Re: Email Web of Trust - some actual experience

John Levine <[email protected]> wrote:
> Russ Nelson has been experimenting with a web of trust.
> 
> See http://web-o-trust.org/

  From what I can tell, it's just a simple "I connect to you, you
connect to someone else" kind of database.

  Previous discussions on this list already rejected such a design,
due to some very simple arguments.  I already know that there is a
path from me to you.  Knowing the elements along that path doesn't add
any useful information.

> It doesn't work very well.

  Exactly.  This can be proven from some simple analysis of the
system.

> A scoring system like PGP uses might work better,

  Which is why we've been discussing scoring/weighting systems here.

>  but I'm sceptical that one could get a set of weightings consistent
> enough to be useful on a large scale.

  I agree.  That's incentive to design a system where the weightings
are simple to administer, and simple to verify by elements in the
network.

  In contrast, the accreditation systems being discussed on ASRG are
complex to administer, and difficult to automatically verify by
network elements.  The data which comes out of them is thus very
subjective.

  My goal is to work on systems which are emperically verifiable.  I
don't have time to work on political systems, because I just don't
believe that they will work.

  Alan DeKok.