On Saturday, 13 November 2021 at 05:13:20 UTC, Rich wrote:
> To the extent that the author has coherently explained anything, it
> certianly looks that simple.
The most cogent explanations are in the documents currently
on his website. They only tell a tiny part of the story, though, and
they contradict his narrative here.
> Encryption appears to be (<+> == "tuple add" and <-> == "tuple
> subtract", Coo = "change of origin value"):
> C = P(as a tuple) <+> Coo
> And decrypt appears to be:
> P(as a tuple) = C <-> Coo
No - Wizz has it about as correct (as it will get without further
details) in the "vector math" thread.
AOB is adamant that his vectors are the same as the ones used
in physics, i.e. real-valued. His algorithm is explicitly integer with
GCD() usage central and a need to search for an integer solution
to one variable. How the real-valued algorithm is supposed to work
is a total mystery.
> The questions, so far unanswered, are whether Coo is a constant or not,
> and if it is a constant, over what timeframe is it constant. And if
> not a constant, then how/when does it change, and are the changes
There are three sets of numbers baked into the code, I, J, K, each being
a list of 14500 numbers. Why 14500, I have no idea. These numbers
were created something like 15 years ago, and it would appear that
changing them is sacrilege. AOB then uses his weak "scrambling"
algorithm[*] to create II from I, JJ, from J and KK from K. Coo would then
appear to be (A, B, C) where A, B, C are picked from (I, II, J, JJ, K, KK)
as part of the keying process. This requires changing the source.
Coo cycles through all 14500 triples thus created.
[*] the "scrambling" has three parameters:
1: offset - where to start.
2: length - a sublist length that will be reversed
3: repeat - how many times the above reversals will be applied.
With AOB's algorithm, offset + length*repeat must be less than 14500,
(I get about 10^9 choices) making searching all possibilities practical.
The possible permutations of the list is 14500! with a proper shuffle.
This is about 1.111274225*10^54045 permutations.