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

Re: Re: VMs: Criteria for a successful solution



Hello Rene,  

======= At 2004-08-16, 06:32:00 you wrote: =======

>One can be a bit flexible here, but a method that
>only works for two pages, or one word out of
>every 20, cannot be accepted. We already know that
>some things are consistent throughout the 
>MS (some frequent words occur all over) so anyone
>who proposes a solution for only a section of the MS
>should be able to indicate why it would not work
>elsewhere.

Now that's where I think we restrict ourselves to impossibility:
if different keys and same nulls (call them  prefixes, suffixes and what not)
are used for different parts,  we get the mentioned similarity, 
but still are not able to read some parts.
I do not know why the author would do that, but cryptographers change
keys quite frequently. If for instance the coding  was used in pictures,
 it might have been quite different than the one used in text thanks to different 
 possibilities.

   Strong's solution is not a bad example:
>he uses a polyalphabetic cipher with many different
>alphabet tables. These are all parameters that he
>can adjust in order to get meaningful text out.
>If it is not the right solution, it will only be
>possible for a relatively short part of the text.

That is an opportunism. I could however hardly see something which solves the 
whole page  (I mean  sentence by sentence and without artificial add-ons)
not being at the same time the partial solution for the VM. If there were just 
different key used (and same nulls), we may be able to keep the method and search 
for the other key as most likely event. 
 
 
>Well, many label words are the same as plain text
>words. They also largely follow the same paradigms.
>Yet, they are different in that they don't
>follow Zipf's law. An acceptable solution will
>need to come up with a good explanation.

Right, the mutual appearance may not mean the "word"  has same meaning, but 
rather the indicated usage (i.e. it may be the key). 
Say top  row of 1006247.SID:   there are shown some  groups of individual roots 
or bulbs.  The "labels" there are however different for each other picture and what is more 
interesting, the number of roots vary from 1 to 10 - something single plant does not 
have :-).   I do not know statistics, but to me it looks like the appearance of the same 
labe  in the text may  be  more like coincidence than like some rule. 

>It should be valid for a cipher as well. Nulls
>could be part of the explanation, why not?
>
Definitely, but the grammatical rules should  apply more to the plaintext while   
cryptographic rules may not recognize suffixes and prefixes. if it is really encoded 
and simila "suffix type" groups appear, those are more likely the nulls and we should
 call them that way.  
 
>> Right,  but that brings another point: we cannot
>> expect 100 percent solution, based on the 
>> fact that some plaintext words or expressions would
>> be obsolete, unexplainable and some 
>> (maybe) related to contemporary events and so on.
>
>Agreed. There will undoubtedly be errors in the
>MS text as well.

Still, I am optimistic:  if 80, 90 percent of each translated "sentence" 
(they do appear rather long, don't they?) is sensible, it's worth 
testing it further. 

Again, I repeat what I said  in response to Elmar: by finding which points 
were the most difficult to satisfy, we may be able to focus on them. I consider 
Grant's idea quite  intriguing: why it was so easy to encode the VM ( judging by 
smooth  handwriting in the VM  and no apparent mistakes, returns, crossings etc.) 
and so  difficult to decode it?   What does it tell us? I wish I know :-).

jan
 



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