Spammers probe SMTP mail servers on the Internet in an attempt to discover valid addresses at a domain.  They set their computers to try sending email to different addresses using a dictionary of common first name and last name combinations.  Since companies sometimes use a standardized name format as the first portion of an email address (before the @domain) these harvesting attempts may succeed.

Why does this "directory harvest" vulnerability exist?

The vulnerability arises because the receiving mail server under attack provides feedback about the non-deliverability of an email address.  The attacking computer notes this failed attempt and continues with other test addresses. 

For example, with Microsoft Exchange Server 2003, if the intended recipient does not appear in Active Directory, the following error message is returned to the spammer... 550 5.1.1 User unknown.  This signal indicates that the email address is not valid; those that are found to be valid will receive increasing spam from the spammer harvesting your email directory.

By comparison, Praetor never returns any signal or other information to the originating spammer.

What other problems can "directory harvest" attacks cause?

When under a directory harvest attack, the massive volume of attempts with different addresses sent by the spammer will consume a mail server's resources.  At worse, the mail server may be prevented from receiving legitimate email if all available resources are exhausted.