Decoders who are happy with word salads to seed their creative writing do not need anything more complex than simple substitution.
For more inquisitive researchers, the problem is identifying cipher systems that could result in all the bizarre properties of Voynichese. Bowern and Gaskell (You are not allowed to view links. Register or Login to view., 2022) consider several different methods, similarly to the approach mentioned by Nablator. They conclude that "Many manipulations produce outcomes that are similar to Voynich text on at least some metrics... [but] These text manipulations did not produce a single set of changes that mimic Voynich text".
Hauer and Kondrak (You are not allowed to view links. Register or Login to view., 2016) attempted an in-depth exploration of anagrammed substitution ciphers. Their approach is straightforward and sensible in my opinion: writing a custom software aimed at decoding the text under specific assumptions about the nature of the cipher. But before committing such a great effort in experimenting with a specific method, one should be reasonably sure that the method can result in something statistically overlapping with Voynichese (which is not the case for an anagrammed simple substitution). At the current state of research, Bowern and Gaskell's approach seems to me the best line of investigation for those interested in Voynichese as a cipher.
For more inquisitive researchers, the problem is identifying cipher systems that could result in all the bizarre properties of Voynichese. Bowern and Gaskell (You are not allowed to view links. Register or Login to view., 2022) consider several different methods, similarly to the approach mentioned by Nablator. They conclude that "Many manipulations produce outcomes that are similar to Voynich text on at least some metrics... [but] These text manipulations did not produce a single set of changes that mimic Voynich text".
Hauer and Kondrak (You are not allowed to view links. Register or Login to view., 2016) attempted an in-depth exploration of anagrammed substitution ciphers. Their approach is straightforward and sensible in my opinion: writing a custom software aimed at decoding the text under specific assumptions about the nature of the cipher. But before committing such a great effort in experimenting with a specific method, one should be reasonably sure that the method can result in something statistically overlapping with Voynichese (which is not the case for an anagrammed simple substitution). At the current state of research, Bowern and Gaskell's approach seems to me the best line of investigation for those interested in Voynichese as a cipher.