Sender pays requires a set of standards, cooperative implementation,
some non-trivial infrastructure to make it work, and a lot of
time and
energy. Whitelisting is something that can be built on an individual,
site or organizational level without a need of new standards, new
protocols, financial transactions, certification, PKI
infrastructures,
or anything else. It can be implemented today, in fact,
layering on top
of what's there in most cases.
Whitelisting/Authentication delivers an immediate early adopter value - it
reduces the chance that the company deploying the authentication will have
its emails listed as false positives, it reduces the chance the
authentication aware filter deployer will lose emails.
Clearly there is also a network effect, the more people use authentication
the more incentive there will be to use it as content filters can become
more aggressive. Consider the following scenario, 50% of the mail you
receive is authenticated (not difficult if it is all internal). That means
that you can tighten your acceptance criteria on your spam filters, probably
halving the residual amount of spam without the overall false positive rate
being higher than before.
However network effects bend both ways. Another name for network effect, or
'viral marketting', is chicken and egg problem. That is why it is critical
that there be a value to early adopters.
I don't see any such early deployment advantage to sender pays. It provides
zero or negative return until it is widely deployed. If I am the only person
who puts stamps on my email I will pay money out to do so and receive none
in return.
So it is in the self interest of every party to wait until the scheme is
fully deployed before participating. Furthermore there can be very little
confidence sender pays will ever get anywhere so there is even less
incentive. In particular I and many others simply don't believe that sender
pays will decrease spam if successfully deployed. So the deployment
incentive curve is negative.
Sender pays is an expression of an ideology and reason is not going to
provide any answer. Self interest represents an imminent critique in this
instance since the governing ideology is that of Ayn Rand. The sender pays
proposal will never be made in terms any more concrete than it is today. We
will never see an answer to the problem of how mailing lists are going to be
managed in the new free market, or how the charging infrastructure to
allocate costs will be managed.
The costs of infrastructure are easy to overlook. If you look at past
VeriSign press releases you will see that the hardware we bought from IBM to
suport the DNS clusters cost well over 8 figures. The number of lookups
would be much more than for DNS and if real money was involved the running
costs would be much, much more than for DNS, they would be much higher in
any case because each lookup would involve a state change. ATLAS would have
to be much more complex to manage state changes.
Basically any sender pays mechanism is going to cost a minimum of 0.01 cents
per message to operate, it is more likely the cost would be 0.1 cents even
for the most naive scheme. The cost of the infrastructure required to deploy
would be in the tens of millions for a commuinty of users large enough for
the stamps to provide any value in decreasing spam.
Phill
_______________________________________________
Asrg mailing list
Asrg(_at_)ietf(_dot_)org
https://www1.ietf.org/mailman/listinfo/asrg