I understand that bigger ciphers show more recurring patterns and therefore should be easier (or at least faster) to decrypt as you'll have a better statistical analysis, being

However, if we use an evenly grown key, would this still be true? I've been wondering as I'm trying to decrypt a Vigenére cipher which seems (according to the analysis of the average Index of Coincidences) to have a key length of 81 characters in 1052 characters.

My problem is that after I calculated the chi-squared values, there are many very similar ones like 32.xxxx and 33.xxxx. If I'm to take both, I'd have to split the possible keys at this character's position to cover all possible ones - am I right on this? However, I couldn't be sure that those are really the key letters (as the value with 45.xxx could be also the right one).

This would result in many keys decrypting to equally many different plain-texts, which I would have to walk through manually. Is there a way to be more precise on finding the key I may have missed?

Even If I should've missed some letters out of those 81 - shouldn't those "errors" occur just once per miss every 81 enciphered characters? In this case, I just would have to guess about 10 right to get at least one fully decrypted word.

Here we're getting to the problem (or misunderstanding) I'm facing: According to the chi-square values calculated the most probably keys are just decrypting the cipher to nonsense-text and I can't even recognize single words.

How could I work from here further? Are big ciphers considered "easier" to crack because we do have at least the information I got as we may haven't even got those with a small one?

I hope my questions are clear - any help on resolving them would be greatly appreciated.

PS: To make it more visible than the key length is most likely really 81, here's the analysis of average IC values: Avg ICs

The text I supposed to be a Vigenére cipher actually wasn't one, it was a base64 encoded archive file. However, this does not change my question on how to overcome really big keywords appearing in great numbers. At the moment I'd think: take the one that's most probably it (according to chi-sq) and try every one of those.

Any other ideas?

  • $\begingroup$ @tylo I'll have to dive deeper into statistics for this, but my understanding was just that through IC (which shows how likely it is to grab the same character one after the other) we could get the key length, while by using chi-squares we could get the probably best fitting key characters through comparing actual absolute letter frequency to expected absolute letter frequency. However, this is one of the topics (statistical analysis) I want to study more asap. $\endgroup$ – UsuallyNot Oct 21 '15 at 16:45

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Browse other questions tagged or ask your own question.