Page 4 of 6

Re: Giving a talk at this year's Cryptologic History Symposi

PostPosted: Thu Oct 22, 2015 4:58 pm
by doranchak
Here are the details - http://www.isi.edu/natural-language/mt/ ... ght-14.pdf

They use some kind of machine learning algorithm. It recognizes from among 50 different cipher types with accuracy of almost 60%.

Re: Giving a talk at this year's Cryptologic History Symposi

PostPosted: Thu Oct 22, 2015 5:01 pm
by doranchak
A cool thing today was being in the same panel with Dan Olson from the FBI. He gave an interesting talk about the history of the FBI's codebreaking.

If I see him during tomorrow's sessions, what should I ask him?

I also met the author of this paper about the Kaczynski ciphers: http://www.researchgate.net/publication ... 's_Ciphers I requested a copy of the paper because I'm very curious to know how the Unabomber's crypto-system works.

Re: Giving a talk at this year's Cryptologic History Symposi

PostPosted: Thu Oct 22, 2015 5:11 pm
by daikon
doranchak wrote:A cool thing today was being in the same panel with Dan Olson from the FBI. He gave an interesting talk about the history of the FBI's codebreaking.

If I see him during tomorrow's sessions, what should I ask him?


Personally, I'd like to find out how much effort FBI put to decoding Z340, which encryption methods did they consider in particular, and how far did they get with each one. But I highly doubt that he'll answer with any specifics, as it'll likely be "national security matter". For example, I'm pretty sure FBI has the technical capability to brute-force all columnar transpositions of Z340. It would be nice to know if they tried and whether they succeeded or not, so that I can stop thinking about it. :)

Re: Giving a talk at this year's Cryptologic History Symposi

PostPosted: Thu Oct 22, 2015 5:15 pm
by doranchak
OK, I'll try to ask him about that.

Re: Giving a talk at this year's Cryptologic History Symposi

PostPosted: Thu Oct 22, 2015 10:36 pm
by ZodiacRevisited
Great stuff, Dave. Definitely looking forward to seeing the video.

Re: Giving a talk at this year's Cryptologic History Symposi

PostPosted: Fri Oct 23, 2015 6:35 am
by morf13
Dave, did anybody offer any new or valuable ideas, or theories about the Z ciphers?

Re: Giving a talk at this year's Cryptologic History Symposi

PostPosted: Sat Oct 24, 2015 5:37 am
by doranchak
During the Q&A Rich SantaColoma (a Voynich researcher) asked if anyone had applied the 408's key to the 340 and looked for transpositions that might make sense of the result.

I'm pretty sure people have tried the 408's key, which produces gibberish (and not all symbols are in the 408's key). But I haven't heard of anyone trying out different transpositions of the gibberish plaintext.

Also during the Q&A I was asked if Zodiac might have had dysgraphia (an idea glurk brought up a while back: viewtopic.php?p=10861#p10861). The questioner has dysgraphia himself and said he thought Z's letters and ciphers had some of the features of the condition.

Another person I met was curious about the repeating fragments (such as these trigrams: http://i.imgur.com/Jz7f8d6.png). He referred to them as "isologs", a term I hadn't heard of before. He wondered if some superencipherment scheme (additional step after the first encipherment step) might have concealed some patterns but revealed others (i.e., the isologs). He also asked if I knew of any periodicity tests being applied to Z340 (basically, distances between repeating patterns). I think some such tests have been done before but I don't have a good summary of them.

Several folks were intrigued about the various oddities in the 340 that they hadn't seen before (the pivots, odd/even ngram discrepancies, suppression of bigram repeats (but peaks when counted at distance 19), etc). I think those things made people curious about what might be going in the 340's construction.

I met George Lasry, a very skilled Israeli programmer who during a period of unemployment cracked several notable unsolved historical ciphers. This got Google's attention and they hired him. Anyway, I tried to encourage him to add the 340 to his list of projects but he's reluctant because he doesn't know the encipherment scheme. I think that makes many people reluctant to tackle it.

I learned about this paper which reveals Ted Kaczynski's cipher system: http://www.tandfonline.com/doi/abs/10.1 ... ode=ucry20 I really want to get a hold of that paper since I've been trying to get a copy of Kaczynski's cipher-related material from his journals for some time. I met the author at the conference and asked her to email me a copy of the paper - hopefully she will follow through. She was with the FBI gang that showed up for the Crimes and Ciphers panel that Dan Olson, Klaus Schmeh and I spoke in. BTW Dan didn't attend the 2nd day of the conference so I lost my opportunity to ask him more questions. I'll try to contact him via other means. I did ask him if there was any trend of criminals getting more sophisticated with their cryptographic schemes. I got the impression from him that there wasn't.

Dan also mentioned the 340 is still on the FBI's Top 10 unsolved codes list. I missed my chance to ask him what else was on that list - anyone know where to find it?

Elonka Dunin was telling me a story about how she solved someone's challenge cipher (a playfair cipher, I believe). She first failed to solve it, but then realized he may have been drunk when he made it. So she reasoned about the kinds of mistakes a drunk person would make during its construction, and managed to solve it. I encouraged her to work on the 340 because that's the kind of thinking we need from codebreakers -- trying to guess what the cipher writer might have done. But she's reluctant since it's a creepy serial killer case, and she's already preoccupied with Kryptos. (Kryptos side note: She's hosting a dinner tonight and Kryptos sculptor Jim Sanborn and its cryptosystem consultant Ed Scheidt will be attending. I think the group is going to try to squeeze more info from them about K4, the last remaining unsolved portion of Kryptos).

I met Dennis McDaniels, a legendary retired NSA codebreaker who solved the other three parts of Kryptos in 1992. He had a lot of fascinating stories such as helping to break the code heard on a cassette tape, which led to the capture of an escaped fugitive who had been out in the wild for 20 years: http://www.harryjaffe.com/archive/the-fugitive

Great conference, filled with loads of fascinating and talented people! I hope I have learned a few things via osmosis.

Re: Giving a talk at this year's Cryptologic History Symposi

PostPosted: Sat Oct 24, 2015 5:47 am
by traveller1st
Great stuff Dave. Thanks for the summation.

Re: Giving a talk at this year's Cryptologic History Symposi

PostPosted: Sat Oct 24, 2015 6:01 am
by Jarlve
Very interesting ideas!

I'm glad you got the word out on the 340.

Superencipherment, isologs, some stuff to think about. :)

Re: Giving a talk at this year's Cryptologic History Symposi

PostPosted: Sat Oct 24, 2015 6:35 am
by doranchak
My biggest hope is that someone there will take one of the ideas and run with it, finding a solution so we can move on to other things. :)