Out of the 12 clues, we’ve solved (or at least have good guesses for) 9 of them, and have 3 left to go.
With my typical adventure games, if there’s any kind of ambiguity in a puzzle answer, there is immediate feedback from the game itself. For a static puzzle, it’s hard to have airtight certainty in answer.
In a puzzle book, you can, of course, immediately check the back of the book. The problem with Krakit (in a contest context) is that there is no back of the book, and while the rules aren’t specific, I’m guessing entrants did not get feedback in order to correct a wrong submission.
This is a common problem amongst puzzle hunts; where I think we can say we have answers with confidence is where there’s some manner of “cryptography” (in a loose sense) involved, where some operation is required to “decipher” material into a sensible text it would take a seriously major coincidence to have a valid answer be a fake.
For example, in the comments last time Lucian solved the cryptarithmetic puzzle above.
9567 + 1085 = 10652
That’s eight digits accounted for, so I don’t know why the next bit only lists numbers 0-6, which is only seven?
That is, part 2…
… has 0 through 6 repeat, but not 7, 8 or 9. This immediately led me to realize that the only important number is 10652. But what to do then?
The way words are sort-of spelled out suggested to me we simply had a track that needed to be followed. After a few false starts I realized if you start in the upper left, and pick out 1-0-6-5-2-1-0-6-5-2-etc in order, you get the letters mexic/ochih/uahua, where the path resets partway through to the “inner path” (and the number jumps from 2 to 3).
There are all sorts of other ways the clue could have been interpreted, but the absolutely clear spelled-out answer indicates this is a self-confirming static puzzle.
This unfortunately isn’t true for clue 10:
My best guess here is that this refers to Pyotr Ilyich Tchaikovsky and the 1812 Overture (an opus 49). The “to go forward, think retreat” can be interpreted as Napoleon’s retreat and turn around at Borodino, Russia as celebrated in the music. Will Moczarski also pointed out based on the way Russian names work it can be interpreted as Pyotr, son of Ilya, Tchaikovsky.
Still, this doesn’t feel fully certain, and there are some other ideas from the comments (like MendelsSOHN). The lack of an immediate meaning to the retreat part of the clue isn’t the same thing as saying the answer we have is airtight.
We’re down to three completely unsolved puzzles. Two seem like they’d be self-confirming, but that doesn’t mean they’ll be easy to crack.
(I had what felt like a good crack: assume that we mean two-digit hexadecimal, but are only listing the second digit. That is, we are using hex — “curse” — but only taking half of the digits normally needed. The way characters in both regular ASCII and the ZX81 character set works makes it so there are a limited number of choices then for each digit, so it’d be easy to decipher a code. It still just makes nonsense, though.)
(Some guesses are in the comments from last time, but nothing I’d say is substantial. The fact “HEAR GREEN BALLAD” is in quotes seems noteworthy.)
(I still feel like the stars have to be the key here, and a lot of the letters are ignorable; this would follow the same pattern as the SEND MORE MONEY puzzle insofar as that one also has many “superfluous” letters and we are just trying to pick out the correct ones.)
I’m going to take a Krakit break and play some normal adventure games, then unseal the answers next week. Please keep sending your thoughts in the comments! That hopefully gives us enough time to reach enlightenment before we try to do back-solving.
Renga in Blue 
(I’d just like to add that “SEND + MORE = MONEY” isn’t just a solvable cryptarithm, it is a Classic Puzzle, probably the canonical example of its form. Wikipedia tells me it was first published in 1924: https://en.wikipedia.org/wiki/Verbal_arithmetic)
One thought on 12: All of the digits are on the periphery — on the left and bottom edges, in fact — except for the zeroes. On that basis, maybe they’re not actually zeroes! Maybe they’re the Ø used in Danish and Norwegian, including in city names such as Gjøvik and Hørsholm.
I also just posted this in your last entry:
HEAR GREEN BALLAD is an anagram for Ballaghaderreen, where William Partridge died. He was part of the Easter Rebellion, which took place in 1916. I’m not sure about the “math” part of the puzzle. Maybe something to do with REST IN (peace) since he died there. https://en.wikipedia.org/wiki/William_Partridge_(Irish_revolutionary)
So, I would say Ireland (country), Ballaghaderreen (city), 1916 (number).
nice! I’m guessing the math probably was supposed to lead a person with a clue for the anagram, but that’s too good to be a coincidence.
Maybe the two-letter pairs represent the provinces of Ireland somehow? MUnster, leINster, … but SL for uLSter is a bit of a stretch.
It’s very tantalizingly close. If you put two MUs, two SLs, two INs, and three RESTs together, you almost get Ulster, Leinster, and Munster; there is one M, one I, and two S’s left over, and one E short. Then Connacht is the fourth province where Ballaghadereen is. But it doesn’t quite add up!
In clue 6, the hex digits that aren’t used in the grid are 49 BCE, which is the date of Caesar’s crossing of the Rubicon and the start of the resulting civil war.
huh, the BCE possibility is super interesting
It wasn’t really standard use by ’82 but it certainly was available.
Could they be wanting the player to really ignore that much text? hmmm
I was taught BCE in Hebrew school before ’82, but it was definitely a Jewish thing.
also, if this is right, it would probably be Savignano, Italy, which seems to be the modern-day place where the crossing supposedly happened.
For the hex clue, if the gray squares are meant to stand for missing digits, then we have an odd number of them—which is both literally and metaphorically odd, because hex digits tend to come in pairs (one hex digit encodes four bits or half a byte).
If they’re grouped into triplets (which gives us a whole number of groups), then we start with 1#7 twice, which could be meaningful? Or could just be coincidence.
It occurred to me that the ZX81 tape format doesn’t include a checksum, so there might be corruption in the clues. I’ve just compared the screenshots with the Spectrum 16k and 48k tapes from World of Spectrum, and found one difference: in clue 12, the second line starts “H0B” here, but it’s “H05” in both Spectrum versions. (The Spectrum version also has proper apostrophes; the ZX81 fakes them with commas as there’s no ‘ in the character set.)
curious! it also makes me wonder if it wasn’t a dumping error but an error on the company’s end that they fixed a subtle issue
that particular letter doesn’t get used in any of the patterns I’ve tried though.
If H05 is correct, then it’s the only nonzero digit that isn’t in the periphery.