ietf-asrg
[Top] [All Lists]

Re: [Asrg] Stamping

2003-03-21 05:28:22
Scott,

However, if you're looking to make spamming extremely tedious, time
consuming, and difficult for the sender, I think it's worth revisiting
that thread.

That consumes the only resource that is gaining in cost each year.

I agree that technical solutions that use more CPU, bandwidth and
storage than mail transport currently does are fine if they reduce
spam, and (therefore) also reduce human time resource
consumption. Ultimately if spam volumes are reduced, then there will
be some saving in CPU, bandwidth and storage resource which will
offset any increases due to new protocols. Besides that, as you've
pointed out, those resource keep getting cheaper, whereas human
resources are the opposite.

I don't like that, but how about some notion of stamping? Each message
recieved must include a valid stamp. Stamps are only usable once.

This is an interesting idea. I have some questions for you about your
proposal:

Can you please describe more fully your concept of a "partial hash
collision", and "sufficient number of bits". I think I understand that
the first 64 bits of the stamp is input to MD5, and also that the
recipient's email address is input to MD5. If we let "x" be the number
of bits in "sufficient number of bits", are you saying that:

 MD5(stamp[0:63])[0:x-1] = MD5(recipient email)[0:x-1]

(i.e. the left-hand most part of each MD5 hash match)

or perhaps:

 MD5(stamp[0:63])[128-x:127] = MD5(recipient email)[128-x:127]

(i.e. the right-hand most part of each MD5 hash match)

or perhaps that it is up to the receiver to determine which bits match
? And presumably "x" is worked out by the recipient when it does it's
calculation to produce some stamps ? Or are you suggesting that "x" be
a fixed quantity for this protocol ?

Forgive my ignorance, but I'm not familiar with working with partial
hashes, and I don't know of any gotchas. My main concerns would be:
say for a particular recipient email address, could the MD5 hash it
produces severely limit the maximum value of "x" that can be used for
a partial match against hashes of 64 bit quantities ? Given the good
randomness of MD5, I would guess probably not, but it is worth
investigating (if you haven't already). My reasoning is that if that
were the case, then some unfortunate recipients may be easy targets
for spamming because it would be easier to "guess" the first 64 bits
of their stamp.

Are you proposing that the stamp-control protocol piggy-back within
SMTP ? My main reason for asking is that for a sender to acquire a
stamp from a receiver, the request really needs to somehow end up at
the mail server for the receiver (so that the database of valid stamps
can be checked when a mail message containing a stamp is received). I
think it would make sense for it to piggy back within SMTP.

I am also assuming that a stamp cannot be verified as belonging to, or
not belonging to, a particular recipient, except by the recipient
itself. (I say this because one of the options is for the recipient to
create totally random stamps.) Is this correct ?

As an aside from these questions, I haven't found any holes,
yet... :-)

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