Its been a horrible week, but I need the distraction... I've considered this a few times, every time I prepare to suggest it I remember what happened with my idea to test for long subjects, there just isn't enough uniformity. My concern isn't so much uniformity of technical things like tracking time zones and the like, but rather the way the world spins. A system that penalizes (or rewards) based on when (even if cross referenced when it arrived) would still have to deal with localization.
Can a system reliably know a message was sent during daylight/working hours from where it was sent? The only reliable way I can see is if Scott found a way (assuming the recieving server's clock was set correctly) to cross reference the geo code of the senders IP address with the arrival time of the message. BTW, the graph is amazing, how is it made? Dan On Friday, June 27, 2003 17:12, Markus Gufler <[EMAIL PROTECTED]> wrote: >Nachricht >Hi spam-fighters, > >What do you think about a time-dependently hold weight? >Maybe this can be helpfull on certain systems (where all users >work in the same time zone) to reduce FP's. > >For further explanation please see the PDF-file located at >www.zcom.it/decludeupdater/returncodes.pdf (280 kB). > >-The red dots are single messages over 24 hours (x) and their weight (y). >-The blue line is the average value of all weights in this time range >-The yellow line is our current hold weight of 100 points. > (consider it 100% if you hold on the default weight of 20 points) > >Now my suggestion/question: >As you can see, our server processes most legit messages between 8:00 AM and 8:00 PM >So why not increase the hold weight slightly in this time range >and decrease it a little bit on the resting time? (green line) > >Counting our FP's from the last 20 days and increasing the hold >weight during business time from 100 to 110 this will avoid 65% of them. >Naturally the increased hold value let pass some more spam >messages, but with 225 more delivered spam (that has recieved a >weight between 100 and 110 points) from over 14000 hold spam in >the last 20 days this is very few. > >Theoretically we all can create two identical configuration >files with 2 different hold weights and switch between this two >with a scheduled task. No additional ressources are needed. > >...or am I missing something? > >Markus --- [This E-mail was scanned for viruses by Declude Virus (http://www.declude.com)] --- This E-mail came from the Declude.JunkMail mailing list. To unsubscribe, just send an E-mail to [EMAIL PROTECTED], and type "unsubscribe Declude.JunkMail". The archives can be found at http://www.mail-archive.com.
