You seem well read smokie. I like that. A book can really make you drift away into another reality. As a kid, my father had this large bookshelf upstairs filled to the brim with books on UFO’s and cosmology, I could just pick one and go to sleep with it, wonderful memories.
What is the difference between increasing the general iterations and the general iterations factor?
The first determines the iterations of the substitution solver and the latter increases these iterations over time, but not for all solvers. The latter does not apply to the nulls & skips solver. But it does for the regular substitution solver, after every restart the iterations are multiplied by the iterations factor.
Have you experimented with the general hill climber iterations factor and / or do we need that setting?
The HC iterations are multiplied by this factor after every restart. This is something you can use if you do not really know how difficult the cipher is. We kind of decided not to use it for the optimization since it is not really a good way to determine solve rates.
Since I am still convinced that even very simple transposition steps can lead to a cipher that is extremely difficult to solve, here is a small test of my theory:
klHxJMs2q+cmAsKBd ksvDAnE3rhiysrzAf kjBOOPeBI+h8ickde v5DFXYj4EkfoTGrpQ 1+XSY2kAhHkZaT5LW q+W0ivrsjlcxBJm5P VWM+hdxqNKe+YOknP wLLzMkt+oR+pbdCDr c0SlTE+UumV+nYxy+ MdkA8eoIz5qtpNfr0 OqZXR8BgCi6D+WQkr +dyvyzdRjHNFTK0Lx GsUaJbEeIV+d5AZd+ lqKsHWBXmayFnAD8o +zMbNQhLi3E6J0PxC rSoRl65ZHSsrkm9Ad SWrcUjfqB+KL7wyr2 V+xt6ucd55f1nfca+ dshA2M+WHaiEJXA30 ++xY9yo+KNz0qLelJ
The cipher has many features that z340 also has. Perhaps I will be taught a better lesson and the solution will be found in a few minutes. Nevertheless, I can well imagine that this cipher is a hard nut to crack. I’m curious…
Since I am still convinced that even very simple transposition steps can lead to a cipher that is extremely difficult to solve, here is a small test of my theory:
It seems so. I cannot solve it. My transposition solver ran overnight and nothing. I wonder if you used a transposition that my solver does not support or something else. Tell me what you did I am anxious to know! Your cipher has alphabetic cycles such as "hij" and "lmn", probably something you want to avoid.
Hey Jarlve,
just two diagonal transpositions next to each other (underlying plaintext is continuous). The description rectangle was 20×17, the left and right half of which were transposed diagonally. That was transferred in 17×20 and then substituted. I’ve set 25% randomness. Here is a sketch:
Transpositions generated in this way very often have a peak at P19 (without substitution). After substitution, this peak can disappear, but is often retained. I don’t have concrete numbers yet, but I can draw up statistics if you like.
Your cipher has alphabetic cycles such as "hij" and "lmn", probably something you want to avoid.
The alphabetical cycles come from the fact that I didn’t shuffle the symbols when I created the key. I forgot that step. Here’s the key:
ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstu1vwxyz0+23456789 AAAAABBCCDDDEEEEEEEEEEFGGHHHIIIIKLLLMNNNNNOOPRRRRRSSSSTUUUVWXYZ
I did automated tests with z340 that tried all 256 possible variations of the method shown above. The left and right half transposed each with one of the following options:
So 16*16 variations.
I sent the results through my solver. That’s not really meaningful, though, because my solver isn’t really reliable yet. I can also write the 256 ciphers into a batch file and load them into AZDecrypt.
Translated with http://www.DeepL.com/Translator
Oh, blimey… I forgot we already had something like this:
Multiple inscription rectangles are usually stacked either horizontally or vertically, 1-dimensional. But yours looks like it is 2-dimensional which AZdecrypt does not support.
just two diagonal transpositions next to each other (underlying plaintext is continuous). The description rectangle was 20×17, the left and right half of which were transposed diagonally. That was transferred in 17×20 and then substituted. I’ve set 25% randomness. Here is a sketch:
I understand that the picture shows how you read/input the plaintext, but how exactly did you output it? I am not getting a clear decryption.
Oh, blimey… I forgot we already had something like this:
Multiple inscription rectangles are usually stacked either horizontally or vertically, 1-dimensional. But yours looks like it is 2-dimensional which AZdecrypt does not support.
I think you stacked it 1-dimensionally.
just two diagonal transpositions next to each other (underlying plaintext is continuous). The description rectangle was 20×17, the left and right half of which were transposed diagonally. That was transferred in 17×20 and then substituted. I’ve set 25% randomness. Here is a sketch:
I understand that the picture shows how you read/input the plaintext, but how exactly did you output it? I am not getting a clear decryption.
Did you use the same diagonal direction for each rectangle? If not, that may explain why my transposition solver cannot get it.
Did you use the same diagonal direction for each rectangle? If not, that may explain why my transposition solver cannot get it.
Yes. Both rects are using the direction shown in the sketch above.
I understand that the picture shows how you read/input the plaintext, but how exactly did you output it? I am not getting a clear decryption.
With peek-a-boo:
Resize the cipher to 20×17
Select the left half (10×17)
Select "Diagonal" from the "Transposition" DropDown
Klick "Untranspose" and in the next window "Apply" (Default-Direction is the correct one. Same as shown in the sketch)
Do the same with the right half.
Select in Main Menu "Tools -> Encrypt / Decrypt homophonic" and use the key given above
Okay, thanks. I got it now. Is it a good game? It has to be running on the SCUMM engine.
