I have a series of messages (limited to only a-z and A-Z) that are encoded such that the sum is taken of every two ASCII values and concatenated. For instance, the message hello would be decomposed into the pairs he, ll, and o which would then become 104 + 101, 108 + 108, and 111 which finally is output as 205216111. Each pair is guaranteed to be 3 digits by zero-padding.

Is it possible to trivially decode this final number back into the plaintext? The words are in plain English and each word is capitalized. So far I have attempted to enumerate all possible permutations and identify words from a dictionary list but this has proven to be rather tedious with short messages alone.

  • 1
    $\begingroup$ "this has proven to be rather tedious with short messages alone" - surely, you have a computer doing the searching, don't you? $\endgroup$
    – poncho
    Apr 13, 2016 at 2:32
  • $\begingroup$ @poncho Of course, It is just that the possible pairs are many in number and while a dictionary attack can identify words it cannot discern sentences: small articles often end up making a word soup that makes a totally automated approach nigh impossible $\endgroup$
    – J. Doe
    Apr 13, 2016 at 10:02
  • $\begingroup$ Sounds alot like a Trifid cipher. en.wikipedia.org/wiki/Trifid_cipher $\endgroup$
    – user33402
    Apr 13, 2016 at 10:27
  • $\begingroup$ Attacking this is very similar to attacking a many-time-pad. One advanced attack example: crypto.stackexchange.com/questions/59/… $\endgroup$ Apr 13, 2016 at 10:55

1 Answer 1


It would be impossible i think. But if you consider evolving the encrypion system such as changing pair like : "he","el","ll","lo","o" it would be way to easy to decrypt.


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.