spf-discuss
[Top] [All Lists]

Re: Suggest New Mechanism Prefix NUMBER to Accelerate SPF Adoption

2004-08-25 16:39:45

As a non-mathematican I can only say that I think this whole idea of a
"probablility record"  is rubbish, because it allows me to "score" my own
spf records,


Either you have an idea how often a legitimate use of non-approved mail servers 
will occur or you do not.

It is a double-edge sword.

If you say you do not, then you can never declare "~all" or "-all" with 
confidence.

If you say you do, then you would not want the recipient to use a different 
assumption.


but not to control how that score is interpreted by recipients.


You can not control that with current system.

This would at least allow to you convey the information you know (or not 
because the number is optional).

If the recipient ignores it, you are no worse off.  Not all recipients will 
ignore it.


Conversely - I will see other peoples "score" on their records and wonder
what they  _really_  mean?


They will mean that they estimate xx% of the time any use of non-approved 
servers is forgery.  And (1 - xx%) of time is legitimate use.

 This concept introduces too many variables for a
world-wide, functioning system.


You already have those variables because every recipient is inventing his own 
number to use.  Some putting it in Bayesian incorrectly, others assuming other 
things, etc...


 I'll stick with pre-data verification of
non/forgery and then filter in spamassassin - simple, easy to implement,
and - most importantly - it works.  ;-)

What do you do with a "~all"?  What do you do with "?all"?