Re: AI's take on my cipher...

On 7/7/2025 1:18 PM, Rich wrote:

Chris M. Thomasson <chris.m.thomasson.1@gmail.com> wrote:

Actually, it would be fun if they tried to crack it, akin to what was
done to AOB's ciphers. They got completely ripped apart!

 Compared to the complexity of your hmac cipher, the AOB cipherss look
about as simple to crack as the original ceaser cipher (they were
mildly more difficult due to AOB's obsfucation around "lots of
integers").
 That likely has a *whole lot* to do with why no takers on cracking it.
There are few posters left here in s.c in the first place, and for
those few who are left starting with 'hmac' likely moves the "knowledge
and effort to crack it" level well out of the range.

I can bolt on an HMAC to look for verification of an unaltered ciphertext just like any other cipher, but, I wanted to refrain from that for now. And allow for direct ciphertext decryption even if it was altered. My ciphers inherent design (radical diffusion and avalanche) should generate random garbage if even one bit was flipped in a ciphertext by Eve. Actually, the AI wants me to add in a "first pass" where HMAC is used to prevent a decryption from occurring if the check does not pass...
Now, a question, perhaps this should be in a different thread? The "validating" HMAC, not the HMAC my cipher uses internally, should use two different passwords? In other words password A comprised of TRNG bytes is used for actual encryption and decryption, then a another password of TRNG bytes is used for traditional HMAC verification that we all are familiar with? Make any sense to you? Actually, I can use two different algos, HMAC-512 for verify (aka, before any decryption occurs, and HMAC-384 for encrypt/decrypt) Sorry. My stomach is growling. Time to cook. Perhaps, humm. Well, I have everything for chicken piccata. Humm, well fwiw, here is a picture of my Salmon piccata exposed to one of my experimental vector fields:
(Electric Salmon Piccata v:(0.0.0))
https://youtu.be/8d98otjK6pk
Pretty interesting animation. :^)

You really need an 'inside mathematician friend' at the NSA to take a
look, but unless you have one of those at hand, trying to ask us to do
so is just shouting into the void.

I have a sneaking suspicion that you are 101% right. Sigh. Shit. Well... Humm... I can perhaps ask a favor from some of my main contacts in the AMS... They might know some people that can help me. I had to converse with them for the Calendar...