Michael Greshko
banner
michaelgreshko.bsky.social
Michael Greshko
@michaelgreshko.bsky.social
Associate online news editor @Science. Freelance contributor to NYT, SciAm, WaPo, etc., and author of the Deviations newsletter. Former staff writer at National Geographic. Signal: mgreshko.01 https://linktr.ee/michaelgreshko
It’s *very* hard to reconcile the statistics of Voynichese with those of known written languages. But treating those properties as constraints, it’s possible to find manual cipher families that jointly evade those constraints. Doesn’t mean that’s how the Voynich was made, but good to test.
November 29, 2025 at 6:18 PM
Thank you!
November 29, 2025 at 2:48 AM
Thank you, Jürgen! Your work played a really important role in inspiring this cipher.
November 28, 2025 at 3:03 PM
Thank you!
November 27, 2025 at 4:23 PM
This paper was a labor of love, scraped together from scraps of spare time over the past year. I have lots of people to thank. Of the folks I know are on Bluesky, a huge thank-you to @lisafdavis.bsky.social @chirila.bsky.social and @spinfocl.bsky.social!
November 27, 2025 at 3:42 PM
If you want to watch a video describing the Naibbe cipher, I gave a talk on it back in August:

www.youtube.com/watch?v=ByAR...
Voynich Manuscript Day 2025 Full Conference
YouTube video by Voynich Talk
www.youtube.com
November 27, 2025 at 3:42 PM
The Naibbe cipher is more elaborate than known 15th-century ciphers, but its basic design principles were used in 15th-century Italian homophonic ciphers. And, again, it's totally doable by hand while also mimicking the VMS surprisingly well.
November 27, 2025 at 3:42 PM
The cipher—fully described in the paper—basically takes a Latin text, breaks it up into 1- and 2-letter chunks, and then disguises each chunk as a Voynichese word using a series of 6 tables on a letter-by-letter basis.

How do you choose the tables? Playing cards!
November 27, 2025 at 3:42 PM
This began for me over a year ago as simply a personal art project. But the more I tested an early cipher I had created, the more I began to see a path toward an "eerily good" cipher that could mimic lots of properties of the VMS at once. Naibbe represents that cipher.
November 27, 2025 at 3:42 PM
This started with a simple question:
Is it possible to design a substitution cipher—the type of cipher that would have been found in Europe at the time the VMS was created in the early 1400s—that can mimic the statistical weirdness of "Voynichese"?
Turns out, the answer is "yes."
November 27, 2025 at 3:42 PM
Scholars have debated whether the VMS is:
• an unknown language
• meaningless gibberish
• an encrypted language like Latin
Naibbe is a cipher, fully doable by hand with 15th-century materials, that can encrypt Latin as text that behaves a lot like the weird "language" of the VMS.
November 27, 2025 at 3:42 PM
For context, I have been looking through the Epstein emails today because Epstein was a prolific—and notorious—funder of the sciences. See my colleague @policyhound.bsky.social's 2019 article for more: www.science.org/content/arti...
What kind of researcher did sex offender Jeffrey Epstein like to fund? He told Science before he died
Felon said he liked “rebels” and “rarefied air” in 2017 interview
www.science.org
November 13, 2025 at 3:00 AM
The email is HOUSE_OVERSIGHT_019871.txt, viewable here: drive.google.com/file/d/1i2EL...
001 - Google Drive
drive.google.com
November 13, 2025 at 3:00 AM
*typo: magic THROUGH the millennia. A huge thanks to the book's fact-checker and proofreader, too!
September 2, 2025 at 7:22 PM
This book was a massive lift. Huge thank-you to my coauthors Nina Strochlic and Pat Daniels and to our incredible editor @mayamyersbooks.bsky.social, as well as to the whole team at WonderLab Group and National Geographic, with a special shoutout to Kate Olesin and Jennifer Emmett.
September 2, 2025 at 7:15 PM