are superior to the "visualized" fingerprints: They are harder to imitate by an attacker.
However, I realized this argument is invalid: If you -- as a human -- compare two fingerprints by just looking at parts of the texts/image, you haven't done a full verification and thus are vulnerable. Therefore, in principle, the kind of key authentification how it is done in modern smartphone messengers (making a photo of a visual key (encoded in a QR code or similar) where the smartphone then verifies every single bit) is the right way to go when it comes to fingerprint verification. Humans are too impatient and too inaccurate for this job.
I went into this expecting to see a bunch of precompiled randomarts manually tiled to create a composite picture.
I was not expecting to learn the simple rules that underpin how randomart is generated for a hash and then to see the author discover hashes that roughly match the goal.
In crypto, you can dedicate a modest amount of computing power to get a "vanity address" -- hash enough wallet keys and you get an address that starts with 4 or 5 letters of your choosing.
This is basically the same thing for ssh keys... get something that vaguely looks like some shape you desire.
However, I realized this argument is invalid: If you -- as a human -- compare two fingerprints by just looking at parts of the texts/image, you haven't done a full verification and thus are vulnerable. Therefore, in principle, the kind of key authentification how it is done in modern smartphone messengers (making a photo of a visual key (encoded in a QR code or similar) where the smartphone then verifies every single bit) is the right way to go when it comes to fingerprint verification. Humans are too impatient and too inaccurate for this job.