( October 2023: Taking the 'form' offline and adding the Popular Mechanics 2020 link )

February 2020 Popular Mechanics article by Courtney Linder
Nobody Has Solved This Cryptographic Puzzle for 30 Years. Think You Can?


Kryptos K4 Tool (scratchpad) - April 25, 2011
script-php killed July 31, 2022 - long obsolete and useless
Created as my own personal graph paper to take a stab at K4. Click/Tap one of the already tried keywords or use the box(es) to try a new one.

All notes, links and details after the solution (scratch) area.

K4 - Original - TT(font) and right-aligned
OBKR
UOXOGHULBSOLIFBBWFLRVQQPRNGKSSO
TWTQSJQSSEKZZWATJKLUDIAWINFBNYP
VTTMZFPKWGDKZXTJCDIGKUHUAUEKCAR

Original Straight Line
OBKRUOXOGHULBSOLIFBBWFLRVQQPRNGKSSOTWTQSJQSSEKZZWATJKLUDIAWINFBNYPVTTMZFPKWGDKZXTJCDIGKUHUAUEKCAR


KEY from FORM (cleaned) = NOKEYWORD

Form offline July 31, 2022

Keywords already tried: 54005


Decoded with NOKEYWORD straight line K1/K2 method
FPKJNBNTPUJLWTBFHYIPWRITLNDDVNYXCRTQTZQUHWRORAQZFOAEZAYOIVUMHEXKWPEYAGXYDUWYBPQWQWTDTEPJGEHJEIADV
Simple Transposition L-R/T-L (12x8) [*see note x-posit]
FPKJNBNT
PUJLWTBF
HYIPWRIT
LNDDVNYX
CRTQTZQU
HWRORAQZ
FOAEZAYO
IVUMHEXK
WPEYAGXY
DUWYBPQW
QWTDTEPJ
GEHJEIAD
V
 FPHLCHFI
WDQGPUYN
RWOVPUWE
KJIDTRAU
EWTHJLPD
QOEMYYDJ
NWWVTRZH
ABTEBTRN
ZAAEGPEI
NBIYQQYX
XQPATFTX
UZOKYWJD
V
FPHLCHFIWDQGPUYNRWOVPUWEKJIDTRAUEWTHJLPDQOEMYYDJNWWVTRZHABTEBTRNZAAEGPEINBIYQQYXXQPATFTXUZOKYWJDV




Decoded with NOKEYWORD straight line against KRYPTOS 'screen'
SUQXXJKOKBHHEIJDYJVUNITFIDRQNJJCKSOBKQKCSYSPINLPTJICJOIWFDZTFVFGOTSGIEIJQMNJMELHBDVALPEHXSUHBWLLN
Simple Transposition L-R/T-L (12x8) [*see note x-posit]
SUQXXJKO
KBHHEIJD
YJVUNITF
IDRQNJJC
KSOBKQKC
SYSPINLP
TJICJOIW
FDZTFVFG
OTSGIEIJ
QMNJMELH
BDVALPEH
XSUHBWLL
N
 SKYIKSTF
OQBXUBJD
SYJDTMDS
QHVROSIZ
SNVUXHUQ
BPCTGJAH
XENNKIJF
IMLBJIIJ
QNOVEEPW
KJTJKLIF
ILELODFC
CPWGJHHL
N
SKYIKSTFOQBXUBJDSYJDTMDSQHVROSIZSNVUXHUQBPCTGJAHXENNKIJFIMLBJIIJQNOVEEPWKJTJKLIFILELODFCCPWGJHHLN




K1 decode using this program - debugging.
BETWEENSUBTLESHADINGANDTHEABSENCEOFLIGHTLIESTHENUANCEOFIQLUSION




* * * Notes, Links and details (descending date order - newest info at the top) * * *
July 31, 2022 Finally took the php offline (commented out). Over the last decade the sculptor gave out two 'clues'. google searching has much more info.
Dec 17, 2011 More (lunatic) ramblings on CIA's Kryptos: pdf format
May 4, 2011Oops again. More Final Notes:
  • I am not trying to 'steal' the work of anyone smart enough to figure out the magic word(s) (if there are any) to solve K4 using this method. This script (php) does store tried keywords, but only to keep 'the rest of us' (or ourselves) from re-testing. I offered the source code (below) so anyone using this can 'see for themselves' that there is no trickery here - it is only a 'scratch web-page.' This script does not do any 'secret' or 'backdoor' processing, and it does no notification to anyone other than the viewer about the output of the K1/K2 style decoding.
  • If you solve K4 (figure out the magic word and possible transposition) using this and want to protect yourself - print or save the output webpage, go immediately to http://kryptosclue.com/ and validate your solution, then contact me (see below) and I will replace this page with your solution (if you want) or take it 'offline' until you get your published credit for solving K4.

