3054: Scream Cipher

Explain xkcd: It's 'cause you're dumb.
Revision as of 18:21, 21 February 2025 by Mwarren (talk | contribs) (→‎Explanation)
Jump to: navigation, search
Scream Cipher
AAAAAA A ÃA̧AȂA̦ ǍÅÂÃĀÁȂ AAAAAAA!
Title text: AAAAAA A ÃA̧AȂA̦ ǍÅÂÃĀÁȂ AAAAAAA!

    Explanation

    Ambox warning blue construction.svg This is one of 54 incomplete explanations:
    Created by a SCREAMING BOT - Please change this comment when editing this page. Do NOT delete this tag too soon. If you can fix this issue, edit the page!

    A substitution cipher is a simple form of encryption where each letter in the original text is replaced with a unique letter. In the comic's cipher, the replacement letters are all variants of the letter A with different diacritical marks.

    It's named "Scream Cipher" because the written form of a scream is a long string of As, possibly with some other characters at the end (and an excamation point for emphasis) e.g. "Aaaaaah!" or "Aaaaaaagh!". The name is not a reference to IBM's Scream cipher published in 2002.

    In the comic, Cueball texts the cipher for "HELLO", and Megan responds with the cipher for "HI".

    The title text deciphers to "AAAAAA A SCARY MONSTER AAAAAA!"

    In Unicode the cipher can be described as"

    Input Substitution
    U+0041 LATIN CAPITAL LETTER A U+0041 LATIN CAPITAL LETTER A
    U+0042 LATIN CAPITAL LETTER B U+0226 LATIN CAPITAL LETTER A WITH DOT ABOVE
    U+0043 LATIN CAPITAL LETTER C U+0041 LATIN CAPITAL LETTER A, U+0327 COMBINING CEDILLA
    U+0044 LATIN CAPITAL LETTER D U+0041 LATIN CAPITAL LETTER A, U+0332 COMBINING LOW LINE
    U+0045 LATIN CAPITAL LETTER E U+00C1 LATIN CAPITAL LETTER A WITH ACUTE
    U+0046 LATIN CAPITAL LETTER F U+0041 LATIN CAPITAL LETTER A, U+032E COMBINING BREVE BELOW
    U+0047 LATIN CAPITAL LETTER G U+0041 LATIN CAPITAL LETTER A, U+030B COMBINING DOUBLE ACUTE ACCENT
    U+0048 LATIN CAPITAL LETTER H U+0041 LATIN CAPITAL LETTER A, U+0330 COMBINING TILDE BELOW
    U+0049 LATIN CAPITAL LETTER I U+1EA2 LATIN CAPITAL LETTER A WITH HOOK ABOVE
    U+004A LATIN CAPITAL LETTER J U+0041 LATIN CAPITAL LETTER A, U+0313 COMBINING COMMA ABOVE
    U+004B LATIN CAPITAL LETTER K U+1EA0 LATIN CAPITAL LETTER A WITH DOT BELOW
    U+004C LATIN CAPITAL LETTER L U+0102 LATIN CAPITAL LETTER A WITH BREVE
    U+004D LATIN CAPITAL LETTER M U+01CD LATIN CAPITAL LETTER A WITH CARON
    U+004E LATIN CAPITAL LETTER N U+00C2 LATIN CAPITAL LETTER A WITH CIRCUMFLEX
    U+004F LATIN CAPITAL LETTER O U+00C5 LATIN CAPITAL LETTER A WITH RING ABOVE
    U+0050 LATIN CAPITAL LETTER P U+0041 LATIN CAPITAL LETTER A, U+032F COMBINING INVERTED BREVE BELOW
    U+0051 LATIN CAPITAL LETTER Q U+0041 LATIN CAPITAL LETTER A, U+0324 COMBINING DIAERESIS BELOW
    U+0052 LATIN CAPITAL LETTER R U+0202 LATIN CAPITAL LETTER A WITH INVERTED BREVE
    U+0053 LATIN CAPITAL LETTER S U+00C3 LATIN CAPITAL LETTER A WITH TILDE
    U+0054 LATIN CAPITAL LETTER T U+0100 LATIN CAPITAL LETTER A WITH MACRON
    U+0055 LATIN CAPITAL LETTER U U+00C4 LATIN CAPITAL LETTER A WITH DIAERESIS
    U+0056 LATIN CAPITAL LETTER V U+00C0 LATIN CAPITAL LETTER A WITH GRAVE
    U+0057 LATIN CAPITAL LETTER W U+0200 LATIN CAPITAL LETTER A WITH DOUBLE GRAVE
    U+0058 LATIN CAPITAL LETTER X U+0041 LATIN CAPITAL LETTER A, U+033D COMBINING X ABOVE
    U+0059 LATIN CAPITAL LETTER Y U+0041 LATIN CAPITAL LETTER A, U+0326 COMBINING COMMA BELOW
    U+005A LATIN CAPITAL LETTER Z U+023A LATIN CAPITAL LETTER A WITH STROKE

