Subject: Re: Software architecture matters
From: <stephen@xemacs.org>
Date: Mon, 11 Sep 2006 12:03:03 +0900

Santiago Gala writes:

 > There is a severe breakage in the way email works:
 > 
 >    The amount of money that senders pay are not proportionate
 >    to the attention cost they make us receivers incur into.
 > 
 > I'm speaking of spam here, and I think this has been solved (for
 > the moment) much better in the XMPP/Jabber distribution model,
 > where there is *strong* identity and, thus, ways to get rid of 
 > freeloaders burdening the system as a whole.

That's not a solution to the problem of email-like channels.  When
people say that "spam is killing email as it killed Usenet before", I
don't think they mean for person-to-person communications.  Even
security by obscurity works fairly well for that, and if your known
correspondents can handle digital signatures, you can get rid of 100%
of spam, *as long as you are willing to define "spam" as "anything
without a known signature."*  If most people used email only in this
way, there would be very little spam, and nobody would care about it.

The problem faced by email is that many people *want* to receive
*some* unauthenticable messages because establishing identity in
advance is fairly expensive.  But as soon as you allow that, you're
exposed to spam, as "wiki spam" and "blog spam" show.

I'm not sure why digital signatures are so uncommon.  They would
basically eliminate false positives for known correspondents in spam
filtering, which would be a big plus.