Bay 12 Games Forum

Please login or register.

Login with username, password and session length
Advanced search  
Pages: 1 ... 9 10 [11] 12 13 ... 20

Author Topic: The Doctor's Cipher.  (Read 37061 times)

NJW2000

  • Bay Watcher
  • You know me. What do I know?
    • View Profile
Re: The Doctor's Cipher.
« Reply #150 on: August 30, 2015, 03:52:55 am »

Wait, someone might actually solve it within a week with a hint, and then I won't have time to give it to the cipher-genius I know and take the credit :(
Logged
One wheel short of a wagon

Hapah

  • Bay Watcher
  • The nice guy.
    • View Profile
Re: The Doctor's Cipher.
« Reply #151 on: September 30, 2015, 02:49:32 pm »

So! It's been a while. Anyone still working on this?
Logged
I can't be expected to remember the names of everyone I've tried to stab.

Bored? Go read the EVE Chronicles.

NJW2000

  • Bay Watcher
  • You know me. What do I know?
    • View Profile
Re: The Doctor's Cipher.
« Reply #152 on: September 30, 2015, 02:50:54 pm »

Thanks, I'll remind the clever person I asked.
Logged
One wheel short of a wagon

Hapah

  • Bay Watcher
  • The nice guy.
    • View Profile
Re: The Doctor's Cipher.
« Reply #153 on: September 30, 2015, 03:20:18 pm »

Thanks, I'll remind the clever person I asked.
Cool. Anyone else? Thinking of showing my hand and letting you lot weigh in, maybe we can get somewhere if we're all stuck.
Logged
I can't be expected to remember the names of everyone I've tried to stab.

Bored? Go read the EVE Chronicles.

Ozarck

  • Bay Watcher
  • DiceBane
    • View Profile
Re: The Doctor's Cipher.
« Reply #154 on: September 30, 2015, 03:40:13 pm »

I haven't been working on it in a while. I did invite a couple people to work as partners with me, but only one person seemed mildly interested, so I dropped it. Had any one person agreed, I would have let the others know I had gotten a partner, for fairness.

some obvious conclusions I drew a while back:
It's not a substitution cipher

There are at least six strings of symbols of at least eight symbols in length (I think I found eight or ten so far) that repeat. They do not repeat in a noticeable pattern (i.e. in periods of 36 or 40 symbols).

If it's a trigram cipher, there are 74 combinations in use, which means it's not a simple trigram to English alphabet thing. If it's a Baconian cipher then it is still not a direct code to alphabet cipher - there are more combinations than 26 in use. Also, someone above mentioned that certain letters were upper and certain ones were lower case. I believe they made a mistake, and the phi is lowercase.

Skip sequencing has yet to reveal a skip length that yields better results than the no-skip text.

The adfgx cipher is a good possibility, what with the way it scrambles text after translating it to five symbols. the main issue is the number of symbols - there should be an even number, there are an odd number.

My current suspicion is that it is either a scrambled trigram cipher, or that it is a trigram or baconian cipher that uses more symbols than the English alphabet.

It's been a while, so that's all I remember off the top of my head, except for the patterns I found, but I am not posting them yet, not without some added input from someone else. not that they are hard to find anyway, all that takes is patience and word-search talent. Or a simple program that compares substrings.

Hapah

  • Bay Watcher
  • The nice guy.
    • View Profile
Re: The Doctor's Cipher.
« Reply #155 on: September 30, 2015, 06:29:30 pm »

Alright, guess I'll give my thoughts!

I don't know how you guys went at it, but I'll let you know how I started. The first thing I did, after copying the text down and breaking it up into bigrams and trigrams, was to spend a little bit thinking about how Piecewise would design a cipher. I know he's not a fool, and I know he isn't gonna do anything quick and easy like a direct substitution cipher. However, I don't think he's a crypto expert or anything, so I imagine one of his first actions was to look up common ciphers, and more importantly, common methods for breaking ciphers. If he's cooking up his own cipher from scratch, he will include in the encryption process some way to counter these common methods. A little Google searching led me to believe that the most common ways to break codes were frequency analysis (i.e. the letter that appears the most is likely "E", the next most common is "T" and so on) and doubles (like the "O" in Doomarms), a letter that is repeated back-to-back.

