From: dlv@bwalk.dm.com (Dr.Dimitri Vulis KOTM)
Date: Sat, 14 Sep 1996 14:08:46 +0800
To: cypherpunks@toad.com
Subject: Re: Internet Drivers' Licenses
In-Reply-To: <Pine.SUN.3.94.960913202014.2637A-100000@polaris>
Message-ID: <VZ28TD8w165w@bwalk.dm.com>
MIME-Version: 1.0
Content-Type: text/plain


Black Unicorn <unicorn@schloss.li> writes:
> Seriously, what is the import of the "real identifiable human" or the
> "posting under their legal names" point?

I'm sure certain parties out there would like that.

> If an AI program posts quality stuff, what's the difference?

Indeed, an AI program I wrote has been posting excellent stuff in
alt.sci.physics.plutonium - check it out! :-)

> Why the import of true "legal" names?  Why not simply develop reputation
> signatures?

Right now I let all my incoming e-mail collect in one queue.
If I were really bothered by junk e-mail, I'd set up some sort of
filtering that would sort them into three classes by originator:

to be deleted without reading
to be read as soon as possible
to be read at my leisure

the default, for unknown originators, would be to be read at my leisure.

> What your suggestion basically says is "instead of developing our own
> decentralized reputations system for filtering lets use one already in
> place, i.e. the state Department of Motor Vehicles.

If the situation with junk e-mail becomes much worse than it is now,
then I think we'll end up with the following scenario:

1. A spammer gets my name, Igor Chudov's name, and a bunch of other names
from our Usenet postings.

2. The spammer e-mails each one of us, offering to buy X-rated videos.

3. Igor Chudov reads the spam e-mail first and somehow informs my mail-sorting
'bot that this e-mail should be junked.

4. If my 'bot sees the spammer's mail, it junks it.

And I'd do the same for him if I saw it first. :-) Naturally the warning about
junk e-mail needs to be digitally signed. I suppose they could be posted in a
specially designated Usenet newsgroup. The e-mail-sorting 'bot would check
this newsgroup for signed junk-mail notices from trusted parties and junk the
matching e-mails from the incoming queue. I guess it'd have to look at the
body of the mail and not just the headers, which are easy to vary.

This is the kind of project cypherpunks would do if they were writing code,
instead of lies and personal attacks, the way Tim May (fart) does.

> In any event, getting reputation credentials from a decentralized "web of
> trust" is a much more efficient answer, especially where you can assign
> your own levels of trust to each signator.

Yes - take a look at the NoCeM project for Usenet at http://www.cm.org.
Perhaps this technology can be adapted for rating e-mail.

---

Dr.Dimitri Vulis KOTM
Brighton Beach Boardwalk BBS, Forest Hills, N.Y.: +1-718-261-2013, 14.4Kbps