    Transcript

    Ambox warning green construction.svg This is one of 29 incomplete transcripts:
    Do NOT delete this tag too soon. If you can fix this issue, edit the page!
    [Top of the panel has 26 letters of the alphabet, each followed by a hyphen and the letter "A" with a unique diacritical mark for each:]
    A - A
    B - Ȧ G - A̋ L - Ă Q - A̤ V - À
    C - A̧ H - A̰ M - Ǎ R - Ȃ W - Ȁ
    D - A̲ I - Ả N - Â S - Ã X - A̽
    E - Á J - A̓ O - Å T - Ā Y - A̦
    F - A̮ K - Ạ P - A̯ U - Ä Z - Ⱥ
    [Cueball and Megan typing on their phones, Cueball with two hands and Megan with one hand.]
    Cueball's phone: A̰ÁĂĂÅ
    Megan's phone: A̰Ả
    [Caption below the panel:]
    In the Scream Cipher, messages consist of all As, with different letters distinguished using diacritics.

    comment.png  Add comment      new topic.png  Create topic (use sparingly)     refresh discuss.png  Refresh 

    Discussion

    Anyone know a good free all-language OCR tool to help with the transcript? 172.69.67.156 17:30, 21 February 2025 (UTC)

    Found one here: https://www.lexilogos.com/keyboard/diacritics.htm MeZimm (talk) 17:52, 21 February 2025 (UTC)

    The written cipher is very interesting, but where can I hear recordings of the spoken form? Rockymountain 17:31, 21 February 2025 (UTC)

    Here ya go. MeZimm (talk) 17:54, 21 February 2025 (UTC)

    Are Cueball and Megan millenials? Who else would text greetings when they're standing right next to each other? Barmar (talk) 17:38, 21 February 2025 (UTC)

    They could be texting other people. B for brain (talk) (youtube channel wobsite (supposed to be a blag)) 19:37, 21 February 2025 (UTC)
    Engineers and cyberfolk were text messaging their neighbors rather than talking long before it was cool encouraged for social distancing or quarantine! It's always helpful to get a reminder not to do this. 162.158.159.101 20:37, 21 February 2025 (UTC)
    They might not be diegetically in the same room. Comics can get weird with physical space. GreatWyrmGold (talk) 20:34, 21 February 2025 (UTC)

    Likely a pun on "stream cipher"

    Related reference: https://en.wikipedia.org/wiki/Scream_(cipher) 172.68.26.229 17:46, 21 February 2025 (UTC)

    A̦ÅÄ ẠÂÅȀ, A̓A̅ ȀÅÄĂA̱ ȦÁ ÂÁAĂĂA̦ A̮ÄÂÂA̦ A̓A̮ ȀÁ A̱A̓A̱ A ÀÁÂÃA̓ÅÂ ÅA̮ A̅A̰A̓Ã A̭AA̋Á A̓Â A̅A̰A̓Ã ÃA̅A̦ĂÁ! MeZimm (talk) 17:50, 21 February 2025 (UTC)

