[hashcash] let's see if this works
- From: "Eric S. Johansson" <esj@xxxxxxxxxx>
- To: hashcash@xxxxxxxxxxxxx
- Date: Sun, 16 Jul 2006 19:01:45 -0400
it has taken Comcast forever in a day to fix DNS. I think it's right
now but they didn't bother to tell me. They waited instead for me to
forget and then to remember again.
Anyway, hash cash content in case this makes it through.
At a long conversation with an ISP about using hash cash. They liked
what I did in camram with regards to reducing false positives. They
came up with a suggestion which is to not generate stamps at all on till
the other side signals that stamps would be useful.
I think this is a good idea and I will put it into my code base. The
signal I will use is a DNS text record as I proposed some moons ago.
But while is most disturbing in the conversation was the attitude that
the person wanted what camram provided as long as I took out anything
doing with sender pays. They refuse to deal with the fact that with
their suggestion and the camram message stamps policy, stand generation
would be exceedingly low even through moderate levels of adoption.
The primary concern was he didn't want, as an ISP, to generate any
stamps ever for anybody for any reason whatsoever. He would rather put
up with probabilistic filters tuned to his e-mail profile and apply that
to the entire user base (no, he didn't say that exactly, but he did
heavily imply it)
At the same time he rejected the concept that users could or should
generate their own stamps and also rejected that could understand what
stamp generation was all about.
If folks wouldn't mind, I would like to talk about this experience
answer strategies for how to deal with it. Maybe ISPs are not the
target market initially but instead, make tools like camram useful to
small and medium-size corporations? maybe we need to verify that hash
cash is still in some of the content filters listed on the web page.
And if so, try to get them to advertise their availability via DNS so
that sending systems could know when to generate stamps.
I still need to do the audio track to my slideshow that I did for BBN
and if I remember correctly someone asked what was my argument against
the "proof of work doesn't work" paper. (Hint, they did the wrong
economic analysis. It's not cost of generating stamps, its the cost of
not getting their advertising out there. By reducing spammer exposure,
you reduce the revenue proportionally to the reduction. So if you
reduce traffic by 10%, the revenue dropped by 10%. Also remember that
currently, spammers are getting by on the number of people without
filters plus whatever leakage they have through filters (estimated 10%
across-the-board).
by the way, advertising stamp awareness can be dangerous in the
beginning when there are only a few. This is why it would be critical
to have a "this is spam button" so we can track the sources and add them
to our blacklist. The tracking and blacklist should be filter
independent and it should be shareable. Which raises the question are
there any good models for sharing blacklist type information with
untrusted parties in the mix. Or is falling back on the brown list
"current stamp plus 3 bits" model sufficient?
---eric
Other related posts: