A Use for Big Data: Cracking the Voynich Manuscript

I used to dabble in the creation of naturalistic languages. As soon as you have a syllable structure you can start writing things that look like language but aren't.

Once you add grammar you get things that are language, but are still gibberish until you write a dictionary.

The most notable example of this would be Tolkien's Middle Earth. He created tens of languages to various degrees of completeness. Qenya, for instance, is so complete you can write books and poetry in it. It's even taught at Oxford. That's elvish.

But dark elvish, the writing on The One Ring, has only been constructed to the extent that he could write that inscription. Nothing else of it exists.

Or you can even go for a much simpler example - a markov chain. "trained" on an English corpus it will produce texts that look exactly like English. But because it doesn't adhere to any grammatical rules it produces nothing but gibberish.

This guy managed to produce the text with the same characteristics like observed in the Voynich manuscript by using "an encoding device called the Cardan Grille, first described in 1550 by Girolamo Cardano."

http://archive.wired.com/wired/archive/12.09/rugg.html

The largest example I can think of is Codex Seraphinianus, which is clearly inspired by the Voynich Manuscript.

It's fairly obviously gibberish; so much so that you can tell just by looking. But the author wasn't really trying to make it have statistical characteristics that resemble natural language.

Just point your web browser to foxnews.com, they've been doing it for years.

I think people are learning Betteridge's law of headlines [0] so questioning title constructions are more obviously link-bait.

[0] http://en.wikipedia.org/wiki/Betteridge%27s_law_of_headlines)

Yeah, it's kinda linkbaity and probably made that way because either the author / editor doesn't understand big data or thinks it's just another media buzzword, or to actively try and draw readers in that take offense to the title.

And book is hardly "big data"