Hello Rene,
======= At 2004-09-12, 06:04:00 you wrote: =======
>This really is not correct. The word length
>distribution depends on two main questions:
>- which transcription alphabet is used? . . .
There is apparently more things to consider, since we really do not
know the function of the
spaces yet. Let's cathegorize:
1) The spaces are used as delimiters of words as in natural language -
that is I believe the
case you described.
2) The spaces have been inserted as nulls ( in plain language, the spaces
between words
are actually nulls as well - except being delimiters - but that is not
what I mean here :-).
Say if I write the whole VM without spaces - as one word - and then
insert spaces by
random, I get text, where the solver needs only to eliminate the spaces
and guess the
proper spaces by the content. I believe it will give him say 97 percent
certainty, if we are
talking about plain, simple text and provided he can read the script and
knows the language.
However, with limited number of "word lengths" used ( say by throwing
multi-sided dice)
he will not get the VM "word frequency curve" - all word lengths will have
about the same
frequency, right?
3) Same as 2), but more complicated - say length of the next "word" will
be function of the
dice throw and of the length of "preceding word". There are many
combinations and the
results may vary.
4) The "word length" is the function of some encrypting algorithm - but at
the same time
it will simulate the bell curve. I cannot imagine off-hand how it can be
done.
5) Other possibilites - but hard to specify in detail.
In none of above cases is the "word length frequency curve" the result of
PURE
probability, since the VM has similar shape as for other
natural languages. What we seem
to know for sure is that the longer "words" in the VM are somehow missing
in
comparison with majority of natural languages - and the bell curve is
skewed or simply
nonsymetrical. That still does not eliminate the use of natural
language. Interestingly
enough, the AVERAGE "length of the VM words" according to our statistics
is only
about 1 character shorter than for some other languages. Can it indicate
that the total
number of the spaces is about the same, they are just misplaced?
Now talking about Cardan grille - would Ruggs gibberish have spaces and
how would they
be encoded? The use of grill will certainly mess them up :-). Or would
the grill provide new
spaces with the same word frequency curve as the VM? And how?
Regards,
Jan
______________________________________________________________________
To unsubscribe, send mail to majordomo@xxxxxxxxxxx with a body saying:
unsubscribe vms-list