Prev: Dynamic Hill cipher
Next: PE Scrambler
From: J.D. on 2 May 2010 10:47
On May 1, 3:35 pm, Maaartin <grajc...(a)seznam.cz> wrote:
> On May 1, 7:55 pm, "J.D." <degolyer...(a)yahoo.com> wrote:
>
> > On May 1, 8:38 am, bmearns <mearn...(a)gmail.com> wrote:
> > Once you have your secret key and the session-specific seed, the
> > algorithm to produce the output has three stages -- initialization,
> > absorption, and finalization.  The algorithm uses 34 cards -- all 26
> > of the red cards, and the 8 black face cards (the ace, jack, queen and
> > king of both the spades and the clubs).
>
> It looks like all the exact kind if the black cards doesn't matter at
> all, they could be even all the same (e.g., 8 jokers), right?

Certainly. I just chose the face cards because I like them. They're
prettier than the plain old number cards. However it is entirely
possible to do most of the algorithm without even looking at the
fronts of the cards (only the finalization requires you to check and
see what color the cards are in order to remove the black cards).

>
> > Initialization:
>
> [snip]
>
> Here I assume that all black cards are indistinguishable. Either I'm
> wrong again, or the order of input is not very important, since the
> result is the same for different permutations, as the program shows:

For the algorithm as described that's right. However the order of
infusion during initialization will be relevant if at any point the
algorithm is amended to use the black-cards in a face dependent way
(such as the double-cut shuffle you recommended for finalization
procedure: e.g. h J i J t ==> i t h & black Jacks discarded, repeat
for all other black card pairs).

> > And why use the last 8 letters of the key in reverse order
> > during the initialization phase?).
>
> The reason for the reversal is obtaining some sort of equal influence
> on the output (I mean the sum of the "time distances" of insertions to
> the end is the same for the last 8 key letters), right?

That's correct.

>
> At the same time, you don't reverse the seed since the seed is not to
> be kept secret, so you save some work here, right?

I'm not sure what you mean.

> > But preliminary hand testing gives a pretty good rate of 3 to 4 letters processed per minute
>
> Does it include the fact that each letter gets processed twice?
> Assuming no and assuming 4 letters per minute, I come to 2*8/4=4
> minutes for the 8 letter seed and 18/4=4.5 minutes for the key.

Actually, I've gotten faster and faster. I can process an 18 letter
key (including all the time necessary for the initialization and
finalization steps, as well as absorbing the seed twice during the
absorption phase) in about 4 to 5 minutes. i.e. You can start with a
secret key, a seed, and a deck in the right starting order, and five
minutes later have your final output.

From: Maaartin on 2 May 2010 11:18
On May 2, 4:47 pm, "J.D." <degolyer...(a)yahoo.com> wrote:
> On May 1, 3:35 pm, Maaartin <grajc...(a)seznam.cz> wrote:
> > At the same time, you don't reverse the seed since the seed is not to
> > be kept secret, so you save some work here, right?
>
> I'm not sure what you mean.

I was probably wrong anyway. Another try: The last 8 key cards get fed
in last, so you compensate for it using them at the very beginning in
reversed order. There's no such need for the seed.

> > > But preliminary hand testing gives a pretty good rate of 3 to 4 letters processed per minute
> > Does it include the fact that each letter gets processed twice?
> > Assuming no and assuming 4 letters per minute, I come to 2*8/4=4
> > minutes for the 8 letter seed and 18/4=4.5 minutes for the key.
>
> Actually, I've gotten faster and faster.  I can process an 18 letter
> key (including all the time necessary for the initialization and
> finalization steps, as well as absorbing the seed twice during the
> absorption phase) in about 4 to 5 minutes.  i.e. You can start with a
> secret key, a seed, and a deck in the right starting order, and five
> minutes later have your final output.

That's quite good. I'll try it, too.
From: J.D. on 2 May 2010 11:50
On May 2, 11:18 am, Maaartin <grajc...(a)seznam.cz> wrote:
> On May 2, 4:47 pm, "J.D." <degolyer...(a)yahoo.com> wrote:
>
> > On May 1, 3:35 pm, Maaartin <grajc...(a)seznam.cz> wrote:
> > > At the same time, you don't reverse the seed since the seed is not to
> > > be kept secret, so you save some work here, right?
>
> > I'm not sure what you mean.
>
> I was probably wrong anyway. Another try: The last 8 key cards get fed
> in last, so you compensate for it using them at the very beginning in
> reversed order. There's no such need for the seed.
>

Well, you want to be sure every letter of the seed avalanches and has
approximately the same impact on the digest as every other letter of
the seed -- otherwise the same key but with slightly different seeds
(especially seeds that only differ in the last, weakest letter) might
produce outputs that are too similar.

First  |  Prev  | 
Pages: 9 10 11 12 13 14 15 16 17 18 19
Prev: Dynamic Hill cipher
Next: PE Scrambler