Re: SPF deployment, was RFC 3929 on Alternative Decision ...

John Levine writes:

> 
>>>The amount of mail going through SPF tests is still small, and only a few
>>>aggressively wacky sites block mail that fails SPF.
>>
>>With the release of SA3.0, this is likely changing.
> 
> Small amount of mail tested, or few wacky sites?  Both, I suppose.
> 
> If people really are doing tests of SPF or Sender ID or other similar
> schemes, I'd be most interested in hearing about the results,
> particularly numbers of how much legit mail is tagged valid or
> invalid, how much spam is tagged valid or invalid, and how spammers
> are adapting/ Ciphertrust says they've seen more spam than good mail
> pass SPF checks.  That doesn't surprise me, but their numbers were so
> small that I'd like to see some confirmation.

After DNSBLs, SPF is my most successful spam filter.  Here's my 
latest mail report (# of messages rejected/reason).

7951	550 User unknown.
6182	517 HELO mismatch
1310	502 ESMTP command error
1171	511 Blocked by www.spambag.org : Blocked
839	517 Sender Policy Framework error
81	551 DNS lookup failure
74	558-500-sam:558 500 Message rejected by recipient's mail filter (BCC).
67	Connection timed out
35	500 Microsoft virus refused.
22	550-The headers in this message contain improperly-formatted binary…
21	513 Relaying denied.
21	550-This message contains improperly-formatted binary content…
20	550-This message contains text that uses unnecessary base64…
14	517 Syntax error.
<= 10	79 errors for 44 types


SPF blocks ten times more spam than the next most common spam filter.  Also, 
a fairly good portion of the "HELO mismatch" rejects would've probably been 
caught by the SPF checks too.

I have occasional reports of false positives, but they were related to other 
causes.  No false positives reported for SPF.

-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.2.4 (GNU/Linux)

iD8DBQBBhPqQx9p3GYHlUOIRAsfmAJ9cIv/WmDW7n3T+WG5cyuLmnDNzJwCdE+c/
kiQA1ksBkrUKKJMd3+Glx88=
=70BK
-----END PGP SIGNATURE-----