    The Wikifunction from Scream returns "YOU KNOW, JA̅ WOULD BE NEALLY FUNNY JF WE DJD A VENSJON OF A̅HJS A̭AGE JN A̅HJS SA̅YLE!". Hmmm... Mwarren (talk) 19:04, 21 February 2025 (UTC)
    The page you link to came into existence 7 minutes after I had posted this comment ;) I was doing it all manually, using this page and best-guess attempts to interpret what Randall's handwritten diacritics were meant to correspond to. MeZimm (talk) 19:55, 21 February 2025 (UTC)
    updated version: "A̦ÅÄ ẠÂÅȀ, ẢĀ ȀÅÄĂA̱ ȦÁ ȂÁAĂĂA̦ A̮ÄÂÂA̦ ẢA̮ ȀÁ A̱ẢA̱ A ÀÁȂÃẢÅÂ ÅA̮ ĀA̰ẢÃ A̯AA̋Á ẢÂ ĀA̰ẢÃ ÃĀA̦ĂÁ!" MeZimm (talk) 20:01, 21 February 2025 (UTC)

    On Wikifunctions, we implemented the two functions to Scream Cipher and from Scream Cipher --172.70.38.235 18:09, 21 February 2025 (UTC)

    It looks like the wikifunctions are using a different character for "D" than the github project linked in the explanation. Seems as though one's using U+0331 and the other's using 0332. Schiffy (Speak to me|What I've done) 20:32, 21 February 2025 (UTC)
    Could always be made fault-tolerant, accepting both characters. If one wants to be conservative with A-candidates to save them for further alphabetical expansion, well, we might have to ask Randall for proper specification.172.71.160.94 08:42, 3 March 2025 (UTC)

    Is there a logic behind the choices of the letter? I guess A̧ is for C because of the French ç and Å is pronounced like O in some Nordic languages. Also, is it A̱, A̲ or A̲ ? (or something else). 172.71.126.50 18:10, 21 February 2025 (UTC)

    Seems to be mostly visual similarity. Å has an actual O shape added to it. 172.70.110.171 20:19, 21 February 2025 (UTC)

    I give it a week for people to make a translator to and from this cipher. Caliban (talk) 18:20, 21 February 2025 (UTC)

    The Wikifunctions translations above were complete at least 11 minutes before your comment and well within the goal of one week :-) . Mwarren (talk) 19:04, 21 February 2025 (UTC)

    Numbers should be variations of h and/or g. Andyd273 (talk) 18:32, 21 February 2025 (UTC)#

    H > g SqueakSquawk4 (talk) 18:59, 21 February 2025 (UTC)

    Using sed you can encode with sed 's/C/A̧/g;s/D/A̱/g;s/F/A̮/g;s/G/A̋/g;s/H/A̰/g;s/J/A̓/g;s/P/A̯/g;s/Q/A̤/g;s/X/A̽/g;s/Y/A̦/g;y/BEIKLMNORSTUVWZ/ȦÁẢẠĂǍÂÅȂÃĀÄÀȀȺ/' and decode with sed 's/A̧/C/g;s/A̱/D/g;s/A̮/F/g;s/A̋/G/g;s/A̰/H/g;s/A̓/J/g;s/A̯/P/g;s/A̤/Q/g;s/A̽/X/g;s/A̦/Y/g;y/ȦÁẢẠĂǍÂÅȂÃĀÄÀȀȺ/BEIKLMNORSTUVWZ/'. 162.158.159.102 18:41, 21 February 2025 (UTC)

    This is really neat. I suppose `tr` might work too. 162.158.10.242 16:15, 22 February 2025 (UTC)
    I don't think `tr` can be used, at least not the most common (GNU) implementation; it doesn't do multibyte characters (and thus UTF-8 or other Unicode). chaw (talk) 20:32, 23 February 2025 (UTC)
    You'd be able to use Perl's inbuilt equivalent tr/// (or y///) fair enough. I'd have to test for edge cases, but might or might not have to fight with /u or /l post-params (if otherwise desired), in concatonation with /i for case-insensitivity (useful for upconverting "plainText" to "CIPHERTEXT", unless you're also wanting to cater for "á" as well as "Á" in code, so that it can be restored upon decoding). There's plenty of scope for all of this, of course, well beyond the brief spec as given in the comic. 162.158.74.108 20:47, 23 February 2025 (UTC)

