Checking Success of Cryptanalysis: Computer's Stopping Point

[mtanti]

How do you check if a cryptanalysis was successful? I'm talking about a super computer not a human reading the output text. When does a computer stop cryptanalysing?

Or even whilst doing bruteforce at millions of keys per second. When will a computer stop?

 

[chroot]

Very commonly, parts of the plaintext are known before cryptanalysis is begun. For example, many kinds of electronic communication include headers or other kinds of "framework," and the cryptanalysis program could flag any decrypt which contains those known bits of plaintext.

Alternatively, one could use frequency and contact analysis on the candidate plaintext -- if its frequency and contact charts are close enough to those of the plaintext language, it could be flagged for a human cryptanalyst to take a look.

- Warren