I went to the abaca website and read the technical info. It seems like the core problem is how to identify that 2 or more incoming messages are essentially the same message. If each message is truly unique, and cannot be associated with other messages, then the method seems to fail.
But - technically sophisticated spammers have already been introducing deliberate variations in copies of their messages. The abaca website and its white paper says nothing about this, beyond making a cursory claim that the method can indeed find all copies of an underlying message.
So there is an undisclosed and key step in the method. Where operations are performed on an incoming message, to try to remove these deliberate variations. Why isn't this discussed in the website or paper?
Slashdot Mirror
I went to the abaca website and read the technical info. It seems like the core problem is how to identify that 2 or more incoming messages are essentially the same message. If each message is truly unique, and cannot be associated with other messages, then the method seems to fail.
But - technically sophisticated spammers have already been introducing deliberate variations in copies of their messages. The abaca website and its white paper says nothing about this, beyond making a cursory claim that the method can indeed find all copies of an underlying message.
So there is an undisclosed and key step in the method. Where operations are performed on an incoming message, to try to remove these deliberate variations. Why isn't this discussed in the website or paper?