    It would be really funny if someone added an image of Bill Cipher screaming, with the tag "A screaming cipher". It wouldn't reall fit but it'd be funny SqueakSquawk4 (talk) 18:59, 21 February 2025 (UTC)

    Screamingcipher.png I did. B for brain (talk) (youtube channel wobsite (supposed to be a blag)) 19:57, 21 February 2025 (UTC) (EDIT: WOW, that thing is MASSIVE! Can someone please downscale it because I have no idea how. You have permission to edit my comment only for that.) (DOUBLE EDIT: Nevermind, I did it.)
    ThAT's GoLD ⯅A dream demon⯅ (talk) 14:15, 19 March 2025 (UTC)

    Someone started a GitHub repo with a web-based encoded/decoder already: https://github.com/Reginald-Gillespie/StreamCipher Dlech (talk) 19:35, 21 February 2025 (UTC)

    I'd almost want to edit in my repo instead of the current one because mine is objectively better, but I'm new to this and not sure if that's appropriate or not =P (I don't even know if I am commenting correctly) WKoA (talk) 00:17, 22 February 2025 (UTC)

    Look to the likes of the 1190: Time comic, where several different fanbase compilations may have been given. I would anticipate that you could do a decent job of editing "This thing here does..." into "Ways of experiencing it include this [existing one] and that [yours] [with room to add more, if they add up]. Or just mention your link here, let others decide if your claims of (better?) functionality stand up enough to prompt it to be put up alongside/ahead/instead of the other. I am at least intrigued as to how you did it differently.
    And you certainly had a bit of trouble with the signing. Just add ~~~~ to the end of Talk comments and it autoreplaces. No need to go back in and edit (I added the original timestamp back in, for you, just for future reference). Unless of course you forgot to do it the first time... ;) 172.71.178.161 02:41, 22 February 2025 (UTC)

    I know you can decode a substitution cipher by counting letters and replacing common ones like 'E' and then filling in the rest by inspection, but what kinds of automated approaches are there? 162.158.159.105 20:14, 21 February 2025 (UTC)

    HEADS UP: I just changed A̲ (0332 COMBINING LOW LINE) to A̱ (0331 COMBINING MACRON BELOW) as encoding for "D" in the table and the transcript. Rationale: "T" is written with macron, so it's only logical to encode "D" likewise. A "low line" is longer than a macron, and looking at Randall's comic, the line below the "D" is definitely not longer than the one above "T". It would also make no sense to encode "T" with a "combining low line" as well when a single, uncombined character exists. 172.70.114.123 20:29, 21 February 2025 (UTC)

    These options should all be including as different writing styles! I think the longer line makes it clearer that A̲ represents D because the ink comes nearer to forming into a closed curve. 172.68.54.68 15:55, 22 February 2025 (UTC)

    What's the American alphabet? AnAussie 172.68.64.213 01:16, 22 February 2025 (UTC)

    I coined this term. What I meant by it is the alphabet used by americans from the point of view of an american cultural experience (mine), where for example a decorated word like façade is seen so infrequently that the word cedilla may be equated with the letter C. 162.158.62.105 20:24, 24 February 2025 (UTC)

    I expect to win a Turing Award for my proof this cypher is computationally equivalent to https://en.wikipedia.org/wiki/Whitespace_(programming_language) 172.71.158.19 02:57, 22 February 2025 (UTC)

    The page technically uses the incorrect characters for (at least) E, M, N, O, R, S, and T based on the title text shown on xkcd.com. The original title text uses two separate characters (ex. A + 0301 COMBINING ACUTE ACCENT for E), whereas the table uses the combined character (ex. 00C1 LATIN CAPITAL LETTER A WITH ACUTE for E). Alternatively, my browser is just doing something weird. Not necessarily worth updating, but something I noticed when implementing the cipher. Abus (talk) 06:22, 22 February 2025 (UTC)

