> [mailto:[EMAIL PROTECTED] On Behalf Of Jim Fenton
> If we were to include this in the threat document, it would > need to go into a new category because it's not a threat to > the signature mechanism nor to SSP, but rather an attack on > DNS that might be facilitated by DKIM. I'm not sure whether > this is in-scope for the threat document or not, but it would > be an expansion of its current scope to include it. Actually this class of threat is the one class that I think the IESG needs to police above all. I am not very worried about bad crypto being a threat to the Internet, errors get fixed faster than people think. If we had deployed DNSSEC in 2001 and opt-in had turned out to be a mistake as some people's intestines were claimed to indicate it would be fixed by now (at least the Romans made an empirical examination of the entrails they were relying on for prognostication). I am worried about the risk of run amok protocols or protocols with large cascaded amplification effects. DKIM provides for a certain degree of amplification but only one stage is possible and at base there is not actually a major difference in cost between 100 byte packets and 500 byte packets. Switching is going to be the main cost at that packet size. _______________________________________________ NOTE WELL: This list operates according to http://mipassoc.org/dkim/ietf-list-rules.html
