K4 is complete:…Kind of. Okay look, ‘K4 is finished’ sounds much better than ‘50%+ of the known plaintext has been recovered, involving half a dozen steps and leaving us likely one or two steps away from solving the other 50%’ Just stick with me, because honestly, when was the last time anyone had this kind of progress with this stuff? Due to the concerns over magic numbers, we’ll be rendering some simplified animations to demo how you’d do this on pen and paper <3.
K4 Current Reveal:
DONOTLOCATETHEBERLINCLOC
JSEEKNOWTHEWESTWALLOF[XXX]^IMPORTANT NOTE FOR PEOPLE TRYING TO UNCOVER MORE TEXT: Due to the way K4 is set up and even with the ‘clues’ given K4 has the ability to spiral out into a plaintext longer than 96/97 characters [that is one of the neatest things about this particular encryption path]. As a result I am not going to slide these reveals into the raw encryption to keep from causing confusion, since there are like 5 steps you have to take before you can even pull this stuff out. The lagging trail [XXX] are likely nulls, not something unusual but it is weird to see them in triplicate like this, there might be a further interweaving mechanic between the quadrants that allow this to be solved, but that won’t become clearer until the other 50%+ is pulled at.
Forewarning:
Am I an AI bro? Probably could be classified as one, yes but let’s go with tech goblin today. Am I new to cryptography? Absolutely, only been on Kryptos for a few months. Does that make the information presented any less real and verifiable? No. Follow the steps yourself on pen and paper, run the blind code to get the mathematical answers, but don’t just take my word for it. There’s a lot of info here, but I am hoping you’ll all find it approachable and clear.
What was uncovered:
> How K1 password was derived using the morse code
> How K2 password was derived using K1 [which reinforced what we already know about K2 and K3 working together]
> K3 was solved using information from K2 and a clue from the morse code [we knew this already it just confirmed that all puzzles are self referential]
> K4 involves layering everything we’ve seen before: section, quagIII, movement, and introduced a new folding mechanic
> Jim was wrong, this is solvable with AI, and it can be brute forced but neither happen without context
How I figured it out:
[You didn’t expect a straightforward answer here did you? I’ll make this one easy: You only need to tweak the solution to K1]
YECRNGLYDKXVFGWFBTEHOYGPXFOJICJBLOFNICKXFKOVLVBBWOIGCZTJYDNDLGALCCHHZNOJRDKSVETCIASKMFNTBTQYLPTTPDOKMGXOVDYGHMVWXPMJDBGGKFPMHZSJUPLBLTLQILCDFVBQRDDEXEXGUKYDCIAWFQZLIHUHDLLABYWDYGHFSKCRET.
Github for verification:
https://github.com/ElodineOfficial/KrackersAndCheese <Slider proves intended solutions behind K1+K2
https://github.com/ElodineOfficial/KryptosK4 < Solve for K4, minus step7 [that’s a theory for the final 50% missing solve, and is not confirmed, which is why step6 is missing]
The non-technical overview:
Alrighty, let’s make this part sweet and simple english. I got to thinking about Sanborn’s general complaint that the Kryptos community didn’t understand his intended solutions vs the use of brute force computing. So I took a step back to revisit K0 and K1. What I found was a hidden substring in the morse code where ‘Virtually invisible’ is actually a 13 letter english passcode ‘ALLYINVISIBLE’. You can take that passcode, and the word KRYPTOS as a key, and slide across the encrypted text of K1. This yields the latin passphrase: PALIMPSEST. Which we then combine with the KRYPTOS key and the same decryption method we used before [QuagIII] with the new password PALIMSEST to get the solution to K1. This shows us that we were intended to solve the morse code of K0 and use it as a spyglass for K1.
