I've spent a fair amount of time on K4, and my conclusion is that it's simply a poor puzzle. At this point 24 of 97 characters have been revealed, and yet there's seemingly still not enough information pointing to how the known plaintext corresponds to the ciphertext. Over the decades everything reasonable has been tried and eliminated, which means the solution is likely to be unreasonable.
elahieh · 7m ago
I'm inclining to this too. I heard about it first on Usenet in the 90s, and started looking at it again seriously in June 2017 when I came across the Bauer paper.
But after the 2020 clues (another 13 letters), it became clear that it wasn't any single ACA cipher type, and it was probably something very difficult (because of K4's very low index of coincidence, i.e. 0.036 just below "random" 26-letter text at 1/26, plus the huge number of revealed plaintext letters "in place" i.e. letter-for-lettter correspondence).
That plausibly left a combination of two or more well-known cipher types, but if they were somewhat complex ciphers, the chance of solution would be rather remote.
Hence I always thought a "good" end to the puzzle would be like the book "Masquerade" by Kit Williams where the only guy in cahoots with the creator (Bamber Gascoigne) thought the initial puzzle was an unrealistic challenge, but Williams released clues which enabled two schoolteachers to solve it. So that part was satisfactory, even if hardly anybody remembers the solvers' names!
In contrast, the cribs for K4 haven't helped at all.
pjbk · 16m ago
Probably correct. Different from the other cyphers, the number of symbols is short, and correlating part of the plaintext that has been revealed gives poor measures for the full string length. It has been said that the other solutions are required to solve K4, so if the solution relies on something like character alignment, matrix coding or an even more convoluted permutation arrangement, this can look (or directly be) a one-time pad cypher which are arguably the most difficult to solve.
ars · 1h ago
I have discovered a truly marvelous solution to this code, which this text box is too small to contain.
But after the 2020 clues (another 13 letters), it became clear that it wasn't any single ACA cipher type, and it was probably something very difficult (because of K4's very low index of coincidence, i.e. 0.036 just below "random" 26-letter text at 1/26, plus the huge number of revealed plaintext letters "in place" i.e. letter-for-lettter correspondence).
That plausibly left a combination of two or more well-known cipher types, but if they were somewhat complex ciphers, the chance of solution would be rather remote.
Hence I always thought a "good" end to the puzzle would be like the book "Masquerade" by Kit Williams where the only guy in cahoots with the creator (Bamber Gascoigne) thought the initial puzzle was an unrealistic challenge, but Williams released clues which enabled two schoolteachers to solve it. So that part was satisfactory, even if hardly anybody remembers the solvers' names!
In contrast, the cribs for K4 haven't helped at all.