Hi

On Wed, Nov 09, 2016 at 11:40:25PM +0100, Tobias Knecht wrote:
> On another note I find it slightly strange, that in almost every threat
> about abuse-c the topic of data accuracy is brought up, but policy
> proposals like the abuse-c for legacy space has been withdrawn due lack of
> consensus.

This is not a contradiction.

Forcing legacy holders to add "something" to the database is not magically
going to create "good quality data" for that something.

​I agree. 
The abuse-c was established to solve the challenge of cluttered places for abuse contact information. 
This btw is as well a data accuracy problem. Not knowing which of the fields to use and ending up with a wrong one.

The challenge still exists for legacy space.

 
As is the mandatory abuse-c today - it created "something" (so we can now
tout how wonderfully complete our database is), but given that it was
forced upon non-caring people, the *quality* of the recorded abuse-c:
values is not necessarily better than it would have been for "if you care,
please register an abuse-c:".

​I agree as well. 
And again, not having a standardized place to publish abuse contact data will make the data accuracy solution much harder.

​We will always have people that will not care and we can only make it as hard as possible for them to not care.
On the other hand, having a false or non existing abuse contact is already been used as a reputation metric.
Just the fact that it does not make sense for this community does not mean it does not make sense for any other communities. ​

For our data, the data quality is less good than before, as I find it far
too annoying to register abuse-c: for customer networks where the abuse
mails *could* be going directly (our parent abuse-c: points to our
abuse handling team, so mails are going to be handled, but might take
longer to reach the customer).

​I agree as well here. (3 out of 3 ;)​

​And yes abuse-c is not perfect and neither I nor anybody that was working on the proposal said it's gonna be perfect.
That's why we have this conversation now and that's why we will hopefully have a proposal that will fix some of the shortcomings. 
BUT it should under no circumstances contradict the ideas and solutions that have been ​achieved by it so far otherwise we will run in circles as we already do in some points.

​​​The same is the point for legacy space. abuse-c is not solving this issue, so lets solve it.

We might want to start solving things step by step instead of searching for the silver bullet and not getting anything done. 

For many PI holders I have seen auto-generated abuse-c: ("forced!"), which
bascically duplicates the normal contact info.  Yay.

​Before abuse-c was existing we saw 38% non deliverable reports in the RIPE region.
​Meanwhile we are down to 12% 
Yay.

​Thanks 

Tobias​