Reply To: Help with hillclimb
The mystery of the silver bullet › Forums › The Intelligence Room › Help with hillclimb › Reply To: Help with hillclimb
It’s not one character difference in the output but rather one character change in the substitution cipher key causing a big difference in the fitness of the decrypts. We can see that the substitution decrypt is correct as we’ve verified it against the output we get using the BOSS tool.
The issue seems to be that the hill climb tries to change only one character at a time and if it doesn’t find the right letter in (say) position 1 of the key on its first go round (whilst all the other letters in the randomly generated key guess are still wrong) then it doesn’t incrementally start moving towards the correct key. In particular we have seen that when the key guess has duplicate letters (which get removed when decrypting and so ‘shift’ a lot of the other letter mappings) we get very different decrypts from relatively similar keys, which seems to put the hill climb off the right scent! We’re using Challenge 4B to verify this is working so we know what the right key is and what the correct decrypt is.