With me so far? Awesome! So now Sanborn expects you to have a method, and one solution of plain text. We can take the first 13 character string from K1 then ‘BETWEENSUBTLE’ and try this as our new english passcode to slide over the the K2 encrypted text with the other word still being KRYPTOS. This yields [you guessed it] ABSCISSA, which we combine with our consistent key of KRYPTOS and the same method [QuagIII still!] to get the decrypted K2 text. You can actually use a ton of 13 letter phrases to get these 10 letter keys, check out the github for some files that do this for you!
That’s a lot of progress! And an actual workflow that connects stages one by one in an artistic manner that never requires any heavy computing. Everything is all still easily yielded through pen and paper. It’s at this point where things get weird. Because there’s somewhat of a gap in how we get from K2 to K3. YOU know that K3 uses transposition, specifically starting at T [the morse code tells us ‘T is your position’] and using a 192 step each time. But the big names digging into this really made a complex solution involving padding, which is also viable [because K3 is just a word jumble], but never actually connected with the INTENTION behind the solution. It’s currently debated on how that 192 plays into this but that’s neither here nor there right now. I am guessing the coordinates from K2 which are listed as being about 175 feet from Kryptos were actually intended to be 192 as this is the only solution with a built in clue and an artistic build.
We’ve got some great self referential connections all hinting that we don’t need anything else outside of the physical art to do this. But that doesn’t explain all the spelling errors. Spelling errors that when laid out begin Q-U-A…like a QUADRANT or a QUARTER. Considering Sanborn refuses to comment on them [he does this anytime someone starts poking in the right direction] then I’d say that’s a pretty solid angle to explore.
And it’s one that lines up PERFECTLY with the extensive math we’ve seen people crunch around K4. [https://glthr.com/a-fresh-perspective-on-kryptos-k4] Basically the thought is that the IOC scores [how close to english is this] are different in different sections of K4 depending on how it is split and mirrored. There’s also a lot of W’s, and one right in the center. Weird that those W’s don’t disturb any of our revealed cribs. Weird that when we subtract the W in the center [technical note: this is not actually a subtraction that will muck up the math down the road, we’re actually transforming the W to a crosshair] we get four ORDERLY QUADRANTS.
O B K R U O | V Q Q P R N
X O G H U L | G K S S O T
B S O L I F | W T Q S J Q
B B W F L R | S S E K Z Z
────────┼────────────
A T J K L U | K W G D K Z
D I A W I N | X T J C D I
F B N Y P V | G K U H U A
T T M Z F P | U E K C A R
Things are getting a little shaky now, we’re losing a lot of the concrete directions and starting to throw things at the wall to see what sticks. But something did. We need a lot of ingredients for this next step but we have all of them to peel back the first layer of K4. [No revealed text yet, we’ve got to prep the space to get the true encrypted text!] We take the latin single use keys from K1 and K2, the quadrants, and a column based cesar transposition…and we get: Directions. One neatly laid in each quadrant. EAST, EAST, EAST, WEST. With KRYPTOS, ABS, PALM, and PALM returning these respective directions. One key for each quadrant, one direction for each quadrant.
NW: SBXASBXYEASTFUZLRRJSUSGU
NE: JNREASTZDLTFBXUJLBFJDALS
SW: LZYAMAFHKEASTFYBKQCVPIQA
SE: WEDKGSSXZSQQEKQNWESTDIRR
All of this is done with the intended clues. All of this can be done on pen and paper. It’s all artistic direction, simple shifts, and most importantly: Looking for the hidden messages. But we all already knew that, this is cryptography afterall! Speaking of hidden messages K3 and Carter right? Weird that it talks about shining a light through a gate that creates a convenient grid. Oh wait– if we use the password LICHT on a quadrant we start to see IOC scores in line with english!
