Is there an existing filter that could determine if a username@ is 60% or more mis-spelled as compared to real usernames? 60% is arbitrary and would be configurable. If so, that would serve to make a fuzzy honeypot filter for dictionary spam.
Right now all I can say is "kathey" is invalid, but might be a mis-spelling of "cathy" by a harmless stalker.
It's really a tough computer science problem. You have to take the order of letters into account, get a fuzzy pattern match. Could soak a lot of cycles, too.
Calculating a numeric value relating ascii and order of occurrence, a weird crc/genetics, might yield interesting results as one dimension of input. Doing it by single char, double, triple, could even work.
Definitely an AI thing.
Offenders could go in a local rblist, but not be reported to outside spammer databases, unless our self esteem increases logarithmically in proportion to a lack of serious whining.
-Bob
