On Oct 22, 2013, at 9:33 AM, Phillip Hallam-Baker <[email protected]> wrote:

> Trust is not transitive but cost measures are pretty stable
> 
> Producing a web of trust with a thousand bogus entries costs essentially 
> nothing
> 
> Producing such a web with twenty links to verified parties is much
> more expensive
> 
> Incidentally, I can't claim any originality for the extended work
> factors I use as it happens. Although I came to them independently it
> turns out that a co collaborator had gone so far as to organize an
> event on the topic,
> 
> But what I had not appreciated earlier is just what a difference it
> makes to cast the web of trust problem in terms of a work factor
> metric and introduce a combination of peer and ttp trust providers.

Dear Phillip,

Some hope to establish defenses for services that lack mandatory 
authentication.  Without authentication, any possible reaction would be based 
on unconfirmed suspicions as the only rational response.  Lack of 
authentication is often justified as a means to provide anonymity.  While 
enabling individual anonymity is fine, those managing a system that may 
initiate abuse must be authenticated and held accountable (to be responsive to 
reports of abuse).  Only those managing a system should be expected to 
attribute individual abuse based on internal accounts.  As such, this 
management can be done with anonymity as well.

Developing a group form of reputation for unauthenticated services as a means 
to avoid liabilities for errors made in identifying suspected abusive actors 
assumes dilution of these errors is a solution.  This approach can not be fair 
and represents a dangerous easily poisoned system.

Regards,
Douglas Otis






_______________________________________________
perpass mailing list
[email protected]
https://www.ietf.org/mailman/listinfo/perpass

Reply via email to