders of experts - doctors, software engineers, pilots, physicists - watching how they work and think, testing their logic, and uncovering ways to help them solve problems.

 

Rugg calls it the verifier approach, and the Voynich was its first major test. If Rugg gets his way, verifiers will revolutionize the scientific method and help solve other seemingly unsolvable mysteries, such as the origins of the universe or the cause of Alzheimer's disease.

 

Rugg was hardly the first to dream of cracking the Voynich. Ever since the manuscript resurfaced - bookseller Wilfrid Voynich bought it from Italian Jesuits 92 years ago - a stream of formidable scholars have pored over it. Some make pilgrimages to Yale's Beinecke Rare Book and Manuscript Library, where the volume resides. Others download JPEGs of the pages, which are available free on the Web.

 

Rugg saw something different and special about the manuscript: It would make a perfect beta test for the verifier approach. As he read about the Voynich and began applying his method - amassing knowledge about a problem and assessing the kinds of expertise applied so far are steps one and two - he saw that no one had seriously explored the idea that the book was a grand hoax. As Philip Neal, one of the world's leading Voynichologists, says, "It has been argued - I used to argue myself - that the phonetic structure was beyond the powers of a 16th-century forger to create, so that the text must be a real language or an unknown type of cipher."

 

Since none of the experts thought a hoax was plausible, no one had looked very hard for a hoax solution. To compound the problem, many Voynichologists were specialists: linguists, cryptographers, mathematicians, medievalists, and literary scholars. But the ideal Voynich expert - a code-breaking, medieval-savvy hoaxologist - probably didn't exist. And the resulting gap had allowed a major problem to go unsolved for the better part of a century.

 

This "expertise gap" is rife in academia, but few recognize it, let alone know how to correct for it. It starts with the best of intentions. Institutions want top-notch people, so they offer incentives to attract and groom experts. Young grad students learn early that if they want to carve out a niche, they must confine their interests to a narrow field. It's not enough to work in spinal cord regeneration; it must be stem cell-based solutions to the problem. That's great if a researcher just happens to stumble on a perfect stem cell cure. But as specialists get further from their core expertise, the possible solutions - what's been tried, what hasn't, what was never properly examined, what ought to be tried again - get even more elusive.

 

With the verifier approach, Rugg begins by asking experts to draw a mental map of their field. From there, he stitches together many maps to form an atlas of the universe of knowledge on the subject. "You look for an area of overlap that doesn't contain much detail," he says. "If it turns out there's an adjoining area which everyone thinks is someone else's territory, then that's a potential gap."

 

So here's Rugg, studying the Voynich on his own and asking himself: If I were living in the 16th century and wanted to make a book that looked mysterious but was really gibberish, how could I do it cheaply and easily? He deliberately searched for low tech tools capable of generating text that seemed random. In his reading, he came across an encoding device called the Cardan Grille, first described in 1550 by Girolamo Cardano. (See "How to Create an 'Indecipherable' Manuscript")

 

Using such an encoder, Rugg figures it would take a smart fraudster an hour or two to write an entire page. A Voynich-size book might take about three or four months to create with illustrations. The time and effort would definitely be worth it: In the Elizabethan era, Rudolph II, the Holy Roman emperor, became fascinated with the beautifully wrought manuscript (he believed it was the work of 13th-century philosopher Roger Bacon) and paid 600 gold ducats for it - about $30,000 today.

 

The text's author was long-dead. Rugg couldn't watch him work, but he could get inside his head by trying to replicate his pen-and-ink technique. The precision paid off. Once, when the ink blotched, Rugg swore aloud and thought about discarding the table. Then it struck him that the author must have experienced the same thing. What was the best solution? Toss it out? No. Paper was expensive. A new grid? No, too much work. Better to leave the blotch and work around it. That helped him realize that some of the cells in the original grids must have been left blank - the resulting missing syllables yielded a variety of word lengths, giving the faked language even more verisimilitude.

 

Rugg's prime suspect for the hoax is con artist Edward Kelley, a hanger-on in the court of Elizabeth I. Kelley insinuated himself into the household of the queen's astrologer, John Dee, and acted as a medium for angels. Modern scholars think Kelley was a fraud; apparently Dee did not, even when the angels suggested the two men swap wives. (Researchers believe they took the heavenly advice.)

