The Voynich Ninja

Torsten: if what Nick says is true, then it might help if you line out a simple step-by-step overview of how the process went from start to finish. I don't mean in detail, just the general steps. That might be necessary to provide clarity. If I'm not mistaken your hypothesis is that an author wanted to fill a manuscript with pseudo-text and you understand the system he used. How did this go from start to finish?

(07-02-2017, 12:52 PM)nickpelling Wrote: You are not allowed to view links. Register or Login to view.Your explanations makes no sense to me precisely because you see the presence of "al" (3002 occurrencees) and "ail" (29 occurrences) in your network as having equal argumentative value. The former occurs 100x more often the latter: and yet you give not a single practical argument as to why there should be disparities of the order of 100x within such small edit distances of common groups.

It is only possible to write one word at at time but there are multiple ways to change a source word. Since there are multiple ways to change a source word this results in a multi dimensional network. "This means that one word is not related to a specific similarly spelled word. In fact, each word is related to all other similarly spelled words" [You are not allowed to view links. Register or Login to view.]. Similar to 'ail' is not only the word 'al' but also the words 'aiil', 'dail', 'kail', 'sail', 'air', 'ais', 'ain' and 'aim' [see You are not allowed to view links. Register or Login to view.].

Lets take a closer look to the network:
                                               aiiin ( 41) daiiin ( 17)
aiil (  1) daiil (  1) aiir ( 23) daiir  ( 23) aiin  (469) daiin  (863)
ail  (  5) dail  (  2) air  ( 74) dair   (106) ain   ( 89) dain   (211)
al   (260) dal   (253) ar   (350) dar    (318) an    (  7) dan    ( 20)

ol   (537) dol   (117) or   (363) dor    ( 73) on    (---) don    (---)
oil  (---) doil  (---) oir  (  2) doir   (---) oin   (  4) doin   (  4)
oiil (---) doiil (---) oiir (  4) doiir  (  8) oiin  ( 33) doiin  ( 19)
                                               oiiin (---) doiiin (  3)

There is no word 'oil' beside 'ail' and even if 'air' occurs 74 times there are only two instances of the word 'oir'. This can only mean that 'al' is usually not used as source word for 'ail'. Otherwise the word 'ol' should lead to at least one instance of 'oil' and 'or' to more instances of 'oir'. Other similar words to 'ail' are 'air' and 'ain' whereas 'ail' is less frequent then 'air' and 'air' is less frequent then 'ain'. For ['dail' - 'dair' - 'dain'], ['aiil' - 'aiir' - 'aiin'], ['oil' - 'oir' - 'oin']  etc. also the word ending in 'l' is less frequent then the word ending in 'r'. In all this cases also the word ending in '-r' is more frequent then the word ending in '-n'. Because of this statistics I would expect words ending in '-il', '-ir' and '-in' as possible source words for the word 'ail'.

We can check this by looking at the five instances of 'ail' in the VMS. They occur at page f105r, f105v, f106v, You are not allowed to view links. Register or Login to view. and f115v. It is noteworthy that all five pages belong to the star section. 

Page f105r:
[attachment=1130]

<f105r.P2.21>    ypair.opaiin.opail
<f105r.P2.24>    daiiils
<f105r.P2.29>    aiildy
<f105r.P2.32>    air
<f105r.P2.33>    ail
On page You are not allowed to view links. Register or Login to view. beside 'ail' three words ending in 'il' exists (underlined red). This is interesting since only 47 words ending in 'il' exists for the whole VMS. The first word 'opail' can be found in line f105r.P2.21 (Note: A possible source word for 'opail' is 'ypair').  On top of 'ail' in line f105r.P2.32 also a word ending in '-air' exists (underlined purple). Therefore we have four possible source words for 'ail' on this page.

<f105v.P.14>     qokaiir ... aiiin.kail
<f105v.P.16>     okair.aiin
<f105v.P.18>     opair.orair
<f105v.P.19>     lkaiir ... ail
<f105v.P.28>     aiil
On page You are not allowed to view links. Register or Login to view. beside 'ail' two words ending in 'il' exists. One of this two words is the word 'aiil' and the other is the word 'kail'. This is stunning since both occur only once. Moreover both words differ in only one glyph with 'ail'. In this way 'kail' is a possible source word for 'ail' and 'ail' itself is a possible source word for 'aiil'.

<f106v.P.5>      aiin ... dair
<f106v.P.7>      ail
On page You are not allowed to view links. Register or Login to view. beside 'ail' also the word 'aildy' exists. But 'ail' is on the top of page You are not allowed to view links. Register or Login to view. and 'aildy' at the bottom. Therefore 'aildy' is for sure not the source word for 'ail'. A possible source word for 'ail' can be found two lines above 'ail'. This word is 'dair'. 

<f108r.P.11>     ar.aiin
<f108r.P.13>     ar.ail
On page You are not allowed to view links. Register or Login to view. we can found the two words 'ar ail' and two lines above 'ar ail' the two words 'ar aiin'. A possible source word for 'ail' is therefore the word 'aiin'.

