RE: VMs: [ha] [hb] not different languages

[Nick Pelling <incoming@xxxxxxxxxxxxxxxxx>]

Hi GC,

At 13:26 03/08/2003 -0500, GC wrote:
> > Also: a space transposition cipher (ie, one where (real) word endings are
> > typically hidden inside (fake) words) may well retain some
> > semblance of the original word-length.
>
> In that case, the observed word beginnings and endings would then be in the
> middle of words.  I believe you've suggested (correct me if I'm wrong) that
> certain glyphs indicate the break for the word.
> InEnglishwedontneedanybreaksrepresentedtounderstandtheflowofthetextSowhywoul
> dthesebeinsertedinsteadofrealwordbreaks?

My current model predicts that {ch|sh} indicates an abbreviation and 
{c|cc|ccc}{o} indicates a suffix - while a suffix would normally be at the 
end of a (real) word, words could quite easily end in other places - so, to 
be precise, my suggestion is that these suffix glyphs terminate probably 
most (but probably not all) words.

It might well be that <dy> (your "89") may often simply be a null to hide 
word-endings. :-)

As for the possible efficacy or usefulness of my suggested space 
transposition cipher, I give the following example taken from your last post:-

xi nen gli shwedon tyne eda nybre aksrep resen tedytoun ders tan dyt heflo 
woft hete xyt sow hywo uldyt he sebein ser tedin ste ado fre alwor dyb re aks?

The only place this is recognisably the same as your original text is where 
long words get broken up into smaller ones (like "represented", 
"understand") or where pairs of consonants get placed side-by-side (like 
"and the", "word breaks"). But these are precisely the kind of 
statistically (and visually) noticeable things an abbreviatory mechanism 
(whether contraction or truncation) would be designed to try to get rid of, 
so I'm not too surprised. :-)

FYI, it also required an initial null character (I used "x") to hide the 
very first character, to throw the code-breaker very slightly off the trail 
- IIRC, a number of different analyses (was it Mark Perakh's again that I 
just read?) have pointed to gallows-initial paragraphs as often starting 
with (say) a p-gallows prefixing a valid-looking word which appears 
elsewhere - this would perhaps point to paragraph-initial gallows often 
being nulls, much as per my example.

Finally, I also added a few "y" characters to hide where consonants ended 
up side by side - I suspect that this may be the functional equivalent of 
the <dy> pair. All in all, this feels pretty VMS-like to me. :-)

BTW, what do you make of the average word-length of my space-transposed 
text? :-)

>   Also, the middle of the words
> would then take on the conformity we see in the word initials and terminals.
> Why should this be so, even in paired cipher?

If I had it all worked out, I'd be posting up the cleartext instead of 
jousting with you. :-)

But as I've said numerous times before, I only see the paired cipher as the 
back-end coding stage - there's much more going on under the hood than 
simply pairs of letters here - but reducing the verbose ciphertext to a 
conventional ciphertext should enable statistical analyses to produce more 
usable output.

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


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