Comments on: Puzzle master wanted!

Question: Puzzle master wanted!

neurodoc — Sat, 16 Sep 2006 16:20:10 -0800

Can someone direct me on the right path to break this code?

As a fun challange, I've been trying to decode a message that someone gave me. The message was created with some sort of dingbat font, and there are no word breaks. There are a total of 198 characters, made up of 38 unique symbols.

I've converted the symbols into A-Z plus 0-9 plus @ and $, and have looked at the frequency of each character. However, nothing really jumps out (1 character appears 14 times, 2 appear 13 times, 1 at 10 times, 3 at 9, and so forth on down).

Any advice on what to do next to crack this thing?

By: effugas

effugas — Sat, 16 Sep 2006 16:22:23 -0800

Take a look at what's neighboring what. Bigram and trigram distributions can be mighty helpful.

Look for THE.

By: LeisureGuy

LeisureGuy — Sat, 16 Sep 2006 16:42:00 -0800

Read this very interesting novel: PopCo, by Scarlett Thomas. Enjoyable, and she talks a lot about cryptanalysis.

By: Leon

Leon — Sat, 16 Sep 2006 16:45:47 -0800

Possibly stupid thought, but if it's in a dingbat font, then maybe it's just plaintext. Find the original font.

By: scottreynen

scottreynen — Sat, 16 Sep 2006 17:12:04 -0800

How about posting the whole thing here?

By: kindall

kindall — Sat, 16 Sep 2006 17:13:36 -0800

Assuming it's a simple substitution cipher, where one symbol equals the same letter throughout the message, breaking the code is pretty straightforward. 200 characters should be plenty. What you need to know is ETONAIRSH, which is the order of the first 9 most frequent letters used in English, in order of frequency. (Some sources have this ETONAIRSH because N and A are pretty close.)

Letter frequency means that the character that appears 14 times is probably E and the ones that appear 13 times is probably T and O (though the frequencies are close enough that any of them could be E, T, or O). The one that appears ten times is probably N. Look for two-character sequences that contain two of these three symbols; these are probably the word "TO." Three-character sequences that contain two are probably the word "THE," and knowing this gets you the character for H as a bonus. Once you know what your T and O characters are, you can look for two-letter words beginning with O (which are probably "OF" or "ON" and if they don't end in the character that appears 10 times, they are probably "OF") and figure out the F. Single character words are either A or I. TH_T is THAT, etc. A longish word that ends in "ON" can usually ends in "ION" and can thus be used to confirm "I" (it will likely be one of the symbols that appears 9 times). Eventually you will start to be able to guess at words, which will let you fill in more of the letters, and at a certain point it will just all fall into place.

This is pretty basic cryptanalysis and you can probably Google up some good tutorials using that word. Or check the young adult section of your library.

By: AndrewStephens

AndrewStephens — Sat, 16 Sep 2006 17:32:26 -0800

Are you sure its a substitution cipher? Your friend may have used the font as some sort of pictograph based code.

By: neurodoc

neurodoc — Sat, 16 Sep 2006 17:48:14 -0800

Thanks for the help on this, but I think I wasn't quite clear enough in my original message. Some of the challanges I see to the above suggestions:

1) There are 38 unique characters used in the message. I think this means that its something more than a simple substitution cipher (which would, of course, have only 26 characters in it).

2) There are no spaces in the message. All characters run together, so it's impossible to tell a one-letter from a two-letter from a twelve-letter word.

I have managed to figure out the font that was used. It's called "Type Embellishments Pi One", and I found a site that lets me type samples of the font. I'm hoping that at least the symbols that correspond to the keyboard letters A-Z might actually be a substitution for A-Z, but of coure I really have no idea.

I'd rather not post the encoded message, as I have no idea what the decoded version might say!

Keep the help coming!!!

By: Mitheral

Mitheral — Sat, 16 Sep 2006 17:52:14 -0800

38 unique symbols in 198 characters? You've either got a specially contructed sentance (IE: The quick brown fox jumped over the lazy dog! 1234567890.) or something besides a simple substitution cypher. Possibly they are using a series of symbols for space or multiple symbols for some characters or symbols for character pairs (eg: # = ch, * = ee).

It's quite possible if you have no idea of the plaintext or subject that this string isn't decypherable. Something like ENIGMA would be impossible to finesse with such a short cypher text, a brute force approach would be needed.

By: monju_bosatsu

monju_bosatsu — Sat, 16 Sep 2006 18:32:24 -0800

Most dingbat fonts have different symbols for upper case and lower case characters, as well as, potentially, symbols for typographic marks including periods and spaces. so 38 unique characters doesn't necessarily mean that it's not a simple substitution cypher. You just need to take account of the possibility that there may be (1) multiple symbols for the same character depending on case, and (2) symbols for typographic marks.

By: AmbroseChapel

AmbroseChapel — Sat, 16 Sep 2006 19:48:13 -0800

>I have managed to figure out the font that was used. It's called "Type Embellishments Pi One", and I found a site that lets me type samples of the font. I'm hoping that at least the symbols that correspond to the keyboard letters A-Z might actually be a substitution for A-Z, but of coure I really have no idea.

If you've got that far, you can translate the dingbat font into normal characters. But, you haven't done that? I'm confused.

Doing that would resolve the "38 characters" problem right off, by determining if Dingbat X is just uppercase Dingbat Y.

By: neurodoc

neurodoc — Sat, 16 Sep 2006 21:04:53 -0800

Sorry for the confusion...I only figured out the font today, after posting my original question. I'm now trying to solve it under the assumption that upper and lower case letters were used.

Hopefully, I'll have this figured out by Monday! Thanks again for all the help.

By: Zed_Lopez

Zed_Lopez — Mon, 29 Jan 2007 17:21:38 -0800

Goodness, don't keep us in suspense. Did you figure it out?