On Mar 7, 2009, at 6:19 PM, Suresh Ramasubramanian wrote:

> On Sun, Mar 8, 2009 at 7:47 AM, Hector Santos  
> <[email protected]> wrote:
>> Suresh Ramasubramanian wrote:
>>> Most of ADSP has been, so far, an attempt to introduce (sometimes  
>>> ridiculously) fine grained reputation scoring for vendors, and  
>>> clients of vendors.
>>
>> Can you explain where in the specification this is stated?
>
> Not the spec.  The majority of the use cases I have seen proposed  
> for it, on the other hand ..

ADSP's current definition of Author Signature is not compatible with  
what might become typical DKIM signing practices utilizing opaque i=  
values.  Requiring two signatures is a needless waste of resources.

DKIM i= values can help mitigate abuse when the number of problematic  
i= values is limited.  A limited number of problematic i= values  
should not be seen as ridiculous.  There is already a fair amount of  
DKIM replay abuse, where i= values could play a meaningful role.  An  
alternative strategy might attempt to limit DKIM domains to specific  
SMTP clients, but that would make the email less robust.

>> I hope the WG chairs will help keep the WG focus of the prize - an  
>> IETF standard policy layer/protocol for DKIM and not allow out of  
>> scope reputation ideas to ruin it once again as it did for SSP the  
>> past years.
>
> If you call them out of scope where that's going to be their single  
> largest intended application .. well, you are technically right I  
> guess.  Practically ... ?  That's another question.

The challenge is to discuss these issues, especially when everyone has  
a different opinion about what might be a practical mitigation strategy.

-Doug

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

Reply via email to