I understand what the problem is, while the rectangles are stacked 1-dimensionally in the vertical direction, each rectangle on its own is hard to solve since it is cut in 2. Will think about it how to add support for this in AZdecrypt. Perhaps it can already do this, not sure.
I got it now. Is it a good game? It has to be running on the SCUMM engine.
Yes, the game is definitely a milestone in the history of video games (and yes, it still runs today via SCUMM). From today’s point of view it is graphically no eye-catcher, but the story and the humor are great. Maybe you know "Day Of The Tentacle", which is the successor. And a few years ago, a game financed via Kickstarter was released, on which some of the original developers worked. It’s definitely worth every penny: Thimbleeweed Park.
https://www.youtube.com/watch?v=gESZ95d5pkU
I understand what the problem is, while the rectangles are stacked 1-dimensionally in the vertical direction, each rectangle on its own is hard to solve since it is cut in 2. Will think about it how to add support for this in AZdecrypt. Perhaps it can already do this, not sure.
I wonder if too much work should be put into the automated solving of such specific ciphers. Maybe there are other ways to explore alternative ideas. Many diagonal transpositions (stacked or not) with corresponding dimensional adjustments result in the P15/P19 peaks that we see in z340. Perhaps it is worthwhile to look at and evaluate various varieties of diagonal transpositions. Only then see whether automatic solvers are worthwhile. It’s just a suggestion…
And a general question: Do you think that the problem of z340 could be "related" to the example cipher shown by me? So that basically simple but not yet considered transpositions may be the reason that z340 has not yet been solved?
Translated with http://www.DeepL.com/Translator
Thimbleweed Park is on my Steam wish list. I am waiting for a 70% discount. Mark Ferrari is definitely one of my favorite artists. The things he does with color cycling… Will consider Manic Mansion and Day Of The Tentacle.
I wonder if too much work should be put into the automated solving of such specific ciphers.
Even within a period 19 scheme there are like a gazillion things that could have happened. For that I much prefer a solver since it can get creative with mixing operations and hill climbing.
And a general question: Do you think that the problem of z340 could be "related" to the example cipher shown by me? So that basically simple but not yet considered transpositions may be the reason that z340 has not yet been solved?
I was getting skeptical but you have shown at least one example and there should be many more. It could be and it is worth the shot.
Smokie, here is the new cipher: jarlve_p20_3nulls_5skips. It has a solve rate of about 13.5% with 10,000 HC and it this cipher I primarily want to use for the new optimization cycle.
1 2 3 4 5 6 7 8 9 10 11 2 12 13 14 15 16 17 18 19 20 6 21 11 22 23 24 25 26 27 28 24 29 30 31 32 33 4 10 1 24 34 35 24 36 37 34 26 38 39 14 40 1 25 41 24 42 43 44 39 37 45 46 26 12 32 28 47 48 15 49 50 33 43 26 11 34 3 21 51 40 20 48 5 52 35 37 43 3 48 53 14 25 7 42 23 50 54 30 22 43 26 34 32 51 55 24 56 37 57 31 28 15 25 58 10 24 20 23 33 26 19 42 51 10 24 20 59 17 9 15 35 47 48 50 36 28 60 30 42 24 13 2 36 7 23 22 31 60 38 45 43 13 19 14 24 5 59 52 12 44 37 9 57 22 32 10 33 22 28 40 6 30 7 25 26 17 40 24 10 4 61 25 39 21 15 26 29 33 56 47 34 24 34 2 41 24 46 50 33 49 35 51 48 53 24 3 59 43 24 48 42 12 4 37 60 15 24 40 31 24 9 52 29 5 11 54 24 57 42 36 8 5 43 12 13 38 15 55 44 22 48 57 12 10 17 14 45 7 18 11 19 16 29 32 47 22 49 25 24 26 21 2 45 33 39 33 47 10 40 49 51 1 25 11 3 26 35 22 11 5 52 51 34 6 41 15 42 54 43 37 6 57 7 24 41 12 10 36 28 22 20 22 16 36 62 29 48 23 1 11 13 27 10 15 17 14 21 22 40 17 21 32 13 14 14 38 32 60 33 24 63 10 24 47 9 24 53 34 24 ^:<],VTRQO2:&B1.U LY+'V82-;9CGA49_> 7#]O^9)Z9$H)GK01 P^CE9[M%0HX*G&74( W.?6#MG2)<83P'W,J ZHM<W"1CT[;6I>-MG )7359FH/4.C@O9'; #G+[3O9'NLQ.Z(W6$ 4S>[9B:$T;-SKXMB +19,NJ&%HQ/-7O#-4 PV>TCGLP9O]DC08.G _#F()9):E9*6#?Z3W "9<NM9W[&]HS.9P9 QJ_,2I9/[$R,M&BK. 5%-W/&OL1XTY2+U_7 (-?C9G8:X#0#(OP?3 ^C2<GZ-2,J3)VE.[I MHV/T9E&O$4-'-U$! _W;^2BAO.L18-PL87 B11K7S#9=O9(Q9")9
O.k., Jarlve, that sounds good. I put the message in the messages tested tab on the optimization spreadsheet. Looking forward to seeing how optimizations work out. Thanks a lot for your efforts.
I am going to resume testing on Jarlve 3/5, wanting to know where the point of diminishing returns is for homophonic hill climber iterations. I ran 500 restarts at 200,000 10,000 60 15 2, and got 12 solves. Will add 200,000 for another 500 restarts and so on. Curious. Will make entries on the optimization spreadsheet soon.
