i have been thinking about the unsolved cicada 3301 liber primus puzzle lately. i was never involved in solving the original puzzle, but years ago i spent a while trying to solve the liber primus

there are two notable properties of the unsolved sections that the rest of the liber primus does not possess:

1) unexpectedly low repeated rune bigram frequency

2) the DJV-BEI (AE) repeated sequence

---

the first property proves for certain, in my opinion, that the method for encyphering each character involves the previous character in some way

there is no frequency pattern observed if you consider any other 'bigrams' of paired letters (i, i + 2) or (i, i + 3), etc., the pattern is only seen in the case of (i, i + 1)

though, we don't know what exactly about the previous character is being considered

it could be:

a) the previous plaintext character

b) the previous character in ciphertext (like a rolling cipher)

c) some other variable that is updated by each character (also a rolling cipher)

a) and b) are interesting possibilities, c) seems arbitrary and equivalent to b)

the main difference between a) and both b)/c) is that if only the previous plaintext character is considered, the plaintext input to the cipher is limited to two characters

if a rolling cipher is used, every plaintext character so far is involved as an input, though chaos compounds

while visually it seems that only the previous character has an influence, it seems perfectly possible to me that a rolling cipher is used and other potential patterns are obscured by the small amount of ciphertext and non-randomness in the plaintext

in the past i investigated and was satisfied that word/line breaks aren't involved in the distribution of repeated rune bigrams

---

the second property suggests to me that the cipher had an unintentional case of the same output resulting from the same two three-letter words repeated at different positions in the plaintext

this could result from a 'previous plaintext character' cipher if the rune before the D was the same in each case, but in the case of a rolling cipher it would be a 1/29 stroke of luck

it bothers me a great amount that one of the DJV-BEIs is at the end of the unsolved section. this could actually be a coincidence, but i don't know

i would be more satisfied, too, if one of them was at the start of the section instead of the end. it makes me wonder if decoding the ciphertext backwards or in some strange way may be necessary

i tried to theorise about the specific position of each DJV-BEI in the past but got nowhere. i will note that i feel it's slightly convenient and if there was one fewer runes in the second half of the unsolved section that would become extremely convenient. might rehash this later

i do not think any of the ciphertext is redundant, the word/line breaks and other images/codes admist the ciphertext clearly suggest meaning

i should make a word frequency table from the previous solved runeglish sections and possibly guess what DJV-BEI could be. two words appearing together twice is at least *something*

we must also not discount the possibility of a stream cipher, since a stream of the totients of the prime numbers is used in 56.jpg

however, the DJV-BEI repeats suggest to me that if a stream cipher is used, it must repeat at some point. this becomes less likely the longer we propose the stream to be. i don't think it's possible that a long/infinite stream was used, it would be too convenient for the stream to line up with identical words in the plaintext

---

i am most interested in coming up with ciphers and looking at the bigram patterns that result. i won't tire myself out trying to find a solution to the puzzle

i should add images later. in time i may make a full page dedicated to the puzzle