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

Re: VMs: new here

Hi Laura,

At 13:51 16/07/2004 +0100, Laura Aylward wrote:
My name is Laura Aylward, I have recently carried out research on the
manuscript with Gordon Rugg. Whilst doing the work I have been silently
watching the list and thought it was about time I introduced myself. I will
also take it upon myself, and hope Gordon won't mind me doing so, to point you
some results of our work at..

One persistent criticism of Gordon's work is that it is based on an idealised conception of Voynichese, which fails to correspond with what we actually observe.

For example, you have probably read (during your recent lurking) that the VMs only matches Currier's A & B dialects (or languages) somewhat imperfectly - that there is a spectrum between the two, ie it's not a binary choice. Yet again, the VMs manages to make monkeys of our attempts to pigeonhole its properties neatly - some pages are indisputably A or B, while others are a kind of mixture. IIRC, GC claims that paragraphs are normally A or B (but that some pages do contain both), while others assert that intermediate languages/dialects are present. So, I would say that any attempt to reconcile Gordon's grille/table analysis with the VMs would also need to evaluate the degree to which these theories are supported by the evidence.

I therefore find all the assertions made by Gordon about the compatibility of the grille/table hypothesis with Currier language/dialect evidence somewhat wishful thinking at best, and even disingenuous at worst. For example (page 10):-

As expected, different tables produced different outputs. Interestingly, though,
different "dialects" were produced by tables with different amounts of structure, even
when the tables used the same syllables in the same proportions. For instance, in one
batch of outputs, the word 'qoky' appeared in text generated from the low structure
table, yet in no other text using the same syllables in the same proportions, but using
different degrees of structure.

This has clear similarities with the "dialects" of Voynichese.

No, this doesn't. AIUI, Currier's observations of languages were based more on the apparent structure of words and specific letter patterns than on occurrence/non-occurrence of single words. The behaviour of Currier's languages forms the single most testing element of the VMs' stats AFA the grille/table hypothesis goes - really, you would need a very much more detailed understanding and exposition of that in your work in order for the claim of compatibility to be supported. Bear in mind that this is the one thing which might obviously falsify this work... worth examining carefully.

My own criticism of Gordon's work is simply that, by requiring that grilles are moved in a random way (what does the paper mean by "semi-random", anyway? Pseudo-random?), he is trying to perform sleight-of-hand with the information - to move it out of the text and into the random-number generator. But surely the whole point about grilles in the first place was to explain how randomness could be reconciled with the kind of highly structured letter arrangements we see in the VMs?

BTW, given Gordon's interest in information theory, an interesting exercise for you might be to calculate how much "choice" (and hence negentropy) is being exercised in the "semi-random" grille moving stage - and hence how much information is being introduced there. That would give you a specific figure to compare with the various calculated entropy values - if there is a significant mismatch between the two (as I suspect), then you would have been a good scientist (in the best Popperian sense) and perhaps falsified your theory. :-)

Things to think about, at any rate. :-o

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

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