This doesn't suggest "repairing" misreports, it suggests ignoring them as not statistically significant enough to affect a particular account's reputation. So, your automated FBL processing doesn't freeze the account, while the guy with a virus or a compromise gets his account detected and frozen by that very same script. "Handling" or "repairing" misreports any further is just not needed when you look at it that way. --srs On Sat, Aug 13, 2011 at 11:16 PM, Alessandro Vesely <vesely@tana.it> wrote:
This comparison apparently suggests that misreports ought to be handled and repaired, just like viruses or compromised passwords, however statistically negligible they might be. (It could even possible to devise methods to handle misreports so as to minimize the amount of time that the relevant abuse team has to spend on manual investigation.)
Is that what you meant?
-- Suresh Ramasubramanian (ops.lists@gmail.com)