I also know that Piecewise is clever and values cleverness, and that the cipher is likely not complex simply for the sake of complexity: he wouldn't encrypt the thing in such a way that the only way to crack it would be to brute-force it. I imagine that the cipher is likely a short set of rules (or perhaps even a single rule) that you follow when encrypting the text, and I bet the key to the cipher fits easily enough on a single sheet of paper. I don't think it'd be the kind of cipher where you have to know or guess a particular key word or anything like that, you just have to figure out the rules, and the cipher will likely be easy to use when encrypting or decrypting a message.

So! After doing this, I sat down and actually looked at the text with the following thoughts in mind:

  • The cipher will likely not be mechanically complex.
  • The cipher will likely contain some measure to foil frequency analysis
  • The cipher will likely contain some measure to foil "doubles"

I did a very quick search, and it appears I might be right on bullet 3 if the list is in trigrams: no 3-symbol string is ever repeated immediately after itself. Frequency analysis definitely yielded some outliers (λФΔ appears 26 times, θФΔ 18, ΔλФ 12, and ФФΩ 14 times, with the other 70 combinations appearing a single-digit number of times).

My first theory was that the text is broken up into trigrams, and that it was composed of multiple (4) different substitution ciphers. In this scenario, the cipher being used would be changed when a double was encountered, and the triple symbols (θθθ, ΔΔΔ, etc) were not text in and of themselves, but rather the signal to move to a particular cipher when decrypting the text. This had several things going for it, in my opinion:

  • It was a simple mechanism to keep letters from repeating in the text, one that I could certainly see Piecewise inventing
  • It also breaks frequency analysis, as your "E" can now appear as four different combinations of symbols
  • Encrypting the text by hand would be as simple as having a column for each of the ciphers, and moving to the relevant column when the signal comes. It is very easy to use.

As an added bonus, if the message is in fact signed "VonNost", this fit cleanly: the last 18 characters were ΩФΩθθθθλΔΔФФθθΩФФΩ, with the bold added just to make the 3-groups more visible. ΩФΩ would be "N" in a cipher, θθθ tells you to move to a new cipher, and the next 12 characters correspond to 4 letters (the "Nost" in a different cipher). It was also helped by the fact that some strings repeated several times (as Ozarck pointed out), and these strings seem to behave in consistent, predictable ways. For example, the string "ΔλФΔΔΔλФΔ" is repeated multiple times, and "ΔλФΔΔΔ" was ALWAYS followed by "λФΔ" every time it appears in the text.

Alas, it is not that simple: the same 3-symbol input gives a different 3-symbol output with the same cipher-change signal (ΩΩΩ). Also, there were two instances where a cipher change was called to the same cipher (i.e. you're on the ΔΔΔ cipher and the symbol comes to go to the ΔΔΔ cipher). I suppose there could be a special rule to just move a column over if this occurs, but it still doesn't fix the first issue. Could be that the ΩΩΩ and possibly the λλλ are special cases and behave differently (θθθ and ΔΔΔ appear 8 and 7 times respectively, ΩΩΩ 4 times and λλλ only once), but I haven't tried to examine what those special cases could be yet. Could also be that I'm just missing something obvious, give me your thoughts!

I feel like I'm onto something, but I'm just missing a lever or trigger or something like that (though everyone thinks they're onto something until conclusively proven wrong, myself included, lol). My current thinking is that the triple signals (θθθ, ΔΔΔ, etc) are not signals to move to a particular column or cipher, but could instead be a signal to move over a particular number of columns in Piecewise's cipher-sheet (i.e. "go right one column", "go left two columns", etc.) Haven't been able to piece together what values for the triples make things sync up (and it's entirely possible that I'm barking up the wrong tree anyway), but there you are!

