On Wed, 16 Dec 2009, Paul Russell wrote:
On 12/15/2009 20:50, Steve Atkins wrote:
On Dec 15, 2009, at 5:48 PM, Rich Kulawiec wrote:
I think allowing end users access to such a button is a terrible idea.
This is the same population that routinely replies to spam, falls for
phishes, and fails to correctly execute rudimentary tasks like unsubscribing
from a mailing list or trimming quoted material from replies. It would
be simpler and about as accurate to simply check a random number generator's
output for a "spam?/not-spam?" opinion.
Data from actual reality contradicts your (otherwise plausible) reasoning.
Our "actual reality" includes an AOL feedback loop stream that consists
primarily of:
* spam complaints about personal messages from family members or friends;
* spam complaints about replies to messages initially sent by the user who
subsequently reported the reply as spam;
* spam complaints about messages posted to COI lists to which the user had
previously and explicitly subscribed.
None of these messages should have been reported as spam, but they were, and
they outnumber legitimate complaints by a wide margin.
If you don't send any spam, then 100% of complaints will be bogus. The
more revealing metric would be the ratios of proper and bogus complaints
to the total number of messages sent. If you send a million messages (all
non-spam) and get 4 bogus complaints, that is a failure rate of .0004%. If
you take the ratio of bogus to true complaints, that is infinity (since
there are no true complaints) or undefined. In any case, it isn't that
interesting a number.
Your situation, where there are a handfull of valid complaints differs
only in degree. The number of valid complaints does not belong in the
denominator. The denominator should have the number of good messages. Then
you can judge if the users are doing a good job or not.
Daniel Feenberg
NBER
--
Paul Russell, Senior Systems Administrator
OIT Messaging Services Team
University of Notre Dame
_______________________________________________
Asrg mailing list
Asrg(_at_)irtf(_dot_)org
http://www.irtf.org/mailman/listinfo/asrg
_______________________________________________
Asrg mailing list
Asrg(_at_)irtf(_dot_)org
http://www.irtf.org/mailman/listinfo/asrg