Interest level "maxed out".. Also wondering how much quicker your system would run the 408 without the filler. Wondering how much better the plaintext would come out if you remove the 18 cipher text fillers. So if that code really is only 390, I ran these numbers through Wolfram Alpha 390 to the power of 18 (not even sure if that was right) but the amout of extra permutations created was staggering. Pointless posting it up . was huge it was. Same went for the 340.

Ps. Is it like we see in the movies, watching your screen with millions of letters rolling through.

0.6

– 10-20% faster.
– Made program unable to switch between fullscreen and windowed. This caused problems.
– Expanded max threads to 128 if you happen to own…

I see you’ve gone all "A Scanner Darkly" jarlve.

Well observed.

Your signature message, Quint – Jaws? (didn’t google)

Well observed.

Your signature message, Quint – Jaws? (didn’t google)

Et tu. It is.

One of my favorite films and, I thought, an appropriate disclaimer in reference to some of the things I’ve suggested/posted/thought over the years in relation to this case.

Well observed.

Your signature message, Quint – Jaws? (didn’t google)

Et tu. It is.

One of my favorite films and, I thought, an appropriate disclaimer in reference to some of the things I’ve suggested/posted/thought over the years in relation to this case.

Yes I agree, it’s a very good reference to the Zodiac case! I own the 30 year anniversary edition of Jaws, the book and the soundtrack. I love Robert Shaw, who also plays Doyle Lonnegan in The Sting, ya follow.

0.7

Happy to share some much needed functionality. It is now possible to put all ciphers in one file but you need to be careful about a few things. Old functionality remains and can be mixed.

Example:

FILE START
results_sub_directory=group_01/

cipher_information=408.txt, part three
NË£SÃE/¼ººZÆAP·BV
ÐÅXÑWѸF·¾Ã+¹¼A¼B
ºOTµRuÃ+¸ÄyѸ^SÑW
VZÅGyKE¸TyA¼º·LÔ¸
H‚FBX¼¤XADĽL‚¥Ñ
¸ÅÄ··°ÅµPORXQFºGÃ
Z¹JTÔѸ¾JI+ÒBPQW°
VEXÒ¼WI°ÑEHM£¥uIË
END

The first line must be the output sub directory, which will be automatically created under results/. In this format "results_sub_directory=yourdirectory/". Each cipher must be seperated by one empty line (no more, no less) and at the end of each file there may be no more than one empty line. I’ll try to loosen everything up for the next version.

You may have one line in the format "cipher_information=408.txt, part three" at the top of each cipher (no empty line between this and cipher), that will be recognized and bundled in a file called group_list.txt with the scores and everything for after analysis.

Example of group_list.txt output located under your specified directory:

408.txt, part one (1065,673,136,51)
408.txt, part two (1119,684,136,54)
408.txt, part three (1084,662,136,51)

The format between the brackets is (score,ioc,symbols,unique symbols).

An example of this, ciphers/1.txt is provided with 0.7. If anything let me know.

I also added an option "reset thread-flags (r-button)", this only works when threads are set to 0.

There is currently still a problem when either to many threads are called and/or you are using close to 100% of your cpu. This may cause some of the thread numbers to be unused, which you can check on the left hand side of the program window. If so, drop the threads to 0, wait for them to finish and then press the r-button and everything should be okay again. You can also just increase the thread numbers to compensate. I want to add that this problem does not crash the threads, it just mixes up some variables.

0.8

Edit: most of the threading problems I spoke of earlier seem to be fixed. Though I recommend not using 100% of your cpu at any time just yet.

Made a new startup screen where you can click to change values and added the option to only save ciphers which score over a specific value. I changed the name of the program a bit.

0.9

I’ve managed great improvements to the program, it is about twice as fast and score stability and output quality both have improved because all letters can now be used.

The default score system has changed to normalized, which normalizes the scores of ciphers with different character counts quite well. It is possible to make a custom letter file (see corpus/abc.txt) where you type in the letters the program is allowed to assign. Also a custom IoC target is now possible.

I’ve added three presets for "keys per cipher:", very fast, fast and balanced. Very fast and fast offer remarkably good recovery for the time invested. If you can spare the time the balanced preset is the way to go and has a very high recovery rate for ciphers around the multiplicity of the 340.

