Re: PRNG Crypto

On 2024-08-08 7:10 p.m., R Kym Horsell wrote:

In alt.global-warming Nadegda <nad318b404@gmail.invalid> wrote:

Time to trigger the right-wing snowflakes again. Melt, snowflakes, melt!
On Sun, 04 Aug 2024 12:21:53 +0000, R Kym Horsell wrote:
>

In alt.global-warming Creon <creon@creon.earth> wrote:

On Sat, 03 Aug 2024 06:21:23 -0500, X, formerly known as \"!Jones\" wrote:
>

If you have a medical emergency, stop reading, log off, and dial 911!
>

I've wondered about this for a while.  Just asked about it in sci.crypt,
we'll see what they say.

>
I would expect the level of discussion regarding cryptography there to
mirror the deep thought into the politics of guns in talk.politics.guns.

>
Take a look and see.
>
My geek reading time is currently taken up reading _UNIX: A History
and a Memoir_, but I am curious how one would crack it.

....
>
Cracking a "one time pad" can sometimes rely on the encoding being lazy.
A PRNG generally comes in several different forms. Try each form until
you at least get some partial matches.
If you have a quantum computer you can of course test all possible
n-bit numbers for all the parameters of the RNG at the same time.
Strong cryptographic RNG rely on other forms of randomness that may be
harder to crack. E.g. noise in a cpu or computer memory or
thermal noise in a bank of resistors or simular.
But I've noticed with my own machines that have "strong rng" based on
pools of garbage the O/S keeps on hand the rng are not all that random
and seem to have patches where they "get stuck" in long strings of
correlated bit.  A traditional cracker might try to find just parts of the
message that appear to have "pseudo" rather than "real" randomness and just
a few bits every now and then may give some useful knowlege of what
the message was about and who is talking/listening.
Some of this is just the modern spin on "lazy clerk". E.g. nomencaltor
codes were once real polar -- relating words and common numbers with
(usually) one of a selection of random numbers.
Code cracking relied sometimes on the fact the dumb human doing the
encrypting tended to memorize one or 2 choices for a given code number
and use it over and over making enough samples from that coder
easier to crack.

>
OTP is impossible to crack. Even if you use a brute-force search of the

...
 So are nomenclators. They dont use them any more because they are too easy.
 With AI's there are "people" that can 1 trillion times a second
decide whether something is possibly English or some other natural language.
With quantum computers the speed is essentially infinite.
So you're imagined reason for "impossibility" is all too flawed.
Just because someone wrote something that sounded reasonable
in the 1950s is no reason to accept it as true without a rigorous
mathematical proof. :)

Nadegda, you might want to let this one go. We have an old saying here:
you can lead a Horsell to water, but you can't make her drink.
*snicker*
--
"To explain the unknown by the known is a logical procedure; to explain
the known by the unknown is a form of theological lunacy." ~David Brooks
"I get fooled all the time by the constant hosiery parade
in here."                                                     ~Checkmate