(15-05-2026, 11:34 PM)ReneZ wrote: You are not allowed to view links. Register or Login to view.(15-05-2026, 07:08 PM)nablator wrote: You are not allowed to view links. Register or Login to view.(15-05-2026, 07:02 PM)oshfdk wrote: You are not allowed to view links. Register or Login to view.Something looks off with the placement of EVA i and e in this line.
Also ccth... mor.
Bigrams are not enough, you need trigrams statistics to generate plausible Voynichese.
It would work better, but it becomes a bit of a brute force method.
In both cases (bigrams and trigrams) a separate tweak will be necessary to get realistic word spacing.
And you need separate rules for the various Currier languages and dialects...
What I'm seeing from all of those bigrams and trigram outputs, the zipf curve, the word length distribution... it's all emergent. None of my code forces anything. It is all probabalistic but it didn't need to be. All of those probabilities are to 'simulate' the method the scribe likely had in their head. And, if you remember my work on the bigram ED, this all ties into it. I did say I was working on a method to reconstruct the Voynich. I 'think' I have found it. Scribe 2 used the same system as Scribe 1, but their method was different enough that my modelling just doesn't look realistic. I still have a bit of work to do there. It's the same 'mechanism' just a different way of using it. Scribe one however, was much simpler in the way he worked and I believe even Currier saw that.