 

Rugg published his hoax theory earlier this year in the journal Cryptologia. It was bolstered by Laura Aylward, one of his students, who used software to replicate his work. She found that the Voynich's well-known statistical anomalies - its unnatural repetition of certain words, a sudden dropoff in the use of previously common syllables - can all be accounted for if one uses structured tables and grilles. Still, many Voynich scholars remain unconvinced. "I know how they feel," Rugg says. "The rational part of me says it's a hoax; another part says, yes, but what if 10 percent of it is cipher text, a real message mixed in with all the wattle and padding? It's a lovely problem."

 

The verifier approach has been applied successfully to only one lovely problem so far - the Voynich - but Rugg is already preparing to tackle more. "We have to be careful," he says. "We are not going to barge in on people. We want to approach them with humility and tact, so they'll see us as partners."

 

His approach is built on the observation, noted as far back as the 1970s, that experts tend to cut to the chase. In their zeal to get to an answer, they make many little mistakes. (A recent study of work published in Nature and British Medical Journal, for example, found that 11 percent of papers had serious statistical errors.) Experts unknowingly fudge logic to square data with their hypotheses. Or they develop blind spots after years of working in isolation. They lose their ability to take a broader view. If all this is true, he says, think of how much big science is based on flawed intuition.

 

He gestures as if he's sculpting something invisible yet potent. Rugg's face is youthful - sharp nose, prominent forehead, all brain. He's also a paragon of well-roundedness: He studied French and linguistics, taught English in Nepal, dabbled in archaeology, and wrote a paper on the occurrence of left- and right-handedness in Stone Age peoples. Then came psychology.

 

His work bridges a number of specialties. One of the tools in the toolkit, as he says, is a field called judgment and decisionmaking. Psychological studies suggest that experts, defined as someone with 10 years in a discipline, don't have any more reasoning power than the rest of us. What they have is tons of experience. An old doctor, for instance, has seen so many cases of the mumps that he no longer follows methodical reasoning to arrive at a diagnosis. He instead uses a shortcut called pattern-matching: face red and swollen - mumps. Next!

 

Call it what you like - a hunch, an opinion - pattern-matching is iffy. "Sequential reasoning is formal, almost mathematical," Rugg says as we settle down in the campus cafeteria over mochas. "If this, then that. Pattern-matching is fast and efficient. The doctor knows what's wrong with you before you do. That's fine if he's right, but he can be miscued."

 

Besides pattern-matching incorrectly, experts sometimes misunderstand critical terms. People in similar but specialized fields will find it hard to communicate. A hydrogeologist and a petroleum engineer both took Geology 101, yet their visceral understanding of the word rock may be wildly divergent. Luckily, psychologists have developed a battery of methods - called elicitation techniques - to draw out and define what experts know.

 

As a journeyman researcher in expert reasoning, Rugg ferreted out errors and unseen problems in various industrial and office management cases. It was good work for a human-error psychologist, but he wanted to tackle bigger issues. If experts were making mistakes in doctor's offices and factories, he reasoned, they were making them in labs, too. "My gut feeling is that a lot of research involves pattern-matching," Rugg says. "It guides what is investigated and then the design of the study."

 

Sometime in 1996, while having lunch with colleague Joanne Hyde, it occurred to Rugg that he could pull together all the tools psychologists use - elicitation techniques, the vast literature on human error, decisionmaking models, formal logic and reasoning - to create a novel form of problem-solving: a scientific method to verify the methods of science.

 

The verifier method boils down to seven steps: 1) amass knowledge of a discipline through interviews and reading; 2) determine whether critical expertise has yet to be applied in the field; 3) look for bias and mistakenly held assumptions in the research; 4) analyze jargon to uncover differing definitions of key terms; 5) check for classic mistakes using human-error tools; 6) follow the errors as they ripple through underlying assumptions; 7) suggest new avenues for research that emerge from steps one through six.

 

Experts want to believe that their domain is unique, requiring specialized tools, approaches, and thinking. Rugg was saying no, you could use one kit to solve a million problems, in many fields. Many experts would object to such a theory, but computer scientists were intrigued. "In computing, we're interested in understanding knowledge," says Bashar Nuseibeh, a UK researcher whose work was influential in the development of the Unified Modeling Language. "Gordon is asking, Can you look at the commonality between two domains of research and solve problems within them with a single approach? I don't know. It's a theory."

 

Equipped with his verifier toolkit, Rugg tried finding a wealthy partner to get his new method off the ground. At least one pharmaceutical company turned him down. "It was a catch-22," Rugg says. "No one would take a chance on us unless we showed them we could do it first. But I couldn't show I could do it until someone let me test it." That of course is where the Voynich manuscript came in. Now he's poised to tackle a much bigger question.

