Discussion of "A possible generating algorithm of the Voynich manuscript"

[ReneZ]

I think that the results for mod1 and mod2 tell us more than the encryption of Machiavelli: your two methods autonomously generate a network of similar words, as Timm's method does.
You wrote:

pliny_mod2 is interesting, because due to its structure it allows to predict the frequencies of the various edit distances. For all words that are not the same (edit distance 0) the ratio should be 20 : 5 : 1 for edit distances 1 , 2 , 3 , and smaller again for 4.

This seems a promising approach. I was thinking that it might be the case to apply the method to Machiavelli (or to a different text with a more Voynich-like TTR than Pliny), but I believe these ratios are not optimal. The histogram for Voynichese looks more like 15:10:5. I wonder how difficult it would be to tune the method to get results closer to Voynichese?

I can give that a go.

Note that the mod1 and mod2 approach both assign numbers to words, and the number is increased by 1 whenever there is a new word. This means that most of the time, the closest recent word in terms of edit distance is either 0 or the word with a numeral that is 1 less.

The encoding for mod1 is using Roman numerals, so the minimal change can have edit distances of 1, 2 or more, and could be predicted. This is easier for the mod2, which is more regular. It uses 5-'digit' words, and one could call both this and the Roman numerals 'high-endian'. That is: minimal changes are happening on the right-hand side of the 'word'. 

It is an interesting question whether the Voynich words could form some enumeration system. It seems doubtful, because minimal changes seem to happen anywhere in the word. However, a closer study of the word structure might still bring further insights.

Note that edit distance depends on the representation (transliteration). The results for the same text will be different depending on whether it is given in Eva, Cuva, FSG etc.

PS: looking at the _mod2 words again, I wonder if my prediction of 20:5:1 was right, and probably it should have been 20:4:1 . In that case, the percentages would become 65, 13 and 3 which doesn't fit quite as well, and I need to think about that a bit.