    I think it's the browser. Firefox->Page Source gives me A+COMBINING WHATEVER, but w3m -dump_source https://xkcd.com/3054/ | zcat | grep 'img.*title=' returns single characters. The title text here on explainxkcd was copied by TheusafBOT. I trust TheusafBOT and w3m to be so basic and simple that they wouldn't try to do something 'clever' with the characters, whereas I tend to suspect the multi-MB-monster Firefox messed things up. I'm just guessing, though... could maybe someone test with yet another browser - e.g. Edge or Opera? 162.158.159.107 11:52, 22 February 2025 (UTC)
    Argh, no, turns out w3m, and also wget all return A+COMBINING THINGY, but they got merged into the single chars by my xterm when I copied them to some 'identify unicode' web page. Looking at the raw file dumped by wget I see A+COMBINING XX - I think... So I think you are right with your observation! Randall uses A+COMBINING XX for the title text on XKCD (though I really doubt that was intentional), then TheusafBOT merged the characters when it copied the text to create this page. That said, I still think using the merged chars is cleaner. 162.158.159.107 12:20, 22 February 2025 (UTC)
    It may be worth noting the difference somewhere, even though the merged chars are cleaner. Someone copying the characters from this page to encipher text would technically be in violation of the "spec" since Randall used the two character version in the title text. If someone wished to decipher the title text from the original XKCD, this characters on this page would fail. Abus (talk) 18:56, 22 February 2025 (UTC)
    unsure how to indent a sourceblock, see below 172.68.54.167 16:13, 22 February 2025 (UTC)
    $ curl -s https://m.xkcd.com | sed -ne 's/.*id="altText">\([^<]*\)<.*/\1/p' | iconv -f utf8 -t wchar_t | hexdump -C
00000000  41 00 00 00 41 00 00 00  41 00 00 00 41 00 00 00  |A...A...A...A...|
00000010  41 00 00 00 41 00 00 00  20 00 00 00 41 00 00 00  |A...A... ...A...|
00000020  20 00 00 00 41 00 00 00  03 03 00 00 41 00 00 00  | ...A.......A...|
00000030  27 03 00 00 41 00 00 00  41 00 00 00 11 03 00 00  |'...A...A.......|
00000040  41 00 00 00 26 03 00 00  20 00 00 00 41 00 00 00  |A...&... ...A...|
00000050  0c 03 00 00 41 00 00 00  0a 03 00 00 41 00 00 00  |....A.......A...|
00000060  02 03 00 00 41 00 00 00  03 03 00 00 41 00 00 00  |....A.......A...|
00000070  04 03 00 00 41 00 00 00  01 03 00 00 41 00 00 00  |....A.......A...|
00000080  11 03 00 00 20 00 00 00  41 00 00 00 41 00 00 00  |.... ...A...A...|
00000090  41 00 00 00 41 00 00 00  41 00 00 00 41 00 00 00  |A...A...A...A...|
000000a0  41 00 00 00 21 00 00 00  0a 00 00 00              |A...!.......|

    TRIARESIS Does anyone else view the triaresis as a missed opportunity? I'm thinking of Die Aerzte". [1]. Can someone insert the image of the band's logo? 172.71.102.222 17:21, 22 February 2025 (UTC)

    I saw that, when following up on the actual A-diaresisand I quite like the idea that they heavy metal umlauted an actual existing umlaut/diaeresis... If it weren't irrelevent to the comic (and skipped the bit at the top that actually translates the name), I might have relinked to that anchor point. But happily boosting the visibility of it here with a small reply⋯ 172.68.205.122 21:31, 22 February 2025 (UTC)

    Yes, should encode all numbers, in binary, using 'g' and 'h' for 0 and 1.172.70.162.195 17:56, 22 February 2025 (UTC)

