re spam insanity

[Dave Farber <dave () farber net>]

Begin forwarded message:

> From: Dave CROCKER <dhc2 () dcrocker net>
> Date: December 18, 2009 3:52:37 PM EST
> To: dave () farber net
> Cc: ip <ip () v2 listbox com>, Anthony Citrano <a () citrano com>
> Subject: Re: [IP]  spam insanity
> Reply-To: dcrocker () bbiw net

> > From: Anthony Citrano<a () citrano com>
> ...
> > Based on my experience this week, a single email failing Yahoo's  
> "bad
> > word" scanner will forever banish you
> ...
> > Some friends suggested that since I have my own domain and server, I
> > should enable DomainKeys and Sender Policy Framework on it.
> > ...
> > My research brought me to the people in charge of "helping" in such
> > situations: the Bulk Mail people on the Abuse team.
>
>
>
> Anthony,
>
> This actually has nothing to do with Yahoo, specifically, and  
> everything to do with an email infrastructure that is 90-98% spam,  
> and the massive base of well-organized, adversarial email  
> originators that send the junk.
>
> In reality, all of the very large ISPs devote extensive resources to  
> properly distinguishing spam from legitimate mail and the statistics  
> of their work are quite good.  (Their filtering services are nothing  
> as simple as just checking for problematic words; these are AI  
> engines.)
>
> Unfortunately, "quite good" is a long way from "perfect", but this  
> is the nature of the noisiness in the modern, open email  
> infrastructure:  Receivers must employ handling heuristics and  
> heuristics are never perfect.
>
> The advice to use DomainKeys -- which should have been for DKIM, not  
> Domainkeys, at this point; see dkim.org -- and SPF was misleading.   
> These mechanisms merely provide reliable and accurate identification  
> of a party associated with a message.  What they do not do is to  
> assess the quality of the entity associated with that identity.
>
> Further, yes, the current trust-based registration mechanisms, like  
> the one you dealt with at Yahoo, are oriented towards bulk senders,  
> not towards individual mailers.  Such are the economics of these  
> infrastructure services. Unfortunately, interacting with individual  
> users is not cost-effective.
>
> The current reality is that individuals who send mail need to send  
> from an ISP -- specifically, using a domain name and an IP Address  
> -- that has an excellent reputation and preferably one with its own  
> abuse handling services.  And sometimes, you have to encode your  
> message in a way that does not raise spam filtering flags.
>
> The use of authenticated identification, such as with DKIM, creates  
> an opportunity to do a trust-based overlay onto the existing email  
> service. This overlay can have very different handling rules than is  
> required for the current mistrust-based (risk-oriented) open service.
>
> Deployment of the authentication mechanism is proceeding well.  What  
> is still missing is a rich array of assessment (reputation)  
> mechanisms that use the authentication.  Some early assessment  
> services are starting to appear, but only starting.
>
> When an author and/or an operator are part of a trust overlay, the  
> receiver can handle messages with some forgiveness, rather than the  
> current level of Draconian disposition that is necessary.
>
> d/
>
> ps. full disclosure:  I administer the dkim.org web site.
>
> --
>
>  Dave Crocker
>  Brandenburg InternetWorking
>  bbiw.net



-------------------------------------------
Archives: https://www.listbox.com/member/archive/247/=now
RSS Feed: https://www.listbox.com/member/archive/rss/247/
Powered by Listbox: http://www.listbox.com