000000000fdf0c619cd8e0d512c7e2c0da5a5808e60f12f1e0d01522d2986a51
The nonexistent cat wants me to learn more about GANs, mentioned below,
so that I can generate for myself an unbounded supply of nonexistent kitties. 😺

Free images of Lauda!

jamyr and uchegod-21, thanks for your kind words. A clarification for those now joining us:

Quote from: uchegod-21 on Today at 10:53:08 PM

But when Cyrus launched Bitcointalk YouTube channel, I was following the channel until he did Bitcointalk quiz. One of the questions was. Who among the following users cannot return to the forum?

There was couple of names which include Lauda, Vod and others. Cyrus said the correct answer was Lauda, that after leaving the forum he/she was banned. Today I stumbled of the last post of Lauda.

Super-hyper-ultra-banned, not with an ordinary account ban. Lauda got the same ban as Satoshi, whose account is totally disabled to help protect against potential security compromises. From earlier in this thread:

Quote from: theymos on October 19, 2020, 01:50:51 PM

Quote from: Lauda on October 19, 2020, 05:29:24 AM

I will request that theymos ban the "Lauda" account (u=101872)

Done. Lauda is banned in the same way as satoshi, such that it isn't possible to even log into the account anymore.

Continuing discussion of the Lauda Memorial Crypto-Puzzle:

Quote from: garlonicon on Today at 06:44:12 PM

Quote from: nullius on July 04, 2022, 09:40:11 PM

Any new ideas for how to solve the puzzle?

That's the problem: there are too many options.

In honour of Lauda’s thread, it is my pleasure to introduce even in a very informal way some thoughts that I have had on what I call Kolmogorov v. Kerckhoff.

My ideas about this are handwavy, not rigorous—and I know the difference. Nonetheless, I have observed that in practice, the set of all programs with relatively low Kolmogorov complexity is, itself, a keyspace. It is an ill-defined keyspace, which would make many cryptographers and security wonks shudder with horror; but it is a type of keyspace, or rather, I conjecture (*waves hands*) that it could be made one.

Sets of programs have been more rigorously turned into keyspaces of sorts. For instance, Monero’s RandomX POW mining algorithm meets this description in a different way.

From another perspective, I think of it as a crypto-key version of naming the biggest number. Since I specified that I used only standard things, and the solution can be quite easily programmed with readily available tools, the upper bound on the set of possible solutions (= “keys”) is expanded by adding new primitives, new standards, new tools. By an analogy that is to me more philosophical than mathematical, I see that like inventing new notations for expressing bigger numbers.

Anyway, I am using a blackbox program as the secret quasi-key. The program has publicly known inputs. The entire security of the system relies on keeping the program secret. But this, in turn, seems like the nullification of Kerckhoffs’s principle. That is, of course, a security doctrine, not a mathematical rule.

Lest I discourage anyone from attempting the puzzle, I emphasize that I laid out plenty of clues. At least, I think I did. So...

Quote from: garlonicon on Today at 06:44:12 PM

Every time when there are some 32 bytes, then you can put there any hash, so the whole process can be unsafe, but the whole security comes from the algorithm.

...I would not be so cruel! Hashing some data in weird ways with no hint of the inputs would give others an exercise in frustration, not a fun puzzle.

I probably used hashes, and some other things. Hints were dropped about what and how—subtle hints, but sufficient to get me on the right track if I myself ever forget how to reconstruct the private key. Of course, that would simply be jogging my memory. Will others find the hints?

Quote from: garlonicon on Today at 06:44:12 PM

So, I have many ideas, [...]

Good luck. 😼

Quote from: garlonicon on Today at 06:44:12 PM

Also, when it comes to mining transactions, it can be just some kind of timelock: https://www.gwern.net/Self-decrypting-files

Thanks. That’s a good page; I have seen it before, so I guess we read some similar things.

It does not mention that Solana uses a hash-based timelock in its consensus algorithm; they call it a VDF (Verifiable Delay Function), even though it is not a real VDF.

Gwern also has information and experimental writeups on GANs, including StyleGAN. The thiscatdoesnotexist.com site that I have been using as an official source of Lauda images is run by the authors of StyleGAN2. I have been thinking to use Gwern’s info as a general guide to setting up my own Lauda image generator, but I lack the hardware for this type of project (GPUs, etc.). Someday...

GANs for creating nonexistent cats are discussed earlier in this thread, with references. It is my way of inducing people to read the thread, so that they fall under the witch’s spell. ;-)