Re: VMs: VMS generator table

[Nick Pelling <incoming@xxxxxxxxxxxxxxxxx>]

Hi everyone,

> On Friday 05 December 2003 01:08, Jeff wrote:
> Find it here http://www.georgeboolefanclub.com/vmstextgenerator.html

Thanks for putting this up on your site, Jeff: clearly, your ideas have 
moved on from the earlier (Italian-plaintext-centric) set of tables you 
posted (and described) earlier.

However, what would you learn about English if you tried applying the same 
methodology to that instead? I'd suggest... probably not a great deal.

The whole point of building models is to use the process of trying them out 
(on real data) to throw light on what may be driving their behaviour. For 
example, people often use dynamic / multiple regression both to forecast a 
particular share price *and* to investigate what might be leading 
indicators (for example, dollar-sterling exchange rates, OECD consumer 
confidence metrics, etc) for that company.

Most people here who have looked at the VMs for a while will have an 
instinctive feel for what is (and what isn't) Voynichese: many will be able 
to convert that 'feel' into a model (for words or letters). But what do any 
of these models tell us about the VMs itself?

Voynichese contains a viciously tricky interplay between letter-adjacency 
rules, letter-position rules, and semantic content, whereby it's hard to 
tell where one stops and the other starts. This is (I believe) why 
statistics don't help us much, as we're continually measuring intermediate 
data, not final language data.

What, then, is the answer? The first step might well be to stop hoping that 
the VMs is a simply (sequentially) enciphered raw European plaintext - 
there appears to be little or no evidence to support this view. What seems 
far more likely is that its plaintext is a shorthand or private language, 
one probably based on a well-known European language (and furthermore 
probably Italian, Latin, or (say) Provencal), which is then enciphered in 
some steganographically non-obvious way.

I see this as more akin to Navajo encoding (sending a message via a private 
language) than to the Enigma machine (a computationally complex enciphering 
process). Always ask - how could someone have achieved this 500 years ago? 
If it's not an obviously polyalphabetic cipher (and in fact the stats are 
all wrong for that), then it *has* to be simpler than you think (to quote 
Steve E).

The second step is to try to work out what constitutes a single letter in 
the language/cipher. Many of the ciphers in the Milanese cipher ledger use 
pairs of letters to code for single plaintext letters, so this is quite 
reasonable. "qo"? "dy"? "ol"? "or"? "od"? "ii"? "ee"? "eee"? etc. But 
really, using EVA as the basis for your code is building on sand. What do 
you see when you look at the page itself?

The third step... well, to be honest I'm still working on the second step. :-o

The main thing to notice is that, in our quest to understand the VMs, 
statistics have yet to help us significantly (for example, we can't even 
say if it's a hoax or a code based on statistical evidence). Whatever the 
'trick' to the VMs, it's an intensely "stats-unfriendly" one - so relying 
on just one set of stats (as you do) is highly unlikely to move you forward 
either.

Anyway, some stuff to think about as you plan your next move. :-)

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


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