[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

RE: VMs: Personal Guess

Hi GC,

Lances up - another day of jousting begins... :-)

At 22:57 06/08/2003 -0500, GC wrote: 
If you suspect the VMS to be abbreviated (even lightly), I'd like to see
evidence of this.  What do you mean by "abbreviated"?  What statistic
singles out these "abbreviations"?  Usually the purpose of abbreviation is
brevity, something like the 'x' for 'us'.  To fit the 'standard' definition,
it would be reasonable to provide statistics on the single glyphs that act
as "abbreviations", and your rationale behind choosing these above others.

(1) I assert (somewhat uncontroversially) that the VMS' appropriation of the Tironian notae "8" and "9" points to the VMS' alphabet designer's choices being informed by shorthand. In fact, I'd comfortably argue that this alone points to the VMS' not being a natural language in an invented alphabet: why would someone invent an alphabet that not only appropriates two of the few remaining Tironian notae in use, but frequently places them at the beginnings and ends of words?

(2) I strongly suspect that just about every statistic gathered to date may have been misdirected - the strong implication of the pair cipher hypothesis is that statistics based on glyphs would be next to useless. If you also have misgivings about soft spaces vs hard spaces (or if you suspect some kind of space transposition cipher is in use), you'd have much the same kind of suspicion about "word"-based statistics. So: I don't have any statistical evidence - but I don't think that you do either.

(3) My abbreviation hypothesis arose simply out of trying to explain the "curls" / "tears" / "question marks" accessorising the <ch> pair. Under what possible circumstances would a logical, well-thought-out, systematic mode of expression (ie the rest of the alphabet) need to be compromised / accessorised / altered by these (apparently) semi-improvisational marks? After a lot of deliberation, it seems highly likely to me that these represent disambiguation marks (or "hints"), designed to help the decoder guess the correct replacement word. So: I only see <ch> and the various forms of <sh> as abbreviations, but that isn't born of statistical analysis.

As for what I mean by "abbreviated": you talk about "x" for "us", but that's a notae/nomenclature-like tokenisation (see the first cipher of the Milanese cipher ledger for an example of this kind of nomenclature gone crazy).

Strictly speaking, abbreviation (rather than tokenisation) takes two forms - contraction and truncation. Let's look at the Radcliff/Ratcliff shorthand Lord's Prayer:-

        Our Fth wch rt n hvn ; hlwd b y Nm
        Y Kgdm cm Y wl b dn n rth z it s n Hvn

What types of abbreviation are at play here? Clearly, many vowels are being discarded (contraction) and some words are being terminated early (truncation).

The obvious question then comes: what kind of abbreviation do I think is denoted by {ch|sh} in the VMS? I'll try to explain my thoughts by going over a fragment of pair-transcribed text from f24r:-

        k.ar.ch
        t.ar.s.ch.e
        or.cth.ol.qo.d.ol.y.ch

Firstly, I suspect that - for most common words - many vowels have simply been discarded (ie, contraction), and it would be fairly pointless to have a glyph coding for this kind of abbreviation. But for truncation, that's a different matter - especially if the spaces in the text have been somehow obscured.

All of this leads me to predict that the structure of the first word "k.ar.ch" is probably "consonant . consonant . <truncation>".

For the next word - "t.ar.s.ch.e" - I predict that we might well have "consonant . consonant . consonant . <truncation> . <particular noun/verb ending>".

However, I believe that the next word - "or.cth.ol.qo.d.ol.y.ch" - may well be a proper name, or a similar CT word not really amenable to obvious internal contraction, just truncation once it gets beyond a certain length (like "Arist--" for "Aristotle"). Therefore, some of the pairs in this particular word may well code for vowels.

Abbreviation is a natural partner for verbose ciphers, especially if you want to try, as part of encoding an already-existing document, to reproduce the general textual layout of the original. If the subject matter is not a particularly good fit with heavy abbreviation, the alternative is to use a finer quill - this may be part of the explanation for what we observe in the VMS' hands.

Fair enough criticism.  You won't mind filling me in on Strong's "specific
historic model"?  Let's see, Selenus and others up to Strong's time covered
a very wide range, including nomenclators and homophonic substitution
ciphers.  Even most of Friedman's critical papers were out (and read by
Strong) by the time he took on the Voynich. As far as "external historical
research", his interest was in late 16th century, early 17th century cipher,
a little later than the Voynich, but not by much on the evolutionary scale.

Put simply, late 16th Century cipher is what Strong was looking for, and late 16th Century cipher is what he thought he found. He didn't - and you don't - seem to cut the 15th century code-makers much slack: I think they were smarter than we give them credit for. After all, they practically invented modern cryptology.

More than that - According to Strong, he spent over 500 hours of research
(on two pages) before the light broke. <snip>

I'm just talking about two pages' being an insufficient sample for Strong to derive a theory from. If he had been allowed to see "the bigger picture", I'm sure his ideas would have moved on.

> Generally: you and I both see the VMS as a non-obvious polyalpha system -
> but please accept my caution that I think the VMS author is trying to
> sidestep our feeble post-1912 attempts at cryptology by misdirecting us,
> certainly at the letter level and probably elsewhere as well. For
> example,
> I would be completely unsurprised if the order of the letters within each
> word were transposed in some way - perhaps reversed.

I'm not certain what you mean by "non-obvious".  The very nature of cipher
is to make the solution "non-obvious".

No, the nature of cipher is to make the solution mathematically hard (ie, you have to rely on statistics to help you get to the answer) - the nature of steganography is to make the solution non-obvious.

If I think I see something, I go to
the database and pull up the numbers.  If they work in my favor, go a little
further.  If not, rethink it, try something different.

I wish I had even 10% of your faith that current VMS statistics tell us anything about the coding system. Just because it's easier to look for your keys under the streetlight doesn't mean you dropped them there - here, I'm certain that the VMS' author constructed a system with a bright (steganographic) light, ready to be surrounded by (cryptological) moths, but which actually illuminates nothing.

I personally don't
think the VMS author knew or cared about a 'post-1912' world, and certainly
didn't think his manuscript would be the subject of so much discussion.  He
didn't write it to misdirect us, we're doing a good enough job of that all
by ourselves.  He wrote it for his own use, for personal reasons, and
couldn't have cared less about you or me.

What if (as I personally suspect) the exact opposite of every assertion in this paragraph were true? :-p

One of the reasons I enjoy this list is that I get to play the "thesis-antithesis" game with smart people like you - but perhaps one day (soon) we'll get to "synthesis"... game over! :-)

Cheers, .....Nick Pelling..... 


______________________________________________________________________
To unsubscribe, send mail to majordomo@xxxxxxxxxxx with a body saying:
unsubscribe vms-list