0.91

Fixed a minor miscalculation with the IoC. Managed to improve the scores a great deal, especially so for the faster presets to the point where I made the fast preset the new default. Changed the name of the balanced preset to thorough.

Added 2 new buttons, "edit" besides letters which opens the letters file in notepad for easy editing. And "get ioc from corpus" which does exactly that, and is very handy when working with other languages etc. It also takes in account the letter file, meaning, if some letters are not included the ioc will be adjusted accordingly!

If you want to use other languages make sure the corpus you use is the same size as english.txt shipped with the program. This is because some internal values are optimized to this size. When I finally decide to release the program as non-BETA I will probably make a youtube video on how to use the program.

I’ve noticed a bug that when you launch specific games (on Steam) and the program is running in the background, this sometimes causes the program to be minimized and locked to the taskbar. The program continues to run and accept keyboard commands but the window is gone with only a preview in the taskbar and windows-key+tab to judge from. Looking into it.

Edit: I also want to note that support for other languages atm is not perfect because only the 26 letters used in english are considered.

0.92

Massive improvent in recovery rate, scores etc. Several of my test ciphers (1000 samples per) scored a 100% recovery rate on the default "keys per cipher" setting. This is really getting somewhere. Noticed a possible performance scaling issue for multi-processor systems (not multi-core), for the moment you can workaround by launching several instances and distribute the workload. Also noticed a possible lockup/crash with low values for the "keys per cipher" setting, for now I made the minimum 1400000 until it’s sorted out.

It has been a great week for the program and I’m really pleased with the progress I’ve made.

0.93

A small bug fix pertaining to the scoring mechanism. Adjusted some variable types so that they don’t go negative anymore on very large amount of ciphers.

Great tool..still exploring it. Hm..checked it out as far as I could. Some difficulties with it:

1. On my win8-64bit-pc the program seems to crash as soon as I try to modify the number of keys or the score goal. If I simply click on ‘start processing’, however, it works fine. But I can’t modify anything without it to crash..
2. Any possibilities to define vowels or non-vowels?
3. Is the program based on quad-grams? The 340 does not have many of those..
4. Depending on the corpora, the solution is always a different one..

Very fast tool..

QT

Great tool..still exploring it. Hm..checked it out as far as I could. Some difficulties with it:

1. On my win8-64bit-pc the program seems to crash as soon as I try to modify the number of keys or the score goal. If I simply click on ‘start processing’, however, it works fine. But I can’t modify anything without it to crash..
2. Any possibilities to define vowels or non-vowels?
3. Is the program based on quad-grams? The 340 does not have many of those..
4. Depending on the corpora, the solution is always a different one..

Very fast tool..

QT

Thank you for the feedback,

1. I’m not experiencing crashes when changing the values on any of my 4 machines (netbook (win7 32-bit), desktop (win7 64-bit), workstation (win7 pro 64-bit) and ultrabook (win8.1 64-bit)). But since it is a custom "UI", the way it works is that you click on the number of keys "button" the program then only accepts keyboard input, so then you’ll have to input a value and at least press enter to have the mouse working again. I have noticed a crash with the current version when trying to get the program to work with a custom alphabet file.

2. Do you mean, predefine what symbols have to be vowel or non-vowel?

3. The scoring mechanism is similar to that of ZKDecrypto. For every character (minus 3) in the cipher a quadgram is scored. For instance, the word "preview" contains 4 quadgrams: (prev,revi,evie,view) these are then compared to the scores in array (quadgram(25,25,25,25)) derived from the corpus and added to the score. This is brain/intelligence of the program and I’ve done extended testing on using a combination of 2,3,4,5-grams and came to the conclusion that only using quadgrams gave the best speed/quality ratio. http://practicalcryptography.com/crypta … quadgrams/

4. I have recently not tested the program with different corpora but they have to be very close to the size of the original one in order for the program to work optimally. I recommend the current one for english and there is little support for other languages atm. I can guarantee to a high degree that the program will find a solution (strong global optimum) when one is available. When no strong solution is available the output becomes very dependent on the corpora and IoC. If you think you may have a partial solution I recommend you try the same cipher in ZKDecrypto, if it comes up with similar results then there may be something there.