RE: Tag human v. automated submissions> (RE: Misconceptions)

> -----Original Message-----
> From: razor-users-admin@xxxxxxxxxxxxxxxxxxxxx
> [mailto:razor-users-admin@xxxxxxxxxxxxxxxxxxxxx]On Behalf Of Shawn
> McMahon

...

> I have no survey data to back this up, but I suspect that a rather large
> percentage of Razor users don't bother to submit something that's
> already listed as spam.  However, it should be possible to gather data
> on that from the server end.

Perhaps I wasn't clear enough -- this is unrelated to whether or not people
report a message as spam.  It's whether or not they act (report or revoke)
on a message, and what they usually do.  The possibilities are:

(1) Message was not tagged as spam and the user did not report it;
(2) Message was not tagged as spam and user reported it (Razor gets this
information now);
(3) Message was tagged as spam and user did not revoke it;
(4) Message was tagged as spam and user did revoke it (Razor gets this
information now).

Cases 1 and 3 are useful data, in my opinion, that Razor could gather.  Case
1 could help quite a bit to avoid problems like EFF is having, since the
system would see that lots of people received it but didn't report it.  I'd
give higher weight to people who tend to consistently report spam and have
high trust scores, since they're (obviously?) the ones who are paying
attention.  And higher weight to those whose reports are manual, rather than
automated.

One more thought on the human v. automated reports... I wonder if the time
difference between message arrival and Razor reporting would give a good
estimate of whether or not the reporting is from a person or automated?
Seems to me that in most cases, automated reporting would happen very
quickly, while human reports would have varying delays.

Nick



-------------------------------------------------------
This sf.net email is sponsored by:ThinkGeek
Welcome to geek heaven.
http://thinkgeek.com/sf