Hi Leo On Mon, 20 Feb 2023 at 16:37, Leo Vegoda <leo@vegoda.org> wrote:
On Mon, 20 Feb 2023 at 07:25, denis walker <ripedenis@gmail.com> wrote:
On Mon, 20 Feb 2023 at 15:58, Leo Vegoda <leo@vegoda.org> wrote:
[...]
Setting semantics aside... I don't know whether changing definitions — and adding a missing definition is a de facto change — would improve things or make them worse. What research do we have to support the position that it would be an improvement?
An interesting question. But take a step back and consider the current situation. We have an element of mandatory data with no definition. This data has been entered into millions of INET(6)NUM objects but it has no meaning. No one can reliably use this information for anything. And yet we know many people DO interpret this data and assume a meaning for it without any justification. So where we are now is not a good place.
Let me make a quick modified suggestion. We make "country:" an optional attribute in INET(6)NUM objects. The RIPE NCC removes this attribute from ALL existing objects. Consider it as a reset, removing undefined data. We give this attribute a definition related to geolocation. Anyone who chooses to add this attribute with the new definition may do so. Now we have meaningful information that anyone can use in a reliable way.
The RIPE NCC has been publishing country-related data and refining draft text about its unreliable nature for more than 20 years. We know that users infer what they want to infer about what country-related data means. Can we make predictions about changes in user behaviour with any degree of accuracy? If not, are there ways we can influence user behaviour?
My concern is that lots of users will not be aware of the change, or not have the resources to adapt to it in a timely fashion. So, if any change is made, it ought to be accompanied by a change management plan. The alternative is guessing and hoping and I personally believe that is not the responsible path.
I see your point. We have many examples over the years of changes that have been made and some people haven't responded to those changes years later. From the Early Registration Transfer (ERX) project we did 20 years ago there are still INETNUM objects with the combined data from ARIN and RIPE and a comment asking resource holders to update the object. We have three options here. 1/ We never change anything because some people don't take notice. 2/ We do a 'best effort' campaign to let people know. This could involve an occasional/regular newsletter sent to resource holders with details of important changes to tools managed by the RIPE NCC. This could also be published on the website for consumers of the database information. Include a comment in query output referring to this changes page. Users of these tools have some personal responsibility to keep themselves informed of important changes. 3/ In this case we can simply deprecate the meaningless, undefined, useless and misleading "country:" attribute from INET(6)NUM objects. The fact that the RIPE NCC has spent 20 years telling people not to rely on this piece of information for anything, is a good clue that change is needed. If anyone thinks it could be useful for geolocation purposes to have a 'geo' country then we could consider adding a new optional attribute "geo-country:" later. cheers denis co-chair DB-WG
Kind regards,
Leo