Another Look at the Voynich Manuscript | Hacker News Reader

Another Look at the Voynich Manuscript

theatlantic.com

47 pointsfortran7725 comments

Open articleView on HN

Further reading, for those interested:

Hi-res scans of the manuscript: https://archive.org/details/the-voynich-manuscript_202008/

Declassified NSA seminar (1976): https://www.nsa.gov/portals/75/documents/news-features/decla...

Here is an online folio browser:

https://www.voynich.ninja/browser/default.cfm

Summary:

'Till ~2020, Voynich research was pretty much reserved for nut jobs, and the very occasional academic intent upon career suicide.

In 2020, a senior A List scholar published her meticulous, narrow-niche Voynich research - https://muse.jhu.edu/pub/56/article/754633/pdf - which established that the Manuscript was the (physical) handiwork of 5 different scribes.

Since then, a fair number of serious researchers have done work on the Manuscript. So far, their conclusions are similarly dull and limited. (Though they do rule out many cool categories of nut theories.)

It is sort of interesting that it is a group work. It’s odd that somebody would scribble a whole journal full of random stuff, but of course people in the olden days got bored too. Convincing four other people to scribble with you seems a bit harder, haha.

Now that we have GenAI a whole book of random stuff is cheap, but back then it was expensive (and coordinating a group sounds even more expensive!), so we assume it must've had some meaning to someone, and get nerd-sniped into finding a reading.

(on the other hand: https://www.rand.org/pubs/monograph_reports/MR1418.html . It's amusing to see "errata"!)

Check this out: https://scp-wiki.wikidot.com/

(I had no idea this lore exist but got interested why my daughter suddenly started to talk about `scp`, I was wondering is she wants to copy some files between machines...)

how do they rule out if these five hands come from the same person with long periods of not working on the script? My handwriting can certainly change when I take a three year break

Answered in the article:

In certain places, the script was more cramped or more likely to slant as it crossed the page. She tested this observation by picking a letter that didn’t appear often and tracking it across the manuscript’s pages. Its style, she saw, varied among groups of pages but not within those groups. This suggested that the differences—larger or smaller loops, straighter or curvier crossbars, longer or shorter feet—were the product of different scribes rather than of one scribe writing the same letter in different ways.

To reduce the possibility of selection bias, Davis examined other letters and found that their styles shifted in lockstep with the first letter. After months of analysis, she concluded that even if the Voynich had a single guiding vision, it was the handiwork of five different scribes.

>Since then, a fair number of serious researchers have done work on the Manuscript. So far, their conclusions are similarly dull and limited. (Though they do rule out many cool categories of nut theories.)

Not limited enough to rule some things out. Kind of "We're just guessing, but their guesses are incorrect. We do know that much".

>'Till ~2020, Voynich research was pretty much reserved for nut jobs, and the very occasional academic intent upon career suicide.

I wouldn't say that. Probably the most famous Voynich researcher until modern times was William Friedman, who was a famous cryptographer who led the team who cracked PURPLE (Japan's equivalent to Enigma) in WWII. He wasn't viewed as a crackpot but was and is revered in the field of cryptography.

https://en.wikipedia.org/wiki/William_F._Friedman

Voynich Studies seems like one of those interesting fields where hi-tech meets the humanities. One person is using mathematical cryptanalysis, another has spent their life studying medieval inks, somebody else knows about early gynecology.... Archeology is in a similar place at the moment, because of the advent of ancient DNA. Fun for such different academic communities to have to interact with each other.

http://archive.is/dvwsP

I did my own amateur analysis of it once. My idea was to try and use line breaks to derive some information about how words might be delineated.

On a lot of the pages, the text is highly compressed against the illustrations. If you’re writing English you would find this hard to do, because you would need to insert a line break before a word if it’s too long. I couldn’t find any examples where the text overlapped the illustrations. That suggests to me that words are either very short, or graphemes themselves are words. Or, that the text is a meaningless hoax, because it doesn’t adhere to any rules and can be squashed against the illustrations easily.

Or, it's a scriptio continua like classical greek and latin, and line breaks are permitted in the middle of words.

Its doable http://tom7.org/bovex/

short words would be consistent with the serious analysis that the language has to be tonal, Which rules out all European languages and Mostly leaves meso American or sinitic languages; which would be possible as an extra chapter to il millione, the travel report of Marco Polo, albeit reproduced and illustrated in a European context and style

How good is the accuracy of paleological methods? Have they been tested in scenarios where a ground truth later became available?

To me it looks like in same vein as Codex Seraphinianus which has no language but definitely looks like its telling something.

Anyone know if there’s a nice hardbound version of this? Looks like everything on Amazon is garbage.

I have a copy of the YUP facsimile mentioned in the article and its a thing of beauty.

obligatory xkcd https://xkcd.com/593/

Question: is there any history of something like roleplaying games from that area and period?

I could not comment on paywalled article, but I am wondering if someone finally tried to use AI/ML on the manuscript.

This blog is actively writing about such approaches: https://voynichattacks.wordpress.com/

From the Atlantic article: So little was known about the underlying language—if it was a language—that even artificial intelligence, in its current state, lacked the models to decode it. Good AI requires “massive amounts of data to learn from,” Layfield told me. “We simply don’t have that luxury with the text in the Voynich.”

Reply on news.ycombinator.com