How woud an attack work? Spiral text as Rev. 6:14.

[Koen G]

Do I see it correctly that any "known-plaintext attack" cannot work because in the EVA transcription the assignment of the letters is "arbitrary" ?

This is precisely the kind of thing I would like to get a better feeling for during this exercise. I personally see two possible options:

1) No EVA - so also no computational attack: take the actual image of the Voynichese fragment and compare it to the plaintext. Try to work out how they could map to each other. The advantage is that that there is more room for creative solutions, unforeseen realizations, flexible readings. This is closer to how we read actual medieval manuscripts, but there we can use context and expected words to determine unclear readings and even correct scribal errors.

2) Use EVA and computing power, but build in some flexibility. Ideally, the system would be able to parse EVA differently of needed, for example read [ch] or [iin] or [aiin] as single glyphs. ANd take into account two possible readings for the problematic glyphs. Even better would be a system with some general fuzziness, that tries to look for a close match in the realization that a perfect match is always unlikely.

The option (3), look for a 1 to 1 match between raw EVA and a plaintext makes no sense to me.