nanog mailing list archives

Re: Fun new policy at AOL


From: "Stephen J. Wilcox" <steve () telecomplete co uk>
Date: Fri, 29 Aug 2003 00:05:50 +0100 (BST)



On Fri, 29 Aug 2003, Dr. Jeffrey Race wrote:


On Thu, 28 Aug 2003 12:07:30 -0400, Matthew Crocker wrote:

It can be built without choke points.  ISPs could form trust 
relationships with each other and bypass the central mail relay.  AOL 
for example could require ISPs to meet certain criteria before they are 
allowed direct connections.  ISPs would need to contact AOL, provide 
valid contact into and accept some sort of AUP (I shall not spam 
AOL...) and then be allowed to connect from their IPs.  AOL could kick 
that mail server off later if they determine they are spamming.

Now there is an idea!  However an improved variant is to make the
entire internet a 'trust relationship' using the (obvious) steps you
propose.   For several months I have been pondering possible details of
implementing same; see <http://www.camblab.com/misc/univ_std.txt>.
Comments welcome.

Surely it already is ? That is I only announce routes of my customers who I 
trust, my upstreams and peers trust me and what i announce to them, their 
upstreams/peers do and so on. And yet we still have hijacked netblocks and 
ddos's with uncaring sysadmins. Why should email be any different?

And if you do implement such a system, the spammers will just adapt.. the recent 
viruses (sobig) are an example of how spammers can open up end user machines to 
facilitate sending of email, providing they can control such a host they can 
simply relay thro the providers' smtps.. they dont need open relays to send out 
their junk! 

I think we're still trying to treat the symptom tho not the cause. Most of these 
spammers are companies based within our countries, if we can make their kind of 
advertising illegal the spam will reduce (not sure if it will disappear, it 
could be like tax - companies may open offshore offices to facilitate this, but 
we need to keep working on the cause... )

Steve


Current thread: