MY 340
Moderators: Bullitt, Aud8us, Michael Butterfield, johnny5
Re: MY 340
by doranchak » Tue Aug 30, 2011 10:46 am
No way, it's TED!
BK MV BT MA BQ LQ QC GM HP FI RO LA FI HL AL DB YP LM ZD RY TK EN RY TV IL ZT OD OE AB AN RO IQ TL QA PL VA RX AW ZM QL PV YN GX HT OS
BK MV BT MA BQ LQ QC GM HP FI RO LA FI HL AL DB YP LM ZD RY TK EN RY TV IL ZT OD OE AB AN RO IQ TL QA PL VA RX AW ZM QL PV YN GX HT OS
Re: MY 340
by Zsearcher » Tue Aug 30, 2011 10:56 am
That TODOE requires "very mild anagramming" too, as AK likes to call it 
Re: MY 340
by sincos » Tue Aug 30, 2011 2:11 pm
doranchak wrote:No way, it's TED!
BK MV BT MA BQ LQ QC GM HP FI RO LA FI HL AL DB YP LM ZD RY TK EN RY TV IL ZT OD OE AB AN RO IQ TL QA PL VA RX AW ZM QL PV YN GX HT OS
You could be right. I found this:
'HIQDZBNMZTPIAMZTPT' [18]
Y Z H U Q S E D Q K G Z R D Q K G K -9
B C K X T V H G T N J C U G T N J N -6
E F N A W Y K J W Q M F X J W Q M Q -3
H I Q D Z B N M Z T P I A M Z T P T 0
K L T G C E Q P C W S L D P C W S W 3
N O W J F H T S F Z V O G S F Z V Z 6
Q R Z M I K W V I C Y R J V I C Y C 9
H O T A C E E D Z K J O R S I N Y K
Among the first 150 18-letter sequence from the cipher I found 68 where I could find Theodore J Kaczinsky...
/sincos
Ps I don't think glurk is neither Ted or Z.
Re: MY 340
by ALisowsky » Tue Aug 30, 2011 2:26 pm
Since work has been slow... I wanted to elaborate on the frequency distribution a bit:
These are the letter frequencies for glurk's playfair and my340 playfair. For the top charts, To calculate average position, the most frequently occuring letter receives a 1, the least frequent receives a 25. For the bottom two charts, To calculate average frequency, the total number of occurences for that letter are entered.
These numbers are added into the corresponding matrix in the position in which they occur in the solved playfair square. To the right of the values, is an average value for that row. Below the values, is an average value for this row. It was my assumption (incorrect
) that the letters which form the border of the playfair square (row 1, row 5, column 1, column 5) would be the highest frequency, however that was very wrong. The only correlation I did find, is that on both row 5 was the least frequent, and row 1 seemed to be an increased frequency of use.
What does this mean: Nothing. It hides frequency distribution pretty accurately (no big peaks/valleys). I don't think frequency distribution tells us anything about whether a playfair was used. :(
These are the letter frequencies for glurk's playfair and my340 playfair. For the top charts, To calculate average position, the most frequently occuring letter receives a 1, the least frequent receives a 25. For the bottom two charts, To calculate average frequency, the total number of occurences for that letter are entered.
These numbers are added into the corresponding matrix in the position in which they occur in the solved playfair square. To the right of the values, is an average value for that row. Below the values, is an average value for this row. It was my assumption (incorrect
) that the letters which form the border of the playfair square (row 1, row 5, column 1, column 5) would be the highest frequency, however that was very wrong. The only correlation I did find, is that on both row 5 was the least frequent, and row 1 seemed to be an increased frequency of use. What does this mean: Nothing. It hides frequency distribution pretty accurately (no big peaks/valleys). I don't think frequency distribution tells us anything about whether a playfair was used. :(
Re: MY 340
by ALisowsky » Wed Aug 31, 2011 8:08 am
I need some input. Because of the way a playfair encodes, we NEVER output ciphertext in a double letter, unless the following occurs:
Case A: We can have: AB CD DE FG, which when spaces are eliminated yields: ABCDDEFG (ie the appearance of a double letter).
Case B: However we would never have: AB CD EE FG, which when spaces are eliminated also yields a double letter.
If we then apply a homophonic sub, the double letter in case A could become the same symbol, meaning we would have a double letter show up after the substitution. In this case, it means we would have a double letter after substitution that does not rule out playfair.
Is the above assumption true?
If so, I created the following to explain why a playfair is impossible (if we assume right to left, top to bottom message)
The things to notice is if we split one of the double letters and assume they are more like the example in Case A, then continue to pair off the characters from that point, eventually we reach a double letter that is paired together, thereby eliminating the possibililty of a playfair preceding a homophonic sub.
Thoughts?
Adam
Case A: We can have: AB CD DE FG, which when spaces are eliminated yields: ABCDDEFG (ie the appearance of a double letter).
Case B: However we would never have: AB CD EE FG, which when spaces are eliminated also yields a double letter.
If we then apply a homophonic sub, the double letter in case A could become the same symbol, meaning we would have a double letter show up after the substitution. In this case, it means we would have a double letter after substitution that does not rule out playfair.
Is the above assumption true?
If so, I created the following to explain why a playfair is impossible (if we assume right to left, top to bottom message)
The things to notice is if we split one of the double letters and assume they are more like the example in Case A, then continue to pair off the characters from that point, eventually we reach a double letter that is paired together, thereby eliminating the possibililty of a playfair preceding a homophonic sub.
Thoughts?
Adam
Re: MY 340
by sincos » Wed Aug 31, 2011 8:51 am
Your conclusions that it is not a playfair seem valid to me.
Re: MY 340
by doranchak » Wed Aug 31, 2011 9:37 am
I agree with sincos. The only thing I can think of that would strengthen the argument FOR playfair would be if some cipher symbols are truly intended to represent more than one letter, as was found in the 408 as encipherment mistakes (or at least I assume they are mistakes and weren't intended to be multiply encoded).
Re: MY 340
by ALisowsky » Wed Sep 14, 2011 11:51 am
Sorry for being AFK for so long, real life duties calling!
I would like to continue to explore possible encryption methods prior to a homophonic sub, looking for signatures that remain after substitution, and ideally disproving some of these.. so far, we have done a lot (IMO) with this thread, such as all but eliminating vigenere, caeser, playfair encryption (which may have been done in the past, if so, sorry to be redundant but it's new to me).
All of this work of course assumes that its actually encrypted THEN put into homophonic sub.
Just so you know, I have been working on generating a similar string of data for the BIFID cipher technique next, detection methods, etc... I apologize to anyone reading this thread who is bored to tears!
Adam
I would like to continue to explore possible encryption methods prior to a homophonic sub, looking for signatures that remain after substitution, and ideally disproving some of these.. so far, we have done a lot (IMO) with this thread, such as all but eliminating vigenere, caeser, playfair encryption (which may have been done in the past, if so, sorry to be redundant but it's new to me).
All of this work of course assumes that its actually encrypted THEN put into homophonic sub.
Just so you know, I have been working on generating a similar string of data for the BIFID cipher technique next, detection methods, etc... I apologize to anyone reading this thread who is bored to tears!
Adam
Re: MY 340
by doranchak » Wed Sep 14, 2011 7:38 pm
ALisowsky wrote:Just so you know, I have been working on generating a similar string of data for the BIFID cipher technique next, detection methods, etc... I apologize to anyone reading this thread who is bored to tears!
Adam
Count me among the non-bored readers. Keep up the good work!
Re: MY 340
by ALisowsky » Thu Sep 22, 2011 2:44 pm
Wow.. ok the past 2 weeks have been spent trying to find SOME practical way to explore the possibility of a Bifid cipher being used, prior to a homophonic substitution. This is a damn hard cipher to decrypt, and some of the key indicators are nicely concealed when a homophonic is then applied to it. The following is really the only method I found for potentially exploring the bifid from behind a homophonic sub. It's not perfect, but hopefully it's a start.
First up, please check out my thread (its located in Wrench's thread about vertical n-grams) about why a bifid lowers the average number of symbols if a homophonic sub is applied to the ciphertext. Summing it up, we can assume that even the most commonly used symbol would probably not have more than 3 or 4 homophones. This is an important note, as none of this will prove anything conclusively, just (IMO) suggest whether or not a bifid was used.
To begin, here is my methodology. Please point out any errors you find!
1.Encode original cipher using bifid. For this example, we are encrypting the first 36 characters of MY340, using a period length = 6, and an unmixed Polybius square.
2. Encode the resulting cipher using homophonic substitution. Assume each letter occurring more than twice receives 2 symbols.
At this point, we have encrypted a cipher that is encrypted in bifid, then in homophonic sub. After a homophonic sub is applied, we can not apply a chi square test, which is normally used in BIFIDs to determine period length. This complicates things because it means there is no short-cut I could find to examine the z340 and determine accurately a period length.
3. With a standard BIFID, to decrypt, factor the resulting homophonic sub into row/column components and organize by period. To demonstrate naturals/half naturals, the original bifid periods are displayed as well.
4. Using the original bifid, we can identify the naturals (here simply for demonstration). At this point, we can insert the original plaintext to also see the half naturals.
But this does us little good, since we're assuming for this that 340 is encoded first in Bifid, then homophonic sub. As you can tell, the naturals/half-naturals tell us nothing about homophones in the homophonically substituted cipher, or bring us any closer to solving. This is where we have to be a bit...creative.
5. So next I looked for another potential bifid indicator, patterns. (I will be happy to explain this more if anyone really cares to hear it, but basically a BIFID can still leave certain patterns within a period length). I examined the 340 (quickly I might add) using Doranchak’s cipherscope to look for patterns, which are generated when the same word is encrypted inside a given period length. I found a suitable candidate, and went to work.
6. I factored both my340 and the z340 into period 9 breaks to search for naturals.
7. I assumed period length of 9, and broke the cipher into factors. This was done with the Zodiac 340, and also with my340. While not conclusive (due to the homophonic sub applied to the Z340), it also does not support a 9period bifid.
Whew.
Adam
First up, please check out my thread (its located in Wrench's thread about vertical n-grams) about why a bifid lowers the average number of symbols if a homophonic sub is applied to the ciphertext. Summing it up, we can assume that even the most commonly used symbol would probably not have more than 3 or 4 homophones. This is an important note, as none of this will prove anything conclusively, just (IMO) suggest whether or not a bifid was used.
To begin, here is my methodology. Please point out any errors you find!
1.Encode original cipher using bifid. For this example, we are encrypting the first 36 characters of MY340, using a period length = 6, and an unmixed Polybius square.
2. Encode the resulting cipher using homophonic substitution. Assume each letter occurring more than twice receives 2 symbols.
At this point, we have encrypted a cipher that is encrypted in bifid, then in homophonic sub. After a homophonic sub is applied, we can not apply a chi square test, which is normally used in BIFIDs to determine period length. This complicates things because it means there is no short-cut I could find to examine the z340 and determine accurately a period length.
3. With a standard BIFID, to decrypt, factor the resulting homophonic sub into row/column components and organize by period. To demonstrate naturals/half naturals, the original bifid periods are displayed as well.
4. Using the original bifid, we can identify the naturals (here simply for demonstration). At this point, we can insert the original plaintext to also see the half naturals.
But this does us little good, since we're assuming for this that 340 is encoded first in Bifid, then homophonic sub. As you can tell, the naturals/half-naturals tell us nothing about homophones in the homophonically substituted cipher, or bring us any closer to solving. This is where we have to be a bit...creative.
5. So next I looked for another potential bifid indicator, patterns. (I will be happy to explain this more if anyone really cares to hear it, but basically a BIFID can still leave certain patterns within a period length). I examined the 340 (quickly I might add) using Doranchak’s cipherscope to look for patterns, which are generated when the same word is encrypted inside a given period length. I found a suitable candidate, and went to work.
6. I factored both my340 and the z340 into period 9 breaks to search for naturals.
7. I assumed period length of 9, and broke the cipher into factors. This was done with the Zodiac 340, and also with my340. While not conclusive (due to the homophonic sub applied to the Z340), it also does not support a 9period bifid.
Whew.
Adam
Return to November 8, 1969: The 340-symbol Cipher
Who is online
Users browsing this forum: No registered users and 1 guest