The Voynich Ninja

(30-08-2019, 09:43 PM)-JKP- Wrote: You are not allowed to view links. Register or Login to view.Hyde & Rugg's approach and Torsten Timm's approach are almost completely opposite. It worries me when I see people lump them together.



Hyde & Rugg came up with a grille theory (I don't know whose idea it was originally, but the example on the Web was posted under both their names) and applied it to VMS text. They did NOT really study the text.

I'm sorry, but I don't agree. Gordon Rugg based his approach on the results of Jorge Stolfi's analysis of the text, namely the first installment where words seem to be built up of three parts: prefix, stem, suffix.

(This is different from the core-mantle-crust approach that came later.)

I'm not sure what you are disagreeing with, René.

I wasn't talking about where they got their theory (whether it was theirs or someone else's wasn't the point of my post).

Let me see if I can word this better...


What I am saying is that it's crystal clear from their demonstrated example that Hyde & Rugg didn't look at the structure of the text. Whether they got the prefix/base/suffix idea from someone else or came up with it themselves doesn't matter. They adopted it, so presumably they accepted this idea, their example is based on it, they didn't question it, and it's WRONG. Which means they didn't study the text.


So here is how Timm's approach differs from Hyde & Rugg's...

Timm has NOTICED that there are proximity patterns, not only in tokens, but in folios. They're there. I see them. Who else talks about them besides Timm? You can count them on about three fingers.

Most of the "solvers" are way too focused on substituting characters to look at the big picture OR the middle picture. The patterns that Hyde & Rugg missed and that Timm is describing and has mentioned are in the middle picture and I'll be darned if I can think of very many people who have even alluded to them other than Nick. These are extremely important patterns. Hyde and Rugg brushed them off (I can't remember their exact wording but I copied it down because it illustrated not only a lack of research, but a lack of understanding of the textual structure.).


I really don't care whether Timm's conclusions are right or wrong. I'm not even particularly worried if his theory is right or wrong—he is investigating this in ways that further understanding of the text. The Hyde & Rugg method does nothing to further our understanding the text because they are parsing the text incorrectly. They weren't even ready to apply the grille idea. They failed at Step 1.

So... while Hyde & Rugg's idea (which treats the VMS text as filler) and Torsten Timm's autocopying idea might seem superficially similar, they are not the same thing at all. Hyde & Rugg's idea starts and ends with meaningless text. However, autocopied text is not the same as grille text. It does not necessarily have to be meaningless—it's possible to conceive of autocopying methods that encode information (in other words, it might be nonsense text, but doesn't necessarily have to be).

I'll give a simple example. Whatever is different between token1 and token2 and token3 encodes a letter or two or three. You only read the differences, not the similarities. Thus, the next token is copied from the previous one with small differences, but the DIFFERENCES are meaningful. I'm not saying this is how Voynichese is structured, I just want to make it clear that autocopied text isn't necessarily nonsense text. The output from the Hyde & Rugg is nonsense text (and really doesn't match Voynichese patterns at all).

Even if Voynichese turned out to be nonsense text, we have a better chance of discerning it by looking at the text as Timm is doing, than we do by diismissing it with a flawed grille theory.

Timm has been on the hot seat for quite a while now. I've said numerous times I don't agree with all of it (and Nick neatly put his finger on one of my main objections), but he is studying the text and he's done a good job of describing some aspects of it that no one else has written about.

That's why I don't like to see Hyde & Rugg's idea and Timm's research being lumped together. It's insulting to Torsten Timm regardless of whether he has a perfect description of the text, regardless of whether the autocopying idea is right or wrong, regardless of whether Voynichese is meaningless or meaningful.

Getting data and interpreting data are related but different processes and the latter is much harder. It's possible to get the first part right and not draw the right conclusions. Sometimes it takes centuries for data to be interpreted correctly. In the meantime at least he's trying to describe the patterns that most people don't even notice.

I am inclined to the view that there is an overestimate of the useful of TTR in this thread. As already stated it would be expected on the face of it that there would be a higher proportion of unique and less common words amongst labels than amongst sentence text, so the TTR figure definitely has some use. However the TTR figure is a very blunt tool to tell us anything about labels, but we shouldn't be too reliant on this figure.

Davidsch says:

"Yes, this is almost the same as what I wrote, but I would like to add for everybody the observation that it is not necessary that the VMS-labels are labels as you define them by: "nouns, adjectives or also numbers". The labels could also be just a normal piece of text. For example: here......................is..........................she:...........the.....person..........that...........I gave...........the red rose. Where the text is thrown across the page with images."