 

Peter Crome is the deputy head of Keele's medical school and a professor of geriatrics. When he smiles, his eyes narrow to jovial slits. If I were slowly losing my mind, I think, I would want Crome to break it to me. "Alzheimer's is a mystery," he says. "An odd one. Although we know a lot about it, we don't know what causes it. All we can say at this point to someone we think has it is, Come back next year. If you're worse, you have it."

 

Alzheimer's is a classic candidate for the verifier approach: It's been studied by tens of thousands of scientists, yet still defies basic definition. Dozens of causes are under investigation: Could it be toxins in the environment? Aluminum pots and pans? Smoking or not smoking? Drinking or not? Blows to the head? Lack of mental exercise? Depression? Too much education?

 

Rugg met Crome two years ago, while he was working on the Voynich, and proposed verifying Alzheimer's studies to suggest new approaches. Crome realized he was being handed a gift. In addition to becoming too specialized, science has become too dense. The National Institutes of Health's PubMed database alone lists more than 40,000 papers pertaining to Alzheimer's. Regardless of the scientific field - medicine, nuclear physics, mathematics, whatever - PhD students typically take a year just to read what others have done before them, and even then they are usually just scratching the surface. The papers pile up faster than you can turn pages. "I can't stay current in my own area - drugs - let alone all of Alzheimer's," Crome says.

 

So how does Rugg imagine he can tackle this vast field? He's not a doctor, and he freely admits that the sum of his Alzheimer's knowledge has come from the popular press. "The idea is, sit down with these experts, have a cup of coffee, and ask, Now, off the record, what are the main theories in your field?" he says. "Then look at only those key papers in those fields. We would have consultants in each discipline and get them to help us crawl over those papers."

 

Rugg doesn't expect to find evidence of pattern-matching in a printed document, but there are always clues. "You look at the language," he explains. "Sometimes you'll see areas where the authors write, 'This result strongly resembles something else.' That's an opportunity to investigate more closely." Such a phrase is usually built on mountains of assumptions, which an author presumes (sometimes erroneously) that all potential readers understand.

 

As a first step, Rugg is preparing to apply for a grant of about $350,000 - enough to pay for a postdoc and some expenses for three years. For the moment, he's resisted the temptation to look up a single Alzheimer's paper. "I want to approach this material fresh," Rugg says.

 

He's eager to start on the project because medical problems are so urgent. "Alzheimer's is a terrible way to die," he says. "If you could do something to help, you could feel really, really good."

 

Already there are signs that the Voynich breakthrough has vanquished the nagging catch-22. Following Rugg's success with the manuscript, potential collaborators seem to surface daily. Astrophysicists at Keele have agreed to work with him on sorting out seemingly incompatible theories in modern physics. A biochemist at UC San Diego approached him about untangling questions in exobiology, the study of the chemical origins of life. A researcher at the UK's University of Warwick wants to apply the verifier method to attention deficit-hyperactivity disorder. And an emeritus art historian at Rensselaer Polytechnic Institute thinks Rugg can help her resolve mysteries about ancient Greco-Roman settlements.

 

All these collaborations are in the earliest stages. "It could be that I just got lucky once and never again," Rugg says, then pauses. "But suppose we find a breakthrough in Alzheimer's and physics?" At the very least, he'll need a bigger kitchen table.

 

After Thought: There is debate about the features of the manuscript which Rugg's suggested method was not able to emulate. The two main features are lines showing different linguistic features from the bulk of the manuscript, such as the Neal keys, and the statistical properties of the text produced. Rugg argues that these linguistic features are trivially easy to hoax using the same approach with a different set of tables, and would add about five minutes to the time to produce each page; the counter-argument is that this makes the hoax too complex to be plausible. Regarding statistics, Rugg points out that text produced from the same set of initial nonsense syllables but using different table structures shows widely different statistical properties. Since there are tens of thousands of permutations of table design, he argues that it would simply be a question of time to find a design which produced the same statistical properties as “Voynichese”. Whether or not this would prove anything useful is another issue, since it could either be used to support Rugg’s argument, or dismissed as coincidence.

Further counterarguments are:

• There is no historical evidence showing that Cardan grilles were used for this kind of purpose at any time in history
• The radiocarbon dating, codicology and palaeography all indicate that the manuscript was constructed roughly a century before Cardan grilles were devised
• The mathematical concept of randomness was not yet fully formed in the time period proposed

The debate continues. - Wikipedia

Article attribution:

 

© Copyright© 1993-2004 The Conde Nast Publications Inc. All rights reserved. © Copyright 2004, Lycos, Inc. All Rights Reserved. http://www.wired.com/wired/archive/12.09/rugg.html?tw=wn_tophead_6