On Wed, 2004-07-07 at 10:58, Alan DeKok wrote:
Greg Connor <gconnor(_at_)nekodojo(_dot_)org> wrote:
Agreed. MARID-brand LMAP probably won't noticeably reduce spam, or even
forgery, at first, but if it creates a dis-incentive to forging MY domain
name, I will definitely use it.
Summarized in a carefully phrased sentence:
MARID creates a permanent way for domains to permanently give
dis-incentives for the transient problem of forgery.
Domain name forgery is a transient problem, as incidents can occur
sporadically, and the incidents are caused by different people at
different times. The only way to stop this series of problems is with
a solution which will permanently allow people to detect and prevent
these attacks. MARID is one approach which creates a long-lived
solution for a transient problem.
Could you define what you view to be forgery?
Similarly, bandages solve a transient problem, but the *existence*
of bandages is a permanent phenomenom. No one would claim that
bandages "do nothing to reduce injuries" because the problem solved by
bandages is transient in nature.
To put it yet another way, MARID can be thought of as a set of
distributed DNSBL's. Each domain operates its own DNSBL (or
whitelist), which use a well-known format. Peers on the net can look
up information for a domain via DNS, and choose to apply the DNSBL
information (or not).
Listing services are often more robust than a typical DNS server. The
nature of a listing service returns a single record in response to a
single query. Do you see this model being changed with Sender-ID?
None of this is new to the net. DNS already tells peers where to
find information (e.g. MX's), and DNSBL's are already used by clients
to obtain policies which they apply to data flows.
DNS routing information is normally obtained at a connection rate as are
queries to listing services. Isn't the information for Sender-ID
obtained at a much higher than these routing functions you compare it
to?
-Doug