You can find this concept elsewhere or derive it yourself but for those who haven’t considered it, here is a brief analysis of keyword choice.
In forming alphabets such as the one used in Kryptos’ Vigenere table, the keyword can serve to make it more challenging for the analyst.
K R Y P T O S A B C D E F G H I J L M N Q U V W X Z
One thing that must be considered by the encrypter is that the alphabet remains unchanged after the latest occurring letter of the keyword. In the example above, the latest letter is Y so Z will remain in the same position with any use of this keyed-alphabet. The choice of keyword can weaken the strength of the encryption if you choose a short keyword or one that has an early set of letters. Consider the keyword “CAB” and how it will change the alphabet.
C A B D E F G H I J K L M N O P Q R S T U V W X Y Z
You can see that after the first three letters, the keyed-alphabet is basically unchanged. A skillful cryptographer would avoid an analog encryption these days but if forced by necessity he would choose a keyword that allows for a reasonably altered encryption alphabet. This phenomenon applies to monoalphabetic substitution ciphers as well as some of the Polybius/Playfair methods of encrytion. If Ed expected us to use brute force attacks, he would intentionally have Sanborn pick a keyword that allowed for partial recovery of the plaintext via a small keyword with early occurring letters.
For keyed-transpositions, the idea is a little different. Keyword length is important as well as having any repeating letters. I don’t personally know how to do this but the alphabetical makeup of the keyword is important. With “KRYPTOS”, the resulting numerical key is 0362514. If this is a common numerical key resulting from a keyed-cipher then weaknesses become inherent in the system. This is outside of my area of expertise so I’ll only lightly brush that topic.
So what does this mean for K4?
If one is attempting a brute force without the keywords, there is hope that we can recover partial sections of plaintext from a weakly-keyed cipher that will allow us to guess at the keyword or reverse-engineer it by completing words of the plaintext.
Are there other implications from key choice? Yes but I won’t be able to describe them because I don’t know much more than this. Perhaps a cryptology textbook would serve as a better resource but I think the topic is worthy of discussion even coming from an amateur hobbyist such as myself.
Am I claiming I can do this? Not really, I just know it’s a possibility.
I’d like to believe there’s hope for brute force attempts but I think it’d be easier to try with some idea of how we were supposed to find the keywords for the other sections.