[gnso-rds-pdp-wg] authoritative

[Gomes, Chuck]

In preparation for the WG meeting this coming Tuesday, I ask all members to read David's message below and any messages on the list in response to it.  We will try to reach at least a tentative conclusion on this in that meeting.

Realizing that I temporarily halted discussion of issues on the list on Friday, because this is really a definition of terminology and not really an issue of requirements, if anyone has a suggestion for an alternative term instead of 'authoritative', it would be okay to share it on the list.  Thanks to those of you who already responded.

Chuck

-----Original Message-----
From: gnso-rds-pdp-wg-bounces at icann.org [mailto:gnso-rds-pdp-wg-bounces at icann.org] On Behalf Of David Cake
Sent: Thursday, April 27, 2017 1:12 PM
To: RDS PDP WG <gnso-rds-pdp-wg at icann.org>
Subject: [EXTERNAL] [gnso-rds-pdp-wg] authoritative

        I want to just write an (overdue) email recording what I verbalising reported to the meeting earlier this week.
The small group looking at the term ‘authoritative’ and its definition concluded that the term was best avoided.
Essentially, we identified three ways in which the term was in use, and concluded that the potential for confusion was problematic.
        One sense in which it was used was in a technical sense within the DNS. This sense is defined in RFC 7719 (which Andrew Sullivan was a co-author of), and as the DNS and RDS are closely linked systems this presents the potential for confusion. The definition is, however, of no use to the RDS, as it is highly specific to the DNS.
        The second sense is in a more general data-theoretic use. This is conceptually related to the DNS use, though there are significant complications. This sense is the sense in which it was earlier used in discussion of RDS requirements - the concept that within a data system there is a source of the data that is considered to have precedence over other sources of data. For example, in many data systems (including the DNS) data may be cached, and that cached data is not considered authoritative, even though it is a copy of authoritative data, and if that non-authoritative data is ever found to be different to the authoritative data, the authoritative data is assumed to be correct.
        The third sense is a legal sense, that is data that is guaranteed to be correct by some authority. This sense of authoritative was found to be confusing, and had been found to be problematic in the past, including within ICANN. Michael Palage discussed examples where this had happened, and made a strong case that if the term authoritative was to be used, it would have needed to be appropriately qualified.
        With significant potential for confusion, it seemed as if avoiding the specific term, but finding a term that captured the meaning of the data-theoretic sense.A forms of words suggested included "Data from the source repository,”. I felt that while that form of words looked appropriate for our purposes, having untangled the definitional issues that were the focus of the small group, it was appropriate to move discussion back to the main group for further refining of terminology in context.
         It should be noted that previous discussion involving the term in Copenhagen did not intend to preclude the use of non-authoritative data (caching and similar issues were regarded as implementation issues for Phase 2, that should not be precluded by requirements in Phase 1).

        David
_______________________________________________
gnso-rds-pdp-wg mailing list
gnso-rds-pdp-wg at icann.org
https://mm.icann.org/mailman/listinfo/gnso-rds-pdp-wg