D L O C I T
R C O O A E
E N T E N T
C L B O H L
It’s almost like k4 follows k3 where we’re sticking in a light through the grid. But we can’t stop now! The passwords palimpsest and palimpsest were used on the same section with nearly matching IOC scores I wonder if– yeah they’re over and under writing. And can be folded in two different directions. This has to be combined with a knights transposition cipher [another word jumble] to reveal 48/96 fully decoded characters. How would we know its a Knights walk? Well if we look back at the tableau you’ll notice an L here sticks out like a sore thumb, and the position makes it a red herring for hill climbing options if you don’t know this particular type of decryption method exists.
DONOTLOCATETHEBERLINCLOC
We even have a traditionally misspelled word! Then we repeat for the other half. So we are half way through, with Berlin, clock, and a number of other english words on our side but only half the final text you saw at the start. And that’s k4 or at least half of it. Matching the output described by the creator. Following only internal clues. Using blind code to confirm. Clearly Kryptos isn’t solved yet, but I will need a break before I do anymore investigation. Crazy rollarcoaster right? The sudden halt at this point has given me some whiplash.
Technical overview:
https://github.com/ElodineOfficial/KryptosK4/blob/main/README.md
Where the we’ve been going wrong [myself included]:
> Focused so hard on LATIN that we didn’t realize there were english passcodes [k4 showed us there was a German one too]
> Focused so hard on K4 that the methodology behind K1 and K2 solutions were glossed by, afterall, they’d been solved
> Not enough focus on the in puzzle clues, [with respect I’ve seen some people try and tie in the bible or books outside of the quoted Carter Howard passage and that’s just not going to go anywhere, this is a contextually heavy puzzle but it is all internal to itself and general cultural knowledge]
> Not enough focus on the access of information during the time, as Sanborn wasn’t making this with internet popularity in mind. He was thinking about local libraries, internal clues, and perhaps even some pop culture osmosis. Pen and paper were always the focus of this solve.
> We assume corrections are corrections, instead of clues that’ve been messed up so badly that the puzzle may be unsolved without correction [XIDBYROWS vs XLAYERTWO]. Sanborn has said this mixup was a highly unfortunate probability event. I can’t stress enough how every little thing here is intentional even when it seems insignificant. But artist intention and perfection in execution are wholly different, we can’t treat this as an infallible puzzle. This ended up being a hint to how the palimpsest in k4 functioned. So it was critical to correct, I do appreciate Sanborn doing that.
> The end text might not be 96-97 characters, and that alone we can thank Edward Scheidt for that. He really did save the best for last, making a palimpsest that not only folds in on itself but might be able to unfold in a different manner that creates even more than the expected text. Of course, I’d expect nothing less from a man with his credentials! It has been so cool to see this thing sort of ‘open up’ while being so far away from a full answer all at the same time.
On that note. Allow me to be the bad guy for a moment: the Kryptos community in general is composed of good faith actors, and that isn’t always how decryption, hacking, or puzzle breaking happens. This is a social engineering problem, where Sanborn himself has to be considered as he has done so many self referential clues throughout his entire career [not uncommon, I do this too] and we see them begin to bleed through here as well. By having a better understanding of the personal knowledge base, we get a greater idea of what would’ve made sense to use at the time as a first, second, or even obscure third option. The better you understand the man [ethical considerations aside] the easier this puzzle is to undo. K4 is being solved by being the bad guy. By being the hacker, the social engineer, and the goblin. So if you’re going to solve the other half, grab your grey hat!