So! What do you lot have? Any thoughts on this, or your own theories?
Logged
I can't be expected to remember the names of everyone I've tried to stab.

Bored? Go read the EVE Chronicles.

Urist McCoder

  • Bay Watcher
    • View Profile
Re: The Doctor's Cipher.
« Reply #156 on: September 30, 2015, 07:34:13 pm »

This makes more sense then anything I came up with, but what represents spaces?
Logged

Gentlefish

  • Bay Watcher
  • [PREFSTRING: balloon-like qualities]
    • View Profile
Re: The Doctor's Cipher.
« Reply #157 on: September 30, 2015, 07:41:56 pm »

don't have to represent spaces.

Hapah

  • Bay Watcher
  • The nice guy.
    • View Profile
Re: The Doctor's Cipher.
« Reply #158 on: September 30, 2015, 07:43:51 pm »

don't have to represent spaces.
That's my thinking. I suppose it's possible there is a symbol for spaces, but it's not required.
Logged
I can't be expected to remember the names of everyone I've tried to stab.

Bored? Go read the EVE Chronicles.

Urist McCoder

  • Bay Watcher
    • View Profile
Re: The Doctor's Cipher.
« Reply #159 on: September 30, 2015, 07:44:50 pm »

Makes sense, so is the current consensus that it is signed love doctor vonnost
Logged

Hapah

  • Bay Watcher
  • The nice guy.
    • View Profile
Re: The Doctor's Cipher.
« Reply #160 on: September 30, 2015, 07:47:11 pm »

Makes sense, so is the current consensus that it is signed love doctor vonnost
I'm only banking on "VonNost", personally. And even that isn't guaranteed: I just took it as a hopeful sign for my little theories that the crowd-favorite signature fits fine.
Logged
I can't be expected to remember the names of everyone I've tried to stab.

Bored? Go read the EVE Chronicles.

Ozarck

  • Bay Watcher
  • DiceBane
    • View Profile
Re: The Doctor's Cipher.
« Reply #161 on: September 30, 2015, 07:48:51 pm »

With 74 unique trigrams, it's possible to have three separate trigrams represent each letter.

Makes sense, so is the current consensus that it is signed love doctor vonnost
No. I see no real evidence of this. if yo ucan parse out the rest of the text based on this, do so, but I doubt it.

Besides, who was he writing to? probably himself. I doubt this code was meant for any other eyes. You can ask PW, but I doubt you'll get a useful answer.

Hapah

  • Bay Watcher
  • The nice guy.
    • View Profile
Re: The Doctor's Cipher.
« Reply #162 on: September 30, 2015, 07:54:13 pm »

With 74 unique trigrams, it's possible to have three separate trigrams represent each letter.
Wouldn't the ceiling for 3 seperate trigrams be 26*3 = 78? Though not every one will show up 3 times: I doubt we caught a "Z" and "X" in each cipher, for example.

I suspect that some trigrams might be in multiple ciphers as well, i.e. ФФΩ is "B" in cipher one and "X" in cipher two. There's all sorts of wrinkles.
Logged
I can't be expected to remember the names of everyone I've tried to stab.

Bored? Go read the EVE Chronicles.

piecewise

  • Bay Watcher
  • [TORTURE_FOR_FUN]
    • View Profile
    • Stuff
Re: The Doctor's Cipher.
« Reply #163 on: September 30, 2015, 08:26:24 pm »

I'll give you this one for free,
Because I'm a sucker for flattery.
You're right about the key,
and its relative brevity.

Hapah

  • Bay Watcher
  • The nice guy.
    • View Profile
Re: The Doctor's Cipher.
« Reply #164 on: September 30, 2015, 08:29:43 pm »

Hmm...
Logged
I can't be expected to remember the names of everyone I've tried to stab.

Bored? Go read the EVE Chronicles.
Pages: 1 ... 9 10 [11] 12 13 ... 20