On 09Dec24, John Levine via mailop apparently wrote:
> It has been my impression that for many years about 90% of the mail a
> system typically
Does anyone have research which describes a "typical" mail system?
> a lot of attempted deliveries on my small MTAs.
Does anyone have research which correlates attempted deliveries with desired
deliveries
and actual deliveries?
If a spammer wants to deliver a single piece of content to a recipient and
tries with 3
different sender IPs and 4 different "MAIL FROM" addresses is that 12 attempts
for one
desired delivery or one attempt?
Here's some numbers on two systems.
On a 20 year old disposable-address-only system which rejects all invalid
delivery
addresses, about 12% of SMTP transactions triggered in a 5xx response. I don't
know how
many of those are redelivery attempts by spammers reacting to 5xx responses.
On a small sub-domain revivified after 20+ years of non-existance in the DNS,
100% of
attempted deliveries are accepted of which 100% are spam. Here's the arrival
rate since
revivification on 23Mar2023.
---------------|-----------------------------------------------------------------------------------
2023/02/Feb 10|**
2023/03/Mar 198|*****************************************
2023/04/Apr 273|*********************************************************
2023/05/May
396|***********************************************************************************
2023/06/Jun 155|********************************
2023/07/Jul 196|*****************************************
2023/08/Aug 182|**************************************
2023/09/Sep 180|*************************************
2023/10/Oct 133|***************************
2023/11/Nov 182|**************************************
2023/12/Dec 156|********************************
2024/01/Jan 122|*************************
2024/02/Feb 114|***********************
2024/03/Mar 97|********************
2024/04/Apr 88|******************
2024/05/May 75|***************
2024/06/Jun 95|*******************
2024/07/Jul 83|*****************
2024/08/Aug 89|******************
2024/09/Sep 272|*********************************************************
2024/10/Oct 237|*************************************************
2024/11/Nov 256|*****************************************************
2024/12/Dec 65|*************
---------------|-----------------------------------------------------------------------------------
I'd love to see a pattern in this arrival rate, but all I see is Richard Nixon
facing left
with his big nose at May2024 - Rorschach would be proud.
Unfortunately I have no reliable way of distinguishing between recipient
addresses which
remain on 20 year old spammer lists and recipient addresses which have been
recently
scraped off 20 year old web pages.
In terms of ipv6, less than 1% of delivery attempts came from ipv6 addresses of
which
essentially all where from legit gmail/outlook outbound servers.
So there you have it, 12% of delivery attempts on one system and 100% of
delivery attempts
on another. I believe that reasonably covers the range one can expect over a
decent sample
period.
Mark.
_______________________________________________
mailop mailing list
mailop@mailop.org
https://list.mailop.org/listinfo/mailop
