Re: Single model or framework for reputation/trust

Ed Gerck <[email protected]> wrote:
> I'm not going to continue to feed the troll. No one is talking about
> solving human problems here. I am talking about email servers, SMTP,
> software.

  I'm sorry, Ed.  You were the one who brought the human factor into
reputation with your "farmer and chicken" story.  I showed that the
conclusions based on the story were wrong.  It's inappropriate for you
to now claim your intent was to talk about something else.  You're
trying to hide from inconvenient facts.

> Software has no psychological inertia.

  People maintaining and deploying software do.  In addition, software
has *more* intertia than people do, because software doesn't generally
upgrade itself.  So in the absence of active updates, knowledge of
software's past behavior means that you can nearly guarantee it's
future behavior.

>  The purpose of this list is to talk about automated processes to
> define reliance on automated processes. Not to yep about psychology
> or criminal profiling.

  My "yepping" was an attempt to discuss the topic you brought up:
human expectations of reputation.  Since you dislike that topic, I'll
move on to another one.

  As for automated processes, "reputation" in an on-line world is tied
to identity.  Identities can be:

  - known human administering a system
  - domain name
  - SSL certificate
  - IP address
  - known association with other on-line systems

  Reputation can be based on provable on-line behavior:

  - volume of non-spam messages sent (as determined by recipients)
  - volume of spam messages sent
  - existence (or not) of zombie machines in space
    associated with an identity
  - existence (or not) of viruses sent in space
    associated with an identity
  - administrator responsiveness to complaints
  - association with others of known reputation
  - accreditation by others of known reputation

  That evidence can be used by a recipient to get some kind of
confidence level for communications with someone else on the net.
That information won't be perfect, but it will be better than nothing,
and it will be strongly indicative of future behavior.

  Alan DeKok.