The key aspect of my cipher at the moment is not that it "looks" like Voynichese (although this would certainly be desirable), but that it reproduces all the characteristics of Voynichese well: low entropy, predictability of characters and even words, and similarity to LAAFU.
Why might it seem unsuccessful? Because it's still too simple compared to Voynichese, and it only provides a limited set of Voynichese words. It doesn't have to be like that at the moment. You might remember the Naibbe cipher, which creates Voynich-like texts. However, the key is that it simply takes Voynichese "morphemes" like aiin and eor and arranges them in a specific way. This is a completely different principle.
(31-07-2026, 09:47 AM)ololololo Wrote: You are not allowed to view links. Register or Login to view.There's literally nothing for statistical tests yet, and we need to improve the algorithm.
I don't understand how you are going to improve the algorithm if you are not willing to test it? From my point of view you are posting different versions of the algorithm with various tweaks, but to me they look essentially the same, I see no improvement in the generated Voynichese. Which brings it back to the question of what you are trying to achieve and how you measure it.
(31-07-2026, 10:13 AM)oshfdk Wrote: You are not allowed to view links. Register or Login to view.I don't understand how you are going to improve the algorithm if you are not willing to test it?
I'm referring to Python code.
(31-07-2026, 10:13 AM)oshfdk Wrote: You are not allowed to view links. Register or Login to view.Which brings it back to the question of what you are trying to achieve and how you measure it.
Well... Let's say I want to test the capabilities of this cipher and how close it can be to the reference Voynichese. Okay?
(31-07-2026, 10:19 AM)ololololo Wrote: You are not allowed to view links. Register or Login to view.Well... Let's say I want to test the capabilities of this cipher and how close it can be to the reference Voynichese.
How do you measure how close it is?
(31-07-2026, 11:10 AM)oshfdk Wrote: You are not allowed to view links. Register or Login to view. (31-07-2026, 10:19 AM)ololololo Wrote: You are not allowed to view links. Register or Login to view.Well... Let's say I want to test the capabilities of this cipher and how close it can be to the reference Voynichese.
How do you measure how close it is?
By the variety and correspondence of words. Whether the words are similar to the Voinish words and whether the text is sufficiently varied.
I don't know if this will satisfy you...
(31-07-2026, 09:57 AM)ololololo Wrote: You are not allowed to view links. Register or Login to view.The key aspect of my cipher at the moment is ... that it reproduces all the characteristics of Voynichese well: low entropy, predictability of characters and even words, and similarity to LAAFU.
What is the value of entropy and the conditional entropy (predictability) of your cipher?
(31-07-2026, 11:30 AM)oshfdk Wrote: You are not allowed to view links. Register or Login to view. (31-07-2026, 09:57 AM)ololololo Wrote: You are not allowed to view links. Register or Login to view.The key aspect of my cipher at the moment is ... that it reproduces all the characteristics of Voynichese well: low entropy, predictability of characters and even words, and similarity to LAAFU.
What is the value of entropy and the conditional entropy (predictability) of your cipher?
I assume that the predictability is high due to the positional logic of Roman numerals (for example, only I can follow V).
(31-07-2026, 11:50 AM)ololololo Wrote: You are not allowed to view links. Register or Login to view.I assume that the predictability is high due to the positional logic of Roman numerals (for example, only I can follow V).
I think we are lost in translation again. What is the number (like 2.785 or 3.67) of the entropy and the conditional entropy of your cipher?
(31-07-2026, 12:02 PM)oshfdk Wrote: You are not allowed to view links. Register or Login to view. (31-07-2026, 11:50 AM)ololololo Wrote: You are not allowed to view links. Register or Login to view.I assume that the predictability is high due to the positional logic of Roman numerals (for example, only I can follow V).
I think we are lost in translation again. What is the number (like 2.785 or 3.67) of the entropy and the conditional entropy of your cipher?
After rechecking all my encryptions from this thread (without nulls) using Gemini, he gave a result. He said that h1 = 3.4637 and h2 = 1.8224 (characters); h1 = 4.915, h2 = 2.8214 (words).
By the way, I asked if the text follows a binomial distribution. The average word length is 2.54, and the variance is 1.46. The word length is suitable, but the letter frequency is not.
I tossed a coin and it came up heads, which means the numeric cipher theory is wrong. I think my method is no less reliable than yours