Red Herrings are sometimes useful
i_want_links_damit > 6 hours ago
I am examining the Voynich Manuscript as part of a major project. I have a number of suspicions and leads on it that have been deferred as part of the project is to make a tool with a wider scope. In spite of that the system has yielded many insights. One of them is not good. I also realise that if this solution solve it, that is a problem, it ends the adventure and as such so far all efforts have deliberately stopped short of getting too close for now. The one that is not good has found a strong candidate for a system of generating text from other texts but it's only one example and might be spurious despite the probabilities checking out just about. The approach here is that even if generative it must be derivative to some degree from something so it searches for that which covers all other possibilities as well.
The Voynich Manuscript is perfect for this. It's a bit weird. It's kind of average. The problem isn't that all possibilities are ruled out leaving a mystery. The problem is that you can't easily rule them out. It exists at a strange kind of junction. Sometimes I look at the evidence I have found for it being fake which I have not made an effort as of yet to pursue holding myself back and consider the chance of it. It works for lines but not well for single word labels so that needs to be examined. Prior to that I consider the background. People making fakes like this is not unknown for multiple reasons. It's a pattern in which a King would pay people, often two on purpose to compete, to go and collect new books for their library. There is a good chance at least one would cheat and make it up.
Around this time is also when people started working on well, everything, including, yes, DRM, this could be ancient DRM. I'm sorry, I have bad news for you. I recently put this into a Bayesian filter and can confirm that the Voynich Manuscript is in fact ancient spam. The same way you open an email and see it's spam, back then you would go to the library, open a book and see it's spam. What's more at least 5% of it is a virus so don't try to convert it into machine code and run it on your CPU. At this time there were people working on ways to prevent people copying their manuscripts. Creating something like this to tie people up is compatible with the many options of the era. When I consider counters to this such as the images I then also consider that someone making a forgery might consider this and go even more out of their way to make it look like that than something authentic. This is the too authentic to be true test. The book routinely fails the true test derived from UFO studies. There is a barrier that always appears before you can get too close. That is, it always leads you on. It never actually confirms anything. There is a distinct line and drop off that is worrying. You see the same in old pictures of Big Foot, the Lock Ness Monster, etc. There is this point where the resolution always just plummets, bottoms out rather than following the normal continuous curve and this book does that.
I have found a huge number of correlations as anyone else will and its easy to do but where do you move on from there? If you look at the alphabet you see recognisable elements. The approach I'm taking and the tool I'm working on is at phase one of detecting correlations both manually and automatically in a holistic approach. The second phase which has not been properly embarked upon is dating it. Lets say it has that ribbon symbol which is sometimes also a 4 or a 5 in more recent alphabets. The problem is that doesn't tell you must about it. Loads of text have that. The system I am working on does many things and one of them is to detect as many correlations as possible then roughly order or date them.
This is a kind of loose exclusion principle. The more widespread, that is, found elsewhere for example, the less specific. The ideal is to narrow down on as specific traits as close in the timeline as possible to it. Even a very loose preliminary manual application of this methodology has yielded promising results but I shall keep that under my hat. I have enumerated certain characteristics of the text that narrows it into a box though the permutations are still quite extreme.
At the moment this system is in testing and only the first proof of concept version. It has already shown some abilities. This includes detecting ancient language patterns by accident such as Indi-European, ancient human transits as far as to the Americas, ancient conduits and a strange ancient alternative to GMT with Greece and Egypt along the meridian instead. You pick up on many things like this in it including natural barriers when plotted on a map. In another case it display the ability to detect different character sets within a text. Numbers are quite easy. If the VM uses numbers then either they are used in a specific way and sparingly or they are like Roman numerals in most cases and using a letter.
I think even without using more than test samples for the tool in early versions a picture has emerged. In particular there are matches for Enochian on sight without even putting it into the tool. When putting in preliminary sequences there is a certain kind of match along some dimensions. Prior to the tool just looking at Enochian rang a lot of bells as being similar. I immediately felt like the VM seems like a precursor to it.
I am likely both right and wrong though I did foresee my error. Enochian or Adamic did not come out of nowhere. When you read things that give you the impression someone just invented it out of the blue that is incorrect. It's all inspired. When you look at Enochian it's clear that it is some garbled mumbo jumbo based on things like treating prior ciphers differently as to intended. Even prior to it things weren't separated. Actual pharmacology with real ingredients that worked was not considered different to magic. There was a split after the era the VM was likely to have been written. Magic and science among other things split. That is quality control, the wheat from the chaff. Alchemy and chemistry, astrology from astronomy. The Voynich Manuscript seems to be from a time and place where these are more fused. Today you go to your hippy friend's basement and there's a Ouija board. You visit your other friend's place who is a scientist and there is a microscope. In this era it wasn't always like that. You visit a single friend and both are in the basement. They are both the scientist and the crystal worshiping freak. There is no well maintained separation.
I don't think that the VM is actually Enochian but in my analysis it has characteristics that seem to share a common ancestor that's quite recent. Enochian if you look at it takes existing functional mechanisms and makes them creative. It is based on the ciphers of the time but does weird things like inverting them. It's clearly the product in part of people asking funny questions like what happens if you decrypt a test already decrypted as well as people trying to interpret the result of the cipher on face value then integrating the concept with other magical notions.
The system I am working on is holistic but a feature of it is to show you specifically thing such as for this correlation which elements contribute the most to it. The problem with a lot of statistical systems is that they just give an output and that's it. An aggregate. The individuals removed. The system I am working on is different. It minimises things like the use of libraries and is all hand written to be able to do things like pull out the pieces of text that cause the correlation or whatever statistical pattern it is for manual review.
The point being is that I would not entirely dismiss Enochian. It's useful at least to get an idea of what was going on at the time and how creative people could be. Not only that but it likely has correlations if the language was generated or uses a cipher through shared ancestry. You can clearly see an ancestral precedent to Enochian in things such as Cistercian symbols. It is useful to include Enochian text and wordlists so that if it matches you can check to find out why. I did this with it matching Pacific Islands, Zulu, Vietnamese and Matan really well which so far has an explanation for increased rates of coincidence. Even if a red herring it told me something about the way the language is mutated.
There is something about Fijian that's quite interesting where it seems to have quite a restricted character set which is used heavily against itself into a box. Other characters exist with diacritics but it seems in many texts these are either stripped out or people just don't bother to use them. It's a Latinised foreign language. Although it might be different and for other reasons VM might have some boxing as well. There are signs of two sets of types of symbols but the signal is not yet clear enough. It's not as obvious as numbers normally are but like I said it matches some texts using Roman numerals quite well.