Reply To: Challenge 10B
I know this isn’t exactly what we were set, but I’ve been having a go at solving 10B as a 16×36 grid as I thought it should be possible and although the order of the cards on each axis doesn’t matter using this method I’ve set them to the same order as that used in the offset foursquare so the patterns show up.
After many attempts at using a similar algorithm to that used when solving a 5×5 grid I found that the only way I could get an improving result was by going through each position in the grid and swapping the pair of characters in that square with every other position in the grid and using simulated annealing and a quadgram test at each position. Each iteration took about 1 minute and after about 100 iterations it looked quite good and after 548 iteration it got to the best result as the next output was at 1265 iterations and the quadgram score was better but the result looked worse which was confirmed later when I checked the percentage accuracy.
At this point I had approximately 85% accurate result when compared with the plaintext, so I re-ran the algorithm using quintgrams (which took 10 minutes per iteration) using the output from the previous program as the starting point and got an improved accuracy of approximately 95%, with the same problem I was getting before of the result getting worse (at around 37 iterations).
I’ve started reading the text and manually updating the key and since I had not actually read the plaintext the guesses I’ve been making have not been influenced by prior knowledge, I’ve also ignored the patterns showing up in the key as I wanted to show that it could be solved as if using a random key.
For anyone interested these are the keys I’ve got so far (which give a 97.6% accurate result):
key1 =
“SBAAOGSHMDAWSKADCWSHUDOWSHXDOWGHBBOISFGDOVSHLDOWSLABOWSMPMOPSHADOCSYPDOWPBADOMSCADOWSUMDOGSGADOAUMKYLSHDSNBKSIHDGWSHADOWTKACOMUVADOWSNADOLTFAKRHLFEFLUBCEKGIEFMVLIBCEFGDICEFSIMTEAGIBCEHMIGCEFGIPYEFXIDCERXIYTEFHILTELLOXXEQGIICEFGIRRELGILQEAKIWUDPHGXSEAMXVCAFGDMCEFGIYXELGIXPENGIPCEQGIBGEFGUKSRNKXKLMNPXKYMOMVKLMNPTBLVNPUBNSNBHXLHNQCKLMNPAFAONVKHLFNFKHBMPPPQVQNYQQLXNPUQUMBPHVLBNFQXRVNPVQABRGFMWPXXXVIBNVKKWMNFUWAINDNWLOUQQQEVNCWBLBNPQSTKDWYRTMVQYRVQXWYRTYVHYRTWVBYXMUFQIYTOFOGRTUPDYRCUUXYRFHVEVRCUWHTWTWPCARTUVMQTTULXYRXQQCYRTHVUYCQKWKKFKUHNBRTCBYQRTUCVYRKXCTYRVUXWYRTUBPYRTPQHK”
key2 =
“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”
With one character from each key making up each square and with the first 36 character pairs (and subsequent 36 character pairs) making up each column of the grid.
Since about a third of the pairs of the characters in the grid are not used in the plaintext there are quite a few random characters in the key.