> From: Rolf E. Sonneveld [mailto:[email protected]] 
> Sent: Tuesday, February 08, 2011 2:53 PM
> To: Murray S. Kucherawy
> Cc: [email protected] WG
> Subject: Re: [ietf-dkim] Interesting recent statistics
> 
> [...]
> 
>> 1) There's a slow but steady increase in signed message rates:
>> http://www.opendkim.org/stats/report.html#signing_trend (ignore the first 
>> row, that month was skewed;
>> also the current month is, of course, incomplete)
> 
> Good news! Although... we don't know whether the increase is due to spammers 
> that adopt DKIM or legit
> senders adopting DKIM.

We have some hints in the data, since I have some correlation between 
DKIM-specific message data and SpamAssassin results.  I just need to build the 
right SQL query.  I'll post again when I have that.

>> 3) Overall pass rates are pretty high, even when they transit MLMs:
>> http://www.opendkim.org/stats/report.html#passfail
> 
> What I don't understand here, what's what. I see four figures:
> 
> Signatures: 1362786 Ignored: 482 Pass: 1275158 Failed (body): 16027
> 
> But this row does not provide the correct pass/fail rate in total, does it? I 
> thought the pass rate was
> (1,362,786 - 16,020) / (1,362,786), or if we subtract the ignored: (1,362,786 
> - 482 - 16,020) / (1,362,786).
> This, however, is not correct: the number of failed is not the total number 
> of failed, only the body failed
> (as is indicated). Is it possible to add a fifth figure: failed (header), to 
> this row in order to have a
> complete list of figures?

Pass/fail is an indication of whether or not the signature verified.  A 
signature fails to verify if there was a header hash mismatch or a body hash 
mismatch.  Thus:

        failed(header) = total - ignored - pass - failed(body)


_______________________________________________
NOTE WELL: This list operates according to 
http://mipassoc.org/dkim/ietf-list-rules.html

Reply via email to