ietf-asrg
[Top] [All Lists]

RE: [Asrg] How to defeat spam that uses encryption?

2003-03-31 09:51:49

If I were a spammer (I am not) I'd start encrypting messages 
to throw off
content filtering. Public keys are easily obtainable and are readily
associated with good email addresses.

We have not seen this to date but it is certainly a possibility,
one that cannot be fully addressed within the straghtjacket
of pretending SMTP is an end-to-end architecture.

The purpose of S/MIME encryption is to provide strong end-to-end
encryption. There are actually a large number of customers who
can't use that and so it is not unusual to see configurations
where a mail proxy either has the private decryption key or 
checks to see whether all encrypted mail is encrypted under an
escrowed key.

Stock brokers cannot by law use end-to-end encryption for reasons
that become clear during the Enron and dotCom bubble market
investigations. In the military any use of encryption has to
be controlled. You don't want people sending classified info
to the enemy.

These are edge cases however, in many cases you really do want 
true end-to-end. For example in a lawyers or doctor's office.

A non edge case is a device that does not have end to end
capability like my Rim pager. Here the problem is the 
combination of constrained bandwidth and processing power. 
On the other hand the protocols employed in this case are
highly customized and don't require standards.


There are two issues here:

1) The need to block DoS levels of spam at the perimeter

2) The need to use client side information to block spam


What is needed is a richer communication channel between
the MUA and MTA. In particular the client needs to be able to
send feedback to the mail server 'this is spam' or 'this is
needed'. Then the mail gateway can respond differently to
floods of mail from the same IP which in some cases may be
legitimate and in others may not.

For example, Congress changes it's mail server IP address.
Suddenly the Whitehouse gets a deluge of email from a new 
source. Ham or Spam?

This is a very tricky one because once that pornography has
got onto the mail server it is part of the record and deleting
it is a criminal offense that can lead to a congressional
witch hunt.


Also, but not quite as complex:
A few years ago, I 'invented*' a way to encode a whole web 
page as a long
string. This string is a variable in JavaScript. 

No active code of any sort should ever be run unless it is
from an authenticated source. HTTP pull constitutes a weak form
of authentication, SMTP push does not.

In practice though these schemes leave a signal that is far
more distinct than the appearance of certain words, in the 
same way that F-R-E-E is far more likely to indicate a spam
than free.

Just a thought of what's to come...

That sort of thing is already happening.

                Phill
_______________________________________________
Asrg mailing list
Asrg(_at_)ietf(_dot_)org
https://www1.ietf.org/mailman/listinfo/asrg