In the overnight I was hit with a few suggestions and comments regarding this:
  • K4 might be double (or triple or more) encoded. I am aware of that. If someone out there wants to 'branch this code' and make a multi-decoder out of it, OK by me. I don't have the time right now for those kinds of things. I still can't figure out how to get one keyword (PALIMPSEST and ABSCISSA) de-puzzled out of the available clues; I would make myself insane trying to figure out how to multi-multi guess at and then process/decode against all the possible variations (K1/K2 alphabet, plain, Kryptos Tableau, etc.), with shifting and transposing...
  • How can I trust you won't claim my solution? Addressed above. You don't have to take my word for it, though it's good - I'll give you the sources and you can setup your own version on your own computer/server.
  • It says NOKEYWORD but the output is not K4 original. Sorry about that, it is actually decoding using NOKEYWORD as a keyword. I was going to make it skip the decode function if blank, but I just ran out of time/energy for it. My 8x12 transposition grid using the original K4 lettering is not helpful anyway - if K4 is 'just' mixed up letters (not encoded/encrypted) then this page does not matter/help anyway. To see the K1/K2 transposition without decoding: click/tap or enter K as a keyword (because of the 'stacking under K' method Sanborn used, the K alphabet decodes to encoded text).
  • The K1 'debugging' text never changes. It shouldn't, I left it there to make sure there was nothing wrong with the K1/K2 decode method (function). If you see anything other than 'BETWEEN...IQLUSION' in there then something is horribly wrong with this page/script/php. Please let me know if that happens.
  • K4 might be a [complex] shift. I considered this also, 2 or more shifted alphabets (Caesar) run in sequence. I honestly (should never assume, but I did) thought that if K4 is 'just a shift/replacement' the many smarter-than-me decoders would have solved it many years ago. The frequency and pattern (and big computational solutions) to K1/K2 provided by Gillogly and Stein kind of (pretty much) made me think that simple-shifting had been ruled out long ago. In reality, this theory is basically what solved K1/K2 anyway, just using modified alphabets (KRYPTOSABC...) and a keyword for 'out of sequence' sequencing. I am not going to write a webpage to 'run' a plain alphabet shift against K4, I really/truly (assume) believe (just because I do) that this has been tried somewhere, sometime by someone.
  • Net-Cred! I emailed Elonka Dunin yesterday with a link to this mess. She replied: Very nice! Worth posting to the group. So, now I'll join the group, read all kinds of good ideas and lose many (more) hours in Kryptos :)

Marking out (clocked out) at approx 10.33 hours playing with this. PHP source code available if you want it - use my 'homepage' and decode my email ( Home Page )
May 3, 2011(MAYDAY/SOS-Local: RIH OBL - CIA solved one of their mysteries - GJ CIAUSNJSOC!) 'this page going live' (linked)

Of Final Note(s):
  • The keyword submission is 'cleaned' by this script. Any characters outside of A-Z (nonprintable or numbers) are removed. This self-protects against possible injection. Bot/Frequency attacks will get the host IP firewalled using 'standard' frequency/DOS watch-ware.
  • K4 may not solve with the same double-key-poly method as K1 and K2. This is just a 'high speed scrap paper' way to test out 'could be keywords.'
  • I tried a bunch of QU keywords because: Q is the misspelled letter in ILLUSION (K1); U is the misspelled letter in UNDERGRUUND; and E is the misspelling (?) in DIGETAL. Very elementary idea but it stuck out to me - and I still do not understand where on earth 'the breakers' figured out the other 2 keys.
  • I tried KZLDKRNSHGNFIVJY as a key because that's what is checkmarked on Sanborn's original docs.
  • Another 'thought' I had was that - if it is a one-time-key like some say - you (I probably won't) could write a fast script to take the encoded text or the tableau screen and 'run it' (at least 97 letters of it) thru as the key. It would not be that hard to 'rotate' the lettering thru this decoder-script in 97 character chunks. Instead of HUMINT - let the computer idenfify 'BERLIN' (just raw letters and/or actual placement).
  • De-puzzle-ing Kryptos might not be in my ability set - I'm still trying to figure out how PALIMPSEST and ABSCISSA were 'come up with.' I have read a few pages from breakers and I just cannot get the lightbulb to come one about that. So, a keyword toy.