What it all means:
If you made it this far maybe you are wondering how the hell people were supposed to figure it out. You weren’t intended to, your parents and grandparents though? They’d remember what a big deal Carter’s expedition was. How Tutmania swept Berlin, the wall separating the city into half [and even further within those halves to create quadrants] where west and east wasn’t just about geographical locations but centralized power. Sanborn remembers this, he was probably there to see it all. And a lot of folks who were there had to walk by that funny Uhr. Mengenlehreuhr wasn’t some random clock, it predated the final stop of the Tutmania tour and was considered a Sci-fi marvel. Being known as the ‘Set-Clock’ was just a happy accident that tied it even further into the Egyptian craze sweeping Berlin. There were even postcards that had the clock printed over many artifacts! Pictures of them haven’t surfaced in any catalogs [I could only find text blurbs about them], but seeing how Sanborn has a bunch of this stuff framed in his home [I wish I didn’t know so much about this guy], I bet you he has one. I bet it served as the inspiration for this whole puzzle. Even if it didn’t, the cultural tells are all there! And this sort of homogony of understanding or experience is something lost in the digital era. This is literally a lucid memory that Sanborn enjoys, of his time learning and exploring. The last half of K4 likely involves that Set-Clock/Uhr he passed by while enjoying being in Berlin during Tutmania. Regardless, Sanborn acted as an awful puzzlemaster here, and only ever gave out clues that could be useful after you’d ALREADY discovered them yourself. I think this is probably a failure in approaching things from the cryptographers position vs that of the artist [not the cryptographic designer, as much of that goes to Mister Edward Scheidt and for his part his work is AMAZING, I have so much respect for this guy.]. Maybe that was the intention, but when you pair that with a donation system for answering I start doubting that Sanborn actually wants anyone to solve this and hoped it would be some great mystery to be studied through the ages.
Backstory you don’t care about:
I only started working on this project out of spite and I’m only continuing out of spite. I initially started working on this because I’m an AI educator and some of my students had asked if Sanborn was right about AI’s inability to break K4. And he wasn’t, don’t get me wrong K4 hasn’t been cracked by LLM alone [well, now that this info is public that will change over a few months to a year], but saying that algorithmic machines built to emulate people couldn’t figure this out, or that custom systems made with the intent to study Kryptos couldn’t punch holes into it: that’s just false. This whole thread is proof of that. I won’t be giving specifics, because no free advertising on my watch. I’ll just say this took multiple systems, far more than just LLMs to bust open.
For a man who used to go on about how he was so proud that his art could sit and ‘make people think for even ten minutes’, this sucked. Mister Sanborn is a man who has made it clear over the years that the brute force computing methods used to solve K1 and K2 are not his intended solution, and he’s been somewhat sad to see that his actual design has never been figured out. Until, you know, obviously right now. I messaged Mister Sanborn only to thank him for the puzzle, and to let him know that I was confident I had found his intended solutions, and that I’d hope they’d help to pull at some new strings to unravel K4. Happy to leave things there and go back to being a casual.
I was honestly shocked to get any email back, let alone one within a few hours. I’d made it clear I just wanted to pass on my thanks and hoped the community could take that intended solution and put it to good use. At which point I was informed that not even a thank you could be passed on without the Kryptos Fee System [yea, still not gonna submit for the official solve but I am sure someone will send this in for confirmation!]. That got me ticked off enough to kick the can again only to find it actually went somewhere this time. In my email I said it could be weeks or years to move from understanding how K1 and K2 passwords were built to solving K4.
It didn’t take years, or at least, I’m hoping it won’t anymore now that we’ve gotten this far. But remember! We get by with a little help from our friends. It wasn’t a solo effort and this is where I get to brag a little! The hours I spent on this were insane and my Mother, who I rarely get to see due to her failing health, made it to me again this summer. She said solving Kryptos with me was going to be her last great adventure. She is a former pastor and teacher with a PHD that crunched clues on this like it was nothing. Probably because she ended up taking a lot of the same courses Sanborn did. I guess I’m mainly turning over the last 50% because I don’t want to see the adventure end before the puzzle is solved. One of the first lessons I introduce students to is the idea of ‘one nerd in the basement’ [it’s me, I’m the nerd this time] where if you’re able to execute a project at home, it is likely viable for anyone. And if I’m doing this, I know you can too. I’ll be back in a week to start hitting this again but I hope I come back to ya’ll having cracked the other side of this open! And if not I’ve lined up a ton of experiments, but life is calling me for at least the next week!
K4 is solved, at least in part. And the glory goes to the goblins. See ya nerds.