The Voynich Ninja

I thought l/y pairing was already included at least in pfeaster's summary of the curve-line system? I don't remember the history of curve-line systems very well, but I think I remember l/y mentioned somewhere.

As for qo/t/p, this is an interesting and new pairing for me, it appears to match nicely with a recent discussion here, regarding the first q in the MS (in the present order of pages): You are not allowed to view links. Register or Login to view.

(27-01-2025, 12:18 AM)Bluetoes101 Wrote: You are not allowed to view links. Register or Login to view.I'm not sure why this took me this long to notice, but I think there is a good case (if you think CLS has any merit) that EVA: L, is the line equivalent of EVA: Y. 

I agree.

I think that this was also already pointed out in Brian Cham's original document (possibly co-authored with David Jackson): You are not allowed to view links. Register or Login to view.

What it may lead us to is another question.

(27-01-2025, 08:38 AM)ReneZ Wrote: You are not allowed to view links. Register or Login to view.

(27-01-2025, 12:18 AM)Bluetoes101 Wrote: You are not allowed to view links. Register or Login to view.I'm not sure why this took me this long to notice, but I think there is a good case (if you think CLS has any merit) that EVA: L, is the line equivalent of EVA: Y. 

I agree.

I think that this was also already pointed out in Brian Cham's original document (possibly co-authored with David Jackson): You are not allowed to view links. Register or Login to view.

What it may lead us to is another question.

Thank you for the link, I tried to find this article earlier today and couldn't remember the authors' names or the title.
Now I see that they actually highlighted the possibility that l is the minim version of y in the introduction below.

[attachment=9877]

Thanks I must have glossed over that part

In my opinion it is not correct that "l" is "line" only. My theory is that the truth is on the paper. It is a line with a flick "\c" or as I show "K" (Line>Curve). 

I won't waffle with examples, but.. a couple.. 

1. Lines don't follow "l". Curves do follow "l".
   
2. "ol" and "lo" are as prevalent in the text as "ai" ia", one does not conform, the other is foundational to the system.
   
3. "yoy" shows up zero times though it is conforming (ccc). "lol" shows up 166 times and is non-conforming (\c\). 
   

A Voynichese webpage search of "il" and "li" quickly shows my general theory, that "it starts as a line but does not also end as a line".

I'll hopefully have something substantial to show soon. I heard someone quote Currier as saying "it doesn't mean anything, but there it is" in regards to these curve and line patterns, and I think I probably agree. I'm not sure it will ever really mean anything, but its there so here I am poking and prodding it to see if it bites.. but you are certainly in no danger of a "I've solved it!" whatever the results 

Perhaps that is a limitation to the importance of 'what follows what' rather than to the classification of each character...

I agree. However, I also think "you can't have your cake and eat it too" in regards to considering a system which shows typical Voynich word construction. 
If we call "lines", "lines" as in the base line glyph construction method, the system does not work. 

I've gone fairly off in the weeds past the CLS now, I'm debating if it is correct etiquette to give it a new name or not.. but anyway,

Here is "f6r", I have highlighted which words my system shows as (currently) non-conforming.
I have crossed out (underlined where overlap) each word I believe fail original CLS.

In my view it has worked pretty well to point out 4 "words" where 3 are unique and 1 appears twice. 

This is a cherry picked page for a good example, however my config currently scores 10 full Currier A pages of text at 93.9% conformance.
If I take out my meddling. 
y+l = null, d=K, revert to y=c, l=k, d=c

("l" working as null or K only makes about 0.4% difference, I just had it as that as maybe "y" and "l" should be given the same function.. don't know, but anyway "y" and "d" mapping accounts for the vast majority of the conformance gap)  
81.3% conformance.

There is still every chance in the world I have made mistakes at this stage, which is why I'm not sharing everything yet.. If I have I didn't mean to and will correct.




Line <f6r.1,@P0>:
Line <f6r.2,+P0>:
Line <f6r.3,+P0>: foeear -> [K>c]c>K, cheoeees -> cc[c>c]ccc
Line <f6r.4,+P0>:
Line <f6r.5,+P0>:
Line <f6r.6,+P0>:
Line <f6r.7,+P0>:
Line <f6r.8,+P0>:
Line <f6r.9,+P0>: skaiiodar -> [cK]>[\\>K]>K
Line <f6r.10,+P0>:
Line <f6r.11,+P0>:
Line <f6r.12,+P0>:
Line <f6r.13,+P0>: qocthol -> [K>Q]>X
Line <f6r.14,+P0>:

(28-01-2025, 04:02 PM)Bluetoes101 Wrote: You are not allowed to view links. Register or Login to view.I agree. However, I also think "you can't have your cake and eat it too" in regards to considering a system which shows typical Voynich word construction. 
If we call "lines", "lines" as in the base line glyph construction method, the system does not work. 

I've gone fairly off in the weeds past the CLS now, I'm debating if it is correct etiquette to give it a new name or not.. but anyway,

Here is "f6r", I have highlighted which words my system shows as (currently) non-conforming.
I have crossed out (underlined where overlap) each word I believe fail original CLS.

In my view it has worked pretty well to point out 4 "words" where 3 are unique and 1 appears twice. 

