[Interview] An interview with Stephen Bax

[-JKP-]

One has to study and discern the structure first, and then it might make sense to look for nouns/names or whatever might fit the circumstances. It does no good to look for nouns if the text is a self-replicating system or if it is a numerical system, or steganography, or any other of the dozen ways text might be represented other than in a natural-language substitution mode.

We've been doing that for a century, Jpk. Let's try a different angle and see what comes up. OK, so Stephen hasn't made any public headway on his angle of attack, but that's not to say we should stop.

Most of the computational attacks (both historically and recently), assume spaces are spaces and there are no nulls, even though a high proportion of ciphers have nulls (including medieval ciphers) and many are known to have manipulated spaces (also including medieval ciphers). Most computational attacks assume the glyphs are letters even though they are not positionally structured like natural language. A few have yielded good data but many of them are based on unproven assumptions.


Stephen's approach (at least the one demonstrated in the 2014 video) makes the same assumptions ... that the glyphs in some way represent letters, that the spaces are spaces, and that glyphs are intended to be read sequentially, etc, and, on a meta-level, that they can be generalized to the rest of the text if one can simply figure out "key" words (note that nulls are not even mentioned, probably because nulls are not relevant when one is trying to decode ancient natural languages, but this may be a cipher, not just an alternate set of glyphs). The approach demonstrated in the video is the most common approach. It's the one used in 90% of the "solutions" that I've seen so far. It's the one that is most intuitively obvious and accessible. I'm not saying there's anything wrong with trying it, but it is, in a sense, the "first approach" and it doesn't represent a different angle.


As I keep saying, "What if the spaces are contrived? What if there are nulls, modifiers or markers? What if the vowels are not vowels and we only assume they are due to their placement? What if they are not letters? What if it's steganography?" I even posted an alternate version of Voynichese on the forum that renders the vowel shapes so they don't look like vowels so our minds are not misled into thinking they are and no one seemed interested in exploring that angle even though there are historical precedents for fake vowels in old ciphers. Also, what if it IS a verbose cipher or anagramed (not the way Sherwood suggests, but in an algorithmic way that can be read back)? The common approach doesn't take into consideration any preprocessing that may have been done to obscure the text.


I'm not convinced we have studied the structure sufficiently to understand it. Even though there are a few people genuinely trying to unsnarl it, it still seems to me to be the most under-studied aspect of the VMS.