Marking out (clocked out) at approx 10 hours playing with this. PHP source code available if you want it - use my 'homepage' and decode my email ( Home Page )
4.28.11 Epic Fail(yer) Sort OfWhile patting myself on the back for producing such a 'low profile' workable decodexor... a nag. This is why you don't run off publishing stuff on the internetswebs (I, myself, did not) until all of the 'nags' and 'doubts' are resolved. What happened? I just didn't like it - something. What it turned out to be: K1 and K2 are not solved against the 'screen' (the letter-layout on Kryptos itself). I had been so busy working against transcripts of Kryptos, I had totally missed the 'howto' docs that showed that K1/K2 used a different alphabet, and, K1/K2 do not have the keyword offset to the left. That offset is the 911 part of what I missed - the keyword gets vertically stacked under the K of 'KRYPTOSABC...' If the original text contains a K then the encoded text contains the KEY (unshifted, unsubstituted). Adding a K1/K2 style decode function, leaving the 'screen' because 'it is here, it works, it (K4) might still be encoded that way.' Fast fix, +45 mins (9:45 ish now).
4.28.11 Tech2 hours more on 27th and 1 today (9 total now) got everything working and re-layout more to my liking. About 1 hour spent pushing 'keywords' then hand-verication against the 'screen alphabets'. No brain cells really toasted yet trying to figure out a sensible keyword to try, just abusing things mainly for testing. CRACK produced the least Qs and Xs - but nothing else sensible.
4.27.11 Tech / Platform NotesTechno-babble - codexors do ignore.
The tables and 'layout' get an A in IE8(WinVista); the font, color and general rendering... well MS, sorry, D+. On the other hand, iOS (iPod) and Android (Verizon Droid Phone) get a font,color,renering grade of A and A+ respectively. The iOS and Android layout, A-, some of the 'nested' centering is not working there. The php has been stalled for two days - life in the way of nerdy-skills-hobby-honing. The 'syntax' (BERLIN) highlighter is functionalized and reusable, the mysql stuff is good and working, the 'decode' is still in draft, the 'layout' a transposed is still in-line (not a function). And, as of 12:21pm 4.27 - K4 remains on hold because 'hobby/goof' time is very limited. So far profiling is resultant in 0.0% system use, even fancy arrays and transposition (in an interpreted language no less) have no impact on a modern computing machine. (Time in marked at about 4 hours, including the large waste of ORDINATE time (by hand). An appointment got broken, 2 hours (including hand verify against the kryptos-screen) and more ORDINATE testing = fully functional and debugged 'decoder' function. Only decodes against the KRYPTOS (key 1) and whatever you use for (key 2) 'screen-alphabet' right now - no 'plain alphabet decoder.' Timestamp: 6 hours.
4.25.11 x-positTransposition Probs/Note: You cannot do (or undo) a 'prime' or 'odd' transposition without padding, and padding corrupts the coding/decoding. Because 'simple' (pre-planned) transposing/transpositioning always start and end with the same letters (try it out, put ABCDEFGHI (A...I) into a 9-cube grid, encode/transpose it ADGBEHCFI (A...I), then decode/re-pose it back to original (A...I) - OK, because of that I used a 8x12 grid (96) for the initial layout (no padding, no loss) and just 'carry' (or retain) the last letter. Without encoding/decoding K4 always starts with O and ends with R, so the '97th' letter is not used in the transposition charts - it is simply added to the end on its own line.
4.25.11 HUMINTWhy not just book/dictionary attack K4 with this script? Because, it is my (may be misguided) belief that if a transposition (like K3) is used as well as an encoding (K1/K2) that human intelligence (vision, pattern/word recognition) is necessary to solving this. It would take many thousands of lines of code to look for and flag just 'BERLIN' - this is for fun (and nerdiness) so I am not about to commit hundreds of hours and 1000s of lines of code to it.
4.24.11 Prime TimeOne of the first things I noticed during the 'early K4 document analysis': 97 is a prime number. Then there is the 'artistic stack' - 4 letters (5 if you count Q?) on the first row of K4, followed by 3 rows of 31 (prime number) letters. (Update: checked the originals - Sanborn's graph paper used 31 boxes/cubes. My early 1990s store-purchased [yep, still have it] has 33-1/2. Just a note.)
4.24-4.25.11 Anagram?By 'pulling out the letters' you can spell all kinds of things, I got as far as: BERLIN KRYPTOS KGB CIA QUAGMIRE ... well, it's written down at home right now, it still 'could be' just a very fancy anagram (+ Q?) but I am not putting much more effort in that - personally very limited vocabulary. (And, yes, I know 'Doctor Lecter' that a mixed-up-letter-puzzle is not really an anagram, solve K4 then complain about my eeng-uh-lush.)