<f115v.P.7>       air
<f115v.P.11>     ail
On page You are not allowed to view links. Register or Login to view. beside 'ail' also a possible source word 'air' exists some lines ahead. 

This five examples for 'ail' demonstrate that similar words do co-occur within the VMS [see You are not allowed to view links. Register or Login to view.]. Also for 'lkl' it was possible to demonstrate such a result [see my post on You are not allowed to view links. Register or Login to view. of this thread]. This means that the statistics for the VMS and the observations within the text both indicate that similar words are related to each other [see You are not allowed to view links. Register or Login to view.].

Quote:Your autocopying argument requires the presence of a proto-network from which the full network is somehow an extended manifestation (via autocopying): yet you are silent as to what mechanism drove that proto-network. You spend your time talking about your overfeathered chicken but the matter of the missing egg from whence that chicken grew is a debate you prefer to explain away.

Since it makes only sense to talk about the source of a network if you accept that this network does exist I take your chicken and egg example as compliment.

The existence of the script used for the VMS alone indicates that the VMS doesn't include the first drafts the author has made to design all the glyphs used in this script. Moreover it is only logical that it is not possible to copy something from an empty page. To copy the first words for an empty page a second page as source is needed [You are not allowed to view links. Register or Login to view.]. The answer to your question is therefore that the author made some drafts before he started to write the VMS.

(07-02-2017, 01:30 PM)Koen Gh. Wrote: You are not allowed to view links. Register or Login to view.Torsten: if what Nick says is true, then it might help if you line out a simple step-by-step overview of how the process went from start to finish. I don't mean in detail, just the general steps. That might be necessary to provide clarity. If I'm not mistaken your hypothesis is that an author wanted to fill a manuscript with pseudo-text and you understand the system he used. How did this go from start to finish?

It is possible to order the sections of the VMS in a way that two sections with similar frequencies for some key words follow each other. This way a steady development from Currier A to Currier B is revealed [see You are not allowed to view links. Register or Login to view. or You are not allowed to view links. Register or Login to view.].

(07-02-2017, 10:05 PM)Torsten Wrote: You are not allowed to view links. Register or Login to view.It is possible to order the sections of the VMS in a way that two sections with similar frequencies for some key words follow each other. This way a steady development from Currier A to Currier B is revealed [see You are not allowed to view links. Register or Login to view. or You are not allowed to view links. Register or Login to view.].

That is an observation, and a valuable one. But what people don't understand is why this should lead to a conclusion that the text was generated by auto-copying. To make this more convincing, we should understand how the process of auto copying worked.

So... The scribe sits at his desk, pen in hand, staring at a blank page. What happens next?

Can you generate a text the same length as the VMS text using your code and show that it does in fact contain a similar word network to what we see in the VMS?

It's already been pointed out many times that your method would be expected to produce many words which do not in fact occur in the VMS, but I also don't see why it should be expected to reproduce the complete word network (with or without quite a few "extra" words).

In fact the text generated by your app looks quite different from the actual VMS text, and more like what I and others have said it would look like, as I pointed out in the thread about your app a year ago:

You are not allowed to view links. Register or Login to view.

Sam, perhaps it's a good idea to hold a competition based on that: 
who can write the best and shortest program (and publish the source code) that generates the Voynich text.

(08-02-2017, 02:42 PM)Davidsch Wrote: You are not allowed to view links. Register or Login to view.Sam, perhaps it's a good idea to hold a competition based on that: 
who can write the best and shortest program (and publish the source code) that generates the Voynich text.

I've thought that it would be interesting to have a sort of "arms race" where someone creates a code that generates VMS-like text, then people find ways in which the text differs from the actual VMS text, then the code is updated to account for these differences, then more differences are found, then the code is updated again, and so forth.  While I don't think the VMS is procedurally generated gibberish, we could probably learn some things from such an exercise.

Unfortunately, nobody promoting the idea that the text was generated by some procedure seems to be very interested in matching all or even most of the VMS text's actual properties.  Instead people come up with procedures that account for a small number of the VMS text's properties and declare victory.

(08-02-2017, 03:06 PM)Sam G Wrote: You are not allowed to view links. Register or Login to view.Unfortunately, nobody promoting the idea that the text was generated by some procedure seems to be very interested in matching all or even most of the VMS text's actual properties.  Instead people come up with procedures that account for a small number of the VMS text's properties and declare victory.

Yes, exactly, while the question "what can't I explain yet" is just as relevant as "what can I explain".

Davidsch, I like the idea because it would result in a very evidence based approach. But it would require someone with programming skills to take charge of the project.

The problem for me is, that I wrote software to work also the opposite way: take the VMS text and chip away the "finds".
To write a generation program needs a different mindset. I'm not sure if I want to give it a try.

Yes, and it would be a lot of hard work without the guarantee of any neat results. The type of results I'd expect is that aspects x and y are hard to generate in this way. Of course I'd find a result like that rather interesting.