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

Re: VMs: Making a vms with meaning (long)

Hi Gabriel,

I would like very much to have a copy of your text. 

Two things have been nagging for some time:

(1) What class of ciphers has letter sequences that, with a fair 
degree of probability, can be substituted in such a way that the new 
text is almost pronounceable even without using all the vowels 
available? What class of ciphers almost never produces such a text? I 
would think transpositions would be included in the latter unless 
subjected to further manipulation. 

(2) The scarcity of repetitions of two-, three-, and four-token 
sequencies would seem to argue for gibberish, scrambled word order, 
or non-functional word modification.Is it equal to a random 
assortment of the words? Is it, perhaps, even less than that? If so, 
then it must be by design or by consequences of a design. 

I would expect to find many more "chedy daiin"  and longer clusters 
in your text than in the VMS. If I had a talent for cryptology I 
might attempt to discover the meaning of those words in other 
contexts. I do not but still would like to have the text for other 
purposes, chiefly for comparisons with the VMS. I am glad you did 

Ciao ......... Knox

On 27 Jun 2004 at 15:38, Gabriel Landini wrote:

> Hi,
> I have been thinking about Gordon Rugg's "solution" for some time now.
> For sure it is an interesting method to generate nonsense, but I am
> not convinced that it logically follows that the vms is (or could be)
> a hoax of that type.
> Although he has shown one method of generating something like the real
> thing, it still has to be shown that: 1) any statistical properties
> are in fact, similar to the vms, and 2) how likely is one to come up
> with grilles that produce so much text showing the degree of
> consistency seen in the vms.
> Note that Gordon is aware that one needs several grilles to make
> something as big as the vms (otherwise it starts repeating). He is
> surely reverse- engineering the grilles, to fit the vms. However
> extrapolating to a possible hoaxer, it assumes that s/he was intending
> to make something like the vms with a large degree of consistency
> throughout the grilles. To end up with a set of grilles like that, I
> think, may be difficult to achieve without a clear idea of where the
> grille cuts are to be made). What I mean is that (i think) it would be
> quite unlikely to end up with many grilles that all produce vms-like
> text, unless you knew what one is intending to produce.
> So I decided to give some hammering to the 'hoax assumption' ("it has
> not been read because there is nothing to be read") by producing
> something that looks and has similar properties to the vms, AND has
> meaning AND is difficult to crack. The method of encoding I used is
> simple, straightforward and perhaps time consuming. Decoding by the
> author may be relatively difficult (certainly it is possible to read
> back, I am not sure as it could be real time read, perhaps not) but
> cracking by a third party it in its entirety is (I think) *quite*
> difficult.
> Here is some text block cut randomly from the encoded corpus:
>  otal oldar chor lkeedol eer ol dair chedy daiin ockhdar cpheol chedy
>  xar qokaiin y chedy kshdy ololdy aiin char y okeey oldar qokaiin lsho
>  daiin olsheam qoeey chedy dchos pshedaiin shedy d qol key sheol or
>  cpheeedol qokedy qokaiin daiin cthosy chedy ar aiir chedy teeol aiin
>  cheey y cheam oky qokaiin daldaiin loiii ar shtchy chedy aldaiin
>  ydchedy daiin shd okaiin qokain daiin qotcho chedy daiin lchy oloro
> I produced a text of similar size of the vms with this entropy
> statistics on the first 32000 characters using Monkey:
>  NewText		vms
> h0	4.70044		4.64386
> h1	3.85988		3.82127
> h2	2.02343		2.14011
> and it follows Zipf's law AND it has a meaning.
> I can send the text who whoever wishes to have a look.
> I thought of leaving this as a puzzle for the list to solve, but maybe
> it is better to tell how it was made, so even knowing the method, one
> realises that it may be quite difficult to crack it.
> I used a nomenclator to exchange the words for something artificial. I
> used vms words in their ranking order to exchange the plain text (PT)
> words. So the most common PT word is the most common vms word and so
> on. Therefore all character-derived statistics are due to the rules of
> construction of vms words and little to do with the language (since
> most languages follow Zipf's law). Entropy is left almost unchanged,
> and Zipf's law is exactly the same as in the original PT. I imagined
> the author creating a two-way dictionary of imaginary words --that
> correspond unambiguously to real words-- as he writes the vms.
> You do not know neither the PT source nor the language. 
> Any suggestions on how to crack this *in full*?
> Cheers,
> Gabriel

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