I would contest that as I have said elsewhere knowing how to string labels together to make sentence text in practice is in many instances either near impossible or very difficult. This is particularly clear in many instances on the 9 Rosette folio. For the author devising a sensible way to string labels together sequentially and that reasonably being about to read them back seems to me near impossible.

(31-08-2019, 10:44 AM)Mark Knowles Wrote: You are not allowed to view links. Register or Login to view.I am inclined to the view that there is an overestimate of the useful of TTR in this thread. As already stated it would be expected on the face of it that there would be a higher proportion of unique and less common words amongst labels than amongst sentence text, so the TTR figure definitely has some use. However the TTR figure is a very blunt tool to tell us anything about labels, but we shouldn't be too reliant on this figure.

I know Koen can speak for himself, but I see TTR as part of a long tradition of computational attacks, and computational attacks rarely provide answers on their own (nor are intended to). They typically are used to build up a picture that gets filled in, over time, like a jigsaw puzzle.

TTR has value as one of the puzzle pieces. I doubt if Koen ever expected it to be the whole puzzle.

RobGea says:

"Quick list of ways a single label could apply to multiple things:

Attributes / properties
Relationships / mappings
Numeric / quantities
Multiple meanings
Multiple encryptions
Multiple languages
Multiple sounds/Homonymy
Categories
Metadata
Nonsense"

I think this is a good list.

I will take them one by one.

1) It is hard to imagine that a star, a nymph, a pipe, a plant and a rosette feature could have a common property unless the property is so generic that it is pretty superfluous to in practice by meaningles i.e. properties like "good", "bad", "beautiful". This also doesn't explain the similar spellings of repeated words.

2) If we have a relationship mapping such as "figure 2" for repeated labels then we would expect that that text to be referred to in the accompanying text on or near the page with the label on. Alternatively the text would have to occur in a glossary such as the large body of text at the end of the manuscript. This possibilities fall apart on expectation.

3) As far as these repeated words being numbers then one would seem to argue that the words of the format "ok..." or "ot..." would constitute numbers. But what could these numbers mean? If a lone star or nymph etc. has an attached number then how are we to interpret them.

4) I would think when talking about multiple meanings then this is reference to abbreviations or different interpretations of the same word. In practice when the author is reading the text back I would think it often very hard for the author to divine what the intended meaning was as labels tend to be unique plus specific and if the author was already able to identify the item then the label would be unnecessary. Most importantly the very similar spelling of the repeated words does not fit well with this hypothesis.

4) If we have multiple encryptions then presumably the idea is that each page of set of pages has a different encryption key. Where do we find this key? Again most importantly why do the repeated words have similar spellings then, this does not seem consistent.

5) With multiple languages again we have the problem with repeated label words having similar spellings.

6 & 7) Regarding these two I will look up what is intended precisely by these elsewhere in the comments.

8) This is the possibility that makes most sense to me. However I think this only applies to the repeated label words with similar spellings. I believe there is no reason to expect the other label words to be nonsense.

---

JKP: My point is there seems to be a little more talk of TTR on this thread than is really justified given its limitations.

6) On the subject of categories I guess that something like "plant" or "nymph" or "star" is what is being thought of here. This wouldn't make much sense due to the repeated label words across many different label sections and of course as always the similarity of the spellings.

7) On the question of metadata referred to here. I must confess I am not quite sure what is meant here. I can only guess that this is a reference to some other kind of properties/attributes here. Someone could elaborate here if they wish. I understand what "metadata" is, but don't understand what is being thought of in this context.

Again to reiterate there seems to be the assumption that there are two possibilities that either:

A) All the label text is meaningless filter nonsense.

or

B) All the label text is meaningful and has a specific interpretation.

I have made clear that I favour the following alternative.

C) A large proportion of the label text is meaningless filler nonsense and a large proportion of the label text is meaningful text with a specific meaning and interpretation. This seems to me the simplest explanation that fits the observations.

TTR is no golden bullet, but like JKP says, nothing is. Hammers are great but if you have only a hammer you can't build a house. 

TTR is more like a sledgehammer. If you need a big hole in a wall it will do, but you won't use it to hang a picture frame or to paint a door.

So like all tools, TTR is useful if you understand what it can and cannot do. And like so many other tests, it is indicative rather than absolute. 

That said, the fact that the VM labels have a higher TTR than that of any non-label text, VM or otherwise, is a strong indication and it was clever of Rene to suggest this test. It doesn't prove that these are True Labels, but it is still something that has to be taken into account.