Re: VMs: No stats no fun ---> no stats no blinkers! :-)

[Nick Pelling <incoming@xxxxxxxxxxxxxxxxx>]

Hi Rene,

At 13:38 08/08/2003 -0700, Rene Zandbergen wrote:
> --- Nick Pelling <incoming@xxxxxxxxxxxxxxxxx>
> seems ready to disregard statistrics gathered
> about the VMs over the last few decades.

Rather, I'm proposing a connected set of hypotheses which (if true) would 
explain why prior statistical analyses of current transcriptions may well 
have been taking secondary structural artefacts (ie, those of a constructed 
word-game designed to resemble language) as their inputs rather than 
primary structural properties: and hence would have been misdirected.

> Also, everybody up to now has been using images
> of the MS (copyflo, photostats etc) and
> this has not led us anywhere, so perhaps we
> should discard them as well?

The hypotheses I'm proposing are far from that paranoiac - I think the 
reproductions we're now working with (especially the no-ip scans) are 
(apart from their general lack of colour) not too far from primary data. 
And there's nothing wrong with primary data. :-)

> Spass beiseite, it's not the statistics which are
> wrong, but it's the lack of understanding of
> them that could be a problem.

One thing I figured out for myself during my philosophy degree 20 years ago 
is that all data - such as statistics - are dependent on their contexts: 
that there is no such thing as context-free data.

The contexts used for VMS statistics to date have been the obvious ones - 
stroke-based / glyph-based / word-based / line-based / etc. They would have 
been equally obvious 500 years ago - code-breakers then were just as able 
as we are to make inferences from symbol counts. I'm merely suggesting 
steganographic ways of hiding those kinds of statistical artefacts, 
consistent with known cipher practices (pair ciphers, transposition 
ciphers, etc), from which a devious Quattrocento code-designer could have 
constructed a novel coding system.

BTW, surely the manifest lack of consensus - from the statistics - about 
the answer to the most basic question (is the VMS a code, a cipher, a 
natural language, an artificial language, etc?) is some kind of indication 
that the vast statistical machinery we've been using might well have been 
misdirected? In other words, that we may be using a pile-driver to try to 
smash a walnut that has failed to be placed beneath it?

> The stats that have been produced up to now will
> tell you _exactly_ where the text of the VMs is
> unusual. This is where our chances for cracking
> it lie.

To my eyes, the VMS exhibits unusual properties in nearly every statistical 
test that's been done. So, do our chances of cracking it lie nearly everywhere?

> Whatever encryption method you may propose, its
> effect on the statistics of the plain text can
> be predicted. No amount of arm-waving can replace
> a proper analysis.

I completely agree - all I'm trying to do is to be transparent about my 
observations and theories, on the road I'm taking towards a rigorous 
transcription methodology. My hope is that, by using a better transcription 
as their inputs, future statistical analyses will be able to identify 
primary structural properties in the VMS.

> I have quite a bit more to say about this, but
> prefer to keep it short for the moment. There
> are lots of fascinating statistics which Stolfi
> has collected, which still have not received
> ther deserved attention.
> Disregard at your own risk (that goes for me too).

We now have plenty of Internet hosting available to us, so we should be 
able to make significant amounts of this kind of analytical data openly 
available for peer review and discussion!

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


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