This new format seems to be working OK, so I’ll put up another con. This one is a random unknown type from Gutenberg.org. Although grammatical in form, the last few words make no sense with the beginning and in fact are the beginning of another sentence from a different book. My program sometimes grabs a second sentence in order to meet length standards. My Analyzer identified the correct type and my hillclimber solved it quickly.
ksupbatpnzikmpwilpvwgeinpantqvxsibhqbgdiratnziqqidvjqpgeutidyjaksysppgtnjfylnavampisyktxibpysqvgtucipprnaxsizljbne
My hill-climber got it quickly after I guessed the type by trial and error (let’s see, what else haven’t I tried?). I need to do more work on diagnosis!
Was the second sentence fragment from Pepys? There were a number of hits, but that seemed like one of the more famous works.
Yes, it was. I had to look back at the worksheet and search the phrase to get the full sentence, and then search a more unique quote from that to verify it. My program doesn’t identify the source when it randomly chooses a sentence. Gutenberg uses a numbering system in their URLs, so my program just generates a series of random integers and feeds them into a Get statement. The rest is parsing until I find a clean sentence somewhere into the text. Neither the work nor the author are recorded.
The results of my ID tests depended a lot on the maximum period. With a length of 114, some ACA types could go up to a key length of 12. With a max key length set at 12, the correct type was in second place in the random forest test and in first place in all the neural net tests.
My hill-climber solved it after about 8 million trial decrypts, without a crib of course.
The key expander found a two word phrase that would give the key, but couldn’t find a one word key.
The first Gutenberg passage in the plaintext continues (caesar shifted): F RFY TK