Re: [sitefinder-tech-discuss] Pseudo code please

[Andrew Newton <anewton@verisignlabs.com>]

Interesting numbers.

Kee Hinckley wrote:
> At 11:05 PM -0400 10/23/03, Andrew Newton wrote:
>
> I have a startup to run here, so there's a limit to how much free 
> consulting I can afford.  But here's $200 worth.  I did more error 
> checking on this run, so I'll restate my stats for bad-domain bounces.  
> They accounted for 8% of the bounces (the other 7% that I'd reported 
> before were bounces of mail that was forged to be from non-existent 
> accounts at our domain--my logs report both the same, so I didn't see 
> the difference when I did a quick grep last time).

I understand.  Mine came from grepping too.  Interestingly enough, I do 
have a Python script running stats on the maillogs every night, but this 
isn't one of them (I'll have to fix that when I get the time).  So 
rechecking this morning, I did quote a wrong number too.  0.25% was for 
rejects due to DNS failure, 0.2% for non-existent domain.

> This a summary of the 34 million messages rejected by somewhere.com from 
> April 2002 through September 2003 (18 months), with details for messages 
> bounced because the MAIL FROM had a unresolvable domain.  I don't have 
> full stats since then because Somewhere's mail is now being filtered by 
> Messagefire Enterprise.  Numbers are provided as a total count, 
> percentage of all bad domains, and a percentage of the total number of 
> messages rejected.  Mail that was not rejected because of a bad domain 
> was rejected because it went to a non-existent address.  (There are real 
> addresses at somewhere.com, but not more than a dozen or so.)  No other 
> filtering (e.g. DNSBL) was done prior to rejection.  Because of the 
> nature of somewhere.com's email (much of it comes from people giving 
> fake email addresses to web sites, and probably a third of it comes from 
> viruses) not all of the rejected messages are spam.  So the "% total" 
> number should be viewed as a lower bound.  The actual percentage of bad 
> domains in spam alone would be higher.

Perhaps its lack of coffee on my part, but are your stats drawing a 
distinction between unresolvable (as in a lookup failure) vs. non-existent?

-andy


----------------------------------------------------------------
To subscribe or unsubscribe from this elist use the subscription
manager: <http://lists.elistx.com/unsubscribe>