"$MTP" and Extra Added Crypto(TM) (was Re: Throw away the Internet and start over?)

Jeff Bone jbone at deepfile.com
Thu Apr 24 11:03:28 PDT 2003

On Wednesday, Apr 23, 2003, at 13:34 US/Central, R. A. Hettinga wrote:

> At 10:54 AM -0700 4/23/03, James Rogers wrote:
>> Forcing them to run expensive compute operations against
>> their messages would slow them down, but not too much.
> Which is why I like charging them actual money. :-).

Putting prices (and the point here is that evidence of work constitutes 
a kind of price) on the low-level artifacts of communication is IMHO 
the wrong idea.  It's the kind of idea you'd expect from economists and 
the kind of folks that are obsessed with the fact that our money 
doesn't adhere to the gold standard anymore. ;-)

You can increase the hassle-factor all you want, the software isn't 
going to mind. :-)  If you increase the monetary cost of sending 
slightly, then you MIGHT make an impact --- but don't assume that 
simply increasing the resource cost increases the monetary cost.  (And 
FWIW, increasing the cost might not be enough --- does the cost of 
bulk-rate mail keep your snail mail from filling up with crap?  No, it 
doesn't.)  And there are lots of negative consequences that you can 
imagine, starting with the fact that the evidence-of-work idea is just 
generally a colossal overall waste of resources.  (And yeah, FWIW 
that's also an admission that Russell's earlier point re: RESTmail and 
the economics of spam is accurate.)

IMHO, the right approach is to come at this from the top-down;  manage 
the principals involved and the "web of interaction" between them.  
White lists, black lists, filtering, etc. are not only sufficient, 
they're probably necessary regardless of what other mechanisms you have 
in place.  (They just need to be made more effective and easier to use.)

Generally speaking, in a communication system, either people you don't 
know a priori can contact you, or they can't, or they can but they have 
to go through some sort of multi-step introduction / approval process 
(automated white-list building) or be referred (FOAF, web of trust 
mechanisms.)  If you don't model these types of social relationships 
and interactions, you can't manage them.  Period.  If you really want 
to add price tags, then stick them on one of those things:  make users 
pay to be added via an automated introduction / approval process if 
they fail certain other criteria.

NB, tangential, re: managing relationships.  One of the things I was 
particularly proud of in Ding! was the way we handled the buddy 
relationship.  Of course there was the request / approval cycle that is 
common to most buddy list systems;  but we also had a feature called 
"who's watching me" that allowed you to see who had you on your buddy 
list at any point in time.  From that list, you could remove people 
you'd previously allowed to see your presence information, merely 
temporarily disallow them to see / contact you, etc.


More information about the FoRK mailing list