    Why g and h? I saw an early comment saying the same. Why those when letters are A? --Kynde (talk) 18:14, 22 February 2025 (UTC)
    I presume for the "AAAAAAAAAGH!" type thing... (Although, given that a number would then be something like GHHHGGH, as a separate word, I'm not sure it'll look "AAAAAAAGH"ish.
    And is it coded as MSF (42=101010), N-bit (e.g. =00101010), BCD per digit (4=0100 2=0010 =01000010) or some other form? Plenty of scope for interpretation.
    Also, might I suggest E and I (or I and E) for it, instead..? IEIEIE EEIEIEIE EIEEEEIE! 172.68.205.178 19:10, 22 February 2025 (UTC)

    Should "Augh" be added to the versions of screaming? Randall uses it alot, e.g. https://xkcd.com/493/, https://xkcd.com/1401/, https://xkcd.com/1388/, https://xkcd.com/1207/, https://xkcd.com/1226/, https://xkcd.com/780/, https://xkcd.com/990/ -- Drkaii (talk) 18:06, 22 February 2025 (please sign your comments with ~~~~)

    ÅĂÃÃÀ ÂȦA̮Ⱥ! A̯ ȀA̦ÀȀÀȺĂ A̰ ÂÀÀẠ A̤ÀẢ AA̦ÀȀÅA̮ ǍA̦ÀĀ 2916 AÀ A̰ÄA̮ÀÄĂ A̱ÅÀ A̓A̰Ä ǍA̯ÄẠ A̓ÅA̰A̦A̰A̓AĂA̦Ⱥ ǍÀA̦ ȀȦÄA̓AȦA̰AA̯ÀÄ A̰ÄẠ ÄȦĀẢĂA̦Ⱥ AÀ A̰ẠẠ A̯ÄAÀ AÅĂ A̓A̯ȀÅĂA̦, A̰ÄẠ A̱ÅA̯AĂ ÅA̰A'Ⱥ ÅA̰A AÀ A̰ÄA̮ÀÄĂ A̱ÅÀ A̓A̰Ä ẠĂA̓A̯ȀÅĂA̦ AÅA̯Ⱥ A̯Ä 1 ĀA̯ÄȦAĂ ẢA̮ ÅA̰ÄẠ. (A̰ẢÂĀ: ĀȂA̦ A A̧ẢA̯A̰ÁȂ ĀA̰AĀ ȦÁA̋ẢÂÃ ȀẢĀA̰ ĀA̰Á ȦĂÄÁ, AÂA̱ A̱ÅÁÃ ẢĀÃ ȀÅȂẠ ȦA̦ A̱ÅẢÂA̋ A ĀȂAÂÃĂAĀẢÅÂ.) (A̰ẢÂĀ ĀÅ ĀA̰Á A̰ẢÂĀ: ĀA̰Á ȦĂÄÁ A̧ÅÀÁȂÃ 70% ÅA̮ ĀA̰Á ȀÅȂĂA̱'Ã ÃÄȂA̮AA̧Á.) 162.158.90.46 03:57, 23 February 2025 (UTC)

    Can anybody provide pronunciation notes or recordings for those 'A's which are actually used in human languages? Can we transcribe the title text and the "Hello", "Hi" from the comic in IPA or something? 141.101.99.88 10:33, 23 February 2025 (UTC)

    Choose the base letters arbitrarily other than ‘A’ and we get furigana. 物灵 (talk) 05:38, 24 February 2025 (UTC)

    If you accept the definition that a 'word' is a bunch of letters, surrounded by a gap, then 'xnopyt', AAAAAAJJJJJJJ 172.68.71.111 15:40, 24 February 2025 (UTC)

    A̰ 172.69.71.144 16:48, 24 February 2025 (UTC)

    The Trivia "Notes" section reads like pareidolia. -- Rei (talk) 13:16, 24 February 2025 (UTC)
          comment.png  Add comment
    Retrieved from "https://www.explainxkcd.com/wiki/index.php?title=3054:_Scream_Cipher&oldid=366224"