Something more interesting would be to give then nominal scores, say
0.01. Then you can use hit statistics to determine whether or not you
should use the whitelist and how good or bad it might be. (I suspect
results might be surprising.)
{^_^}
On 2011/09/25 10:46, Roger Marquis wrote:
RW wrote:
I took the reduced negative scores recommended by Zimbra, inverted the
values and stored these in my local.cf.
That means you're penalising the better ones more than the worse ones.
Exactly. For those of us who don't track the accuracy of these DNSWLs
the best policy IMO is to simply disable them:
meta RCVD_IN_DNSWL_HI 0
meta RCVD_IN_DNSWL_MED 0
meta RCVD_IN_DNSWL_LOW 0
meta RCVD_IN_IADB_VOUCHED 0
meta RCVD_IN_IADB_DOPTIN 0
meta RCVD_IN_IADB_ML_DOPTIN 0
meta RCVD_IN_BSP_OTHER 0
meta RCVD_IN_BSP_TRUSTED 0
meta RCVD_IN_BONDEDSENDER 0
meta HABEAS_ACCREDITED_COI 0
meta HABEAS_ACCREDITED_SOI 0
meta HABEAS_CHECKED 0
Roger Marquis