As far as I see, what you create is very similar to a "loop" grammar showing possible transitions from character to character. The problem with grammars is the more rules and classes you create, the more flexible you make a grammar, the less specific and less useful it gets. It's very promising that your system correctly predicts (shows as conforming) more words than the original CLS, but could you also estimate whether this happens at the cost of greatly increasing the total number of possible conforming sequences (unattested) that could be produced?

Because the upper extreme of this system is "any character can follow any character", which obviously covers anything and tells us nothing.

There was another thread where @Mauro introduced several very interesting metrics that can be used to compare the efficiency of grammars:

(03-01-2025, 11:28 AM)Mauro Wrote: You are not allowed to view links. Register or Login to view.After introducing the concept of ‘looping’ slot grammars, a generalization of standard grammars, I show how any grammar can be used in a distinctive lossless data compression algorithm to generate a compressed Voynich Manuscript text. This allows the definition of a new metric free from fundamental flaws: Nbits, the total number of bits needed to store the compressed text. I then compare published state-of-the-art grammars and the newly introduced LOOP-L grammar class using the Nbits metric.

Would be great to be able to compare your system with the original CLS and other grammars not based on curve-line system, to see which ones are mathematically superior.

(28-01-2025, 05:03 PM)oshfdk Wrote: You are not allowed to view links. Register or Login to view.

(28-01-2025, 04:02 PM)Bluetoes101 Wrote: You are not allowed to view links. Register or Login to view.I agree. However, I also think "you can't have your cake and eat it too" in regards to considering a system which shows typical Voynich word construction. 
If we call "lines", "lines" as in the base line glyph construction method, the system does not work. 

I've gone fairly off in the weeds past the CLS now, I'm debating if it is correct etiquette to give it a new name or not.. but anyway,

Here is "f6r", I have highlighted which words my system shows as (currently) non-conforming.
I have crossed out (underlined where overlap) each word I believe fail original CLS.

In my view it has worked pretty well to point out 4 "words" where 3 are unique and 1 appears twice. 

As far as I see, what you create is very similar to a "loop" grammar showing possible transitions from character to character. The problem with grammars is the more rules and classes you create, the more flexible you make a grammar, the less specific and less useful it gets. It's very promising that your system correctly predicts (shows as conforming) more words than the original CLS, but could you also estimate whether this happens at the cost of greatly increasing the total number of possible conforming sequences (unattested) that could be produced?

Because the upper extreme of this system is "any character can follow any character", which obviously covers anything and tells us nothing.

There was another thread where @Mauro introduced several very interesting metrics that can be used to compare the efficiency of grammars:

(03-01-2025, 11:28 AM)Mauro Wrote: You are not allowed to view links. Register or Login to view.After introducing the concept of ‘looping’ slot grammars, a generalization of standard grammars, I show how any grammar can be used in a distinctive lossless data compression algorithm to generate a compressed Voynich Manuscript text. This allows the definition of a new metric free from fundamental flaws: Nbits, the total number of bits needed to store the compressed text. I then compare published state-of-the-art grammars and the newly introduced LOOP-L grammar class using the Nbits metric.

Would be great to be able to compare your system with the original CLS and other grammars not based on curve-line system, to see which ones are mathematically superior.

It's surely interesting.

I have first to read an understand exactly how Blutoes101's curve-line system works, then see if my software can manage it, hopefully without too much additional coding. I'll let you know in a few days.

It's an interesting question, and I don't know the answer, but I believe;

CLS
Conforming sequence = 4
Non-Conforming = 4

My stuff
Conforming sequence = 8 
(EDIT: Actually this isn't right, I had only considered sequences which could be non-conforming but instead conformed. I'll figure out a proper list to publish with the completed thing once done)
Non-Conforming = 13


I'd have to read more about loop stuff. I have watched this a couple times if this is the same thing? 
You are not allowed to view links. Register or Login to view.

(if it is) What I took from the talk was that each glyph has a % probability of going to the next, and with some cases this % probability causes a loop. 
I think what I am showing is a bit different, the main difference is that it has no preference or perceived likelihood.

Lets say, I have a bunch of letters for whatever reason - l, k, i, r. Now, in accordance with the system they must be written like "olokair".. you could have also written "lokair", "alokair", ylokair.. so on and so on. Each iteration is pretty samey/uniform but you had choices within the system. It doesn't care what is usual, just conforming. 

See, I'm hoping I missed the "After introducing the concept of ‘looping’ slot grammars" and that is why it sounds like Mauro is casting spells at me  . I may need to backtrack some to get a foothold into that, but I certainly will, just learning what an Nbit and LOOP-L (rappers?) is will help demystify things greatly.

(28-01-2025, 11:11 PM)Mauro Wrote: You are not allowed to view links. Register or Login to view.It's surely interesting.

I have first to read an understand exactly how Blutoes101's curve-line system works, then see if my software can manage it, hopefully without too much additional coding. I'll let you know in a few days.

I'm not quite ready just yet. Probably several days away so definitely no rush. This threads a little crazy as it was a working progress and jumble of working (..non working) ideas. I can share the code when its ready though and send you whatever you need if you send me a message on here.