On 26/Jan/11 18:01, John Leslie wrote:
Alessandro Vesely <vesely(_at_)tana(_dot_)it> wrote:
I propose to use the term "reputation" for the value upon which one
decides whether to send an abuse report to the relevant entity.
For many of us, that's the null set.
Alas.
(I propose that abuse reports are useful when sent to a reputation
service by a customer of that service, and leave others to argue
whether they have any value beyond that. Thus I'm going to hang onto
the name "reputation service" for any such service.)
That's a first step toward a constructive definition of "reputation".
Your latter proposal can be changed by blurring the "customer of"
membership, while enlarging a reputation service by adding more and
more customers. For example, a site might host several virtual
reputation services by assigning higher factors to reports sent by
customers of many of them. But then, are we going to also call
"reputation" the quality that one should have in order to become a
customer of those services?
Alternatively, we can imagine to send abuse reports to, say,
abuse(_at_)arin(_dot_)net, and have them forward them to the relevant ISP (or
RIR) who, in turn, will forward them to the relevant ADMD, and so on.
Then, we can imagine to optimize the resulting traffic by skipping
some of the initial hops and directly addressing a target with a
sufficiently high reputation.
Both approaches should result in the same thing: a global reputation
service.
Now, I'd be happy to use the term "vouching service" for something
determining whether abuse reports would be helpful.
This may be a useful distinction. I'd still consider that vouch=none
for a given domain should mean it has a pathologically low reputation.
_______________________________________________
Asrg mailing list
Asrg(_at_)irtf(_dot_)org
http://www.irtf.org/mailman/listinfo/asrg