Google Groups no longer supports new Usenet posts or subscriptions. Historical content remains viewable.
Dismiss

What Can be Taken Out at This Point in Time.

19 views
Skip to first unread message

Austin Obyrne

unread,
May 11, 2013, 5:23:58 AM5/11/13
to
I am making the point that while it is still early days and possibly too early to make sweeping generalities in cryptography it is safe and timely nonetheless to say what has become apparent to me at least.

The set of integers that is modelled as equidistant points on an arbitrary (i.e. directionless) straight line is hugely vulnerable to cryptanalysis. The reason being the inevitable residual structure that always remains in the numeric ciphertext after the plaintext has been transformed by whatever means in the encryption process.

This structure (i.e. the order of presentation) of the ciphertext as numbers can be only partially obscured but not totally diffused to a point where all traces of its original structure on the number line has been erased, as a result it is completely indelible in this respect.

The upshot is that cryptographers must deliberately seek out extra compensating complexity that will hide this transparency of the raw data as a precursor to encryption, by analogy this is like paying a massive down payment before any intellectual obfuscation of the actual plaintext can start.

In a nutshell, the number system we all know so well is in fact far too well known to be useful in cryptography because of the highly predictable order of the elements i.e. the integers that comprise it. Using everyday numbers from the traditional number line as the primary selection domain of the raw encryption data is tantamount to ‘giveaway’ information being presented on a plate to a cryptanalyst adversary who knows he may safely assume the same order of the raw data that the encrypting person used.

(In entropy terms if you prefer – these numbers have zero initial entropy when they should have much, much more entropy as a basic starting point to the cryptographer for adding more entropy to them during the encryption transformation). They are indeed the worst possible selection domain and indeed so also are other data sets like alphabets and ASCII as the worst possible selection domains for direct selection (i.e. without some substitution before use) of raw encryption data

I have laboured the point many times that one solution to the transparency of using integer numbers as encryption data is to use the widely disparate integer points in three-dimensional space as an alternative to the straight line, sitting duck, traditional number line. Three-dimensional space ensures total disorder to outsiders while being easily retrievable to the legitimate entities within.

This of course takes the encryption methodology into the realm of vector arithmetic which should not be problem to anybody who is properly prepared for crypto research or daily implementation of vector based ciphers in the various infrastructures.

In practice such ciphers are always embedded in computer programs anyway and any non-specialist operator does not need to know anything about vectors so as to implement the software.

There are ciphers up and running for years that demonstrate this. - adacrypt

Austin Obyrne

unread,
May 11, 2013, 6:08:15 AM5/11/13
to
On Saturday, May 11, 2013 10:23:58 AM UTC+1, Austin Obyrne wrote:
> I am making the point that while it is still early days and possibly too early to make sweeping generalities in cryptography it is safe and timely nonetheless to say what has become apparent to me at least. The set of integers that is modelled as equidistant points on an arbitrary (i.e. directionless) straight line is hugely vulnerable to cryptanalysis. The reason being the inevitable residual structure that always remains in the numeric ciphertext after the plaintext has been transformed by whatever means in the encryption process. This structure (i.e. the order of presentation) of the ciphertext as numbers can be only partially obscured but not totally diffused to a point where all traces of its original structure on the number line has been erased, as a result it is completely indelible in this respect. The upshot is that cryptographers must deliberately seek out extra compensating complexity that will hide this transparency of the raw data as a precursor to encryption, by analogy this is like paying a massive down payment before any intellectual obfuscation of the actual plaintext can start. In a nutshell, the number system we all know so well is in fact far too well known to be useful in cryptography because of the highly predictable order of the elements i.e. the integers that comprise it. Using everyday numbers from the traditional number line as the primary selection domain of the raw encryption data is tantamount to ‘giveaway’ information being presented on a plate to a cryptanalyst adversary who knows he may safely assume the same order of the raw data that the encrypting person used. (In entropy terms if you prefer – these numbers have zero initial entropy when they should have much, much more entropy as a basic starting point to the cryptographer for adding more entropy to them during the encryption transformation). They are indeed the worst possible selection domain and indeed so also are other data sets like alphabets and ASCII as the worst possible selection domains for direct selection (i.e. without some substitution before use) of raw encryption data I have laboured the point many times that one solution to the transparency of using integer numbers as encryption data is to use the widely disparate integer points in three-dimensional space as an alternative to the straight line, sitting duck, traditional number line. Three-dimensional space ensures total disorder to outsiders while being easily retrievable to the legitimate entities within. This of course takes the encryption methodology into the realm of vector arithmetic which should not be problem to anybody who is properly prepared for crypto research or daily implementation of vector based ciphers in the various infrastructures. In practice such ciphers are always embedded in computer programs anyway and any non-specialist operator does not need to know anything about vectors so as to implement the software. There are ciphers up and running for years that demonstrate this. - adacrypt

Late supplement.

Forgot to say,

In my view all cryptography to date, which includes the vast number of papers past present and future by the establishment are attempts at providing this compensating complexity when the real solution is to stop using numbers.

Numbers in cryptography are still usable but it requires giving them some substitution change that takes the cryptographer into some alternative methodology other than ordinary number work.

This simplifies the encryption transformation and enables stronger cryptology.

David Eather

unread,
May 11, 2013, 8:43:53 PM5/11/13
to
So you are going to live in your fantasy land?

If you saw a book on Amazon.co.uk that has dozens of reviews and they were
all uniformly bad would you assume it was a good book?

You say you've had 9000 hits on your website but the only comments you've
had are dozens (or hundreds?) saying your cryptography is bad. Not a
single person has said it is any good. What should that tell you.

You're looking for confirmation of your 'genius' from this group but you
are never going to get it. We have all looked at your stuff and have all
independently decided it is crap because it is. You don't even know enough
to understand why it is crap.

Don't waste time with this group looking for what you won't get. It is
better for everyone for you to take your crap elsewhere where someone
might be fooled into thinking/saying it is good or best of all get
yourself to a psychologist because you really need help.
0 new messages