Why I thought 'it could just be a mixed up letter game': KR, YP, TOS(in reverse) in 'plain sight':
(May 4, 11 - adding red on suggestion because IE8/9 just do not render it well.)
K4 - KRYP(TOS) in bold - TT(font) and right-aligned
OBKR
UOXOGHULBSOLIFBBWFLRVQQPRNGKSSO
TWTQSJQSSEKZZWATJKLUDIAWINFBNYP
VTTMZFPKWGDKZXTJCDIGKUHUAUEKCAR

Sanborn said 'clues' and 'directions' regarding the original encypher notes: in the K1 area there is a left and right arrow (in the left margin). I was trying to 'read into K4' top-row-right, 2nd-row-left..., very much 'making a theory work.'
4.24.11 FailedMy 'by hand' effort on Easter was to try to decode using the K1/K2 methods using ORDINATE as a keyword - ordinate is the 'y-axis' (or vertical direction) 2nd part of the 'coordinate set' that ABSCISSA is the first part of. Produced 99% uselessness.
Clues? or NotCIA Berlin Wall Segments The three sections of wall contain the following: (a) [Picture]Sun over mountains and water; (b) [graffiti - hard to read, maybe:] RALE or RALLE; (c) [picture] man-face with long nose, orange eyes and orange drool(?); (d) AND THE WIND CRIES; (e) MNP [N is backwards? byknife! zodiac? nevermind...]; (f) TEAR DOWN THE WALL; (g) Endlich Frei ["Free at last" or "Finally Free"]; (h) DEMOCRACY; (i) [heart picture +]FREEDOM.
Note (Clue?) Q?In K1 ILLUSION is mis-spelled IQLUSION. Actually it is the KEYWORD - PALIMPCEST [correct: palimpsest] that is mis-spelled - at the top of the 'original' encoding page, and only once in the actual encoding - ILLUSION is spelled correctly, palimpcest is not.
November 2010 CluesThe creator of Kryptos gave out some 'clues' on the 20th anniversary of Kryptos' unwrapping. NY Times Article. NOTE:The 'net-consensus' is that he said 'NYPVTT=BERLIN' (64-69) - that is what the reporter wrote (said), not Mister Sanborn. The clue does not necessarily mean that it is a 'straight decode' - it could still be an 'anagram' or 'transposition' along with a decode. Mr. Sanborn gave the NY Times the 'original' graph-paper he used to encode K1-K2, also. In a Wired Story (actually 2, see also: Wired Story 2) Mr. Sanborn said: '...it's a significant clue...' and (paraphrased) '...gave them [NYT] the original codes, more clues...' and he said '...directions. I use the word directions intentionally/purposefully.'
My EffortsShort Version: end of march looked at an FBI-decode thing, led to in april 'broke' (high confidence) Zodiac Z340 (no full decode yet), led to by way of netsearch for 'unsolved crypto' finding out about Kryptos. Easter (4.24.11) had some 'big ideas' (that only lead to obsession with decrypting stuff) which led to making this page because scrappy-paper decryption is just too old fashioned for me. In my 'one day' of playing I did not join 'the groups' and read all ten-buh-zillion webpages about Kryptos - I'm putting my ideas and efforts here and if with all other resources available on Earth someone uses this 'toy/tool' to finally break K4 - yay. Remember us little people in your memoir(s).
Kryptos? K4?Kryptos is an art/sculpture containing (at least) four 'hidden' messages installed at the CIA's HQ in Langley, VA in 1990. Three of the 'main four' have been decoded/decrypted, the final 97 letters (as of 4.25.11) remain unsolved. It's a crytpo-nerd thing to try to solve it (crack it, break it). See Elonka's Kryptos Page (Elonka Dunin) for all of the history, details, FAQ, yahoo! group, etc.

DLW / MLD Home Page