[GNSO-Accuracy-ST] Level Setting

Mon Mar 7 13:54:36 UTC 2022

Michael,

Thanks for your thoughtful email.  IMO, the touchstone question is whether
the data elements serve the *needs* of the parties who receive the data.
This statement includes some important implications.

   1. The focus of attention is on the parties who get the data.  Not the
   registrant, not the registrar, not the registry and not ICANN.  The
   registrant, registrar, registry and ICANN also have interests in this
   process, but if the needs of the parties receiving the data aren't met, the
   rest are irrelevant.

   2. My phrase "parties who receive the data" includes recognition there
   may be conditions controlling whether a party does or does not receive the
   data.  The question of whether a party requesting the data is allowed to
   receive the data, i.e., the access question, is separate.

   3. My choice of the word "needs" is related to what the receiving party
   does with the data.  This is related to but distinct from "purposes" which
   has become a reserved word in our setting.  The list of purposes in the
   EPDP reports is a rough attempt to capture the same notion but forecloses
   any future discussions regarding whether the overall system is actually
   serving the needs of the community.

I have always understood the work of an "accuracy scoping team" is to lay
the foundation for future policy development work.  Accordingly, the
definition of accuracy must include one or more dimensions of variability,
with future policies setting the required levels of accuracy in each
dimension for the various circumstances.  Two of the obvious dimensions are
(1) the degree of certainty that the data is correct when it is supplied
and (2) how recently it has been checked.  The certainty dimension is

0 = accept whatever the registrant supplies

1 = check that the registrant's input fits syntactically for the data
element

2 = check that the registrant's input works operationally

3 = check that the registrant's input is indeed correctly associated with
the registrant

These degrees of certainty apply to *each* of the data elements provided by
the registrant.  For example, it is entirely plausible for a policy to
require level 2 or 3 for the email address provided by the registrant but
perhaps to permit level 0 for the registrant's name or organization.

With respect to recency, possible values are

   - checked when the domain was registered
   - checked annually
   - etc.

Much of what I've said here does not fit cleanly into the binary choice
you've presented, but I'm clearly much closer to the "degree of
correctness" than the other choice.  The other choice will lead to burying
all of the distinctions and shortchange any discussion of the needs of the
receiving parties.

Thanks,

Steve

On Sun, Mar 6, 2022 at 2:32 PM Michael Palage <michael at palage.com> wrote:

> Hello All,
>
>
>
> I am looking forward to a productive ICANN73 public session tomorrow.
>
>
>
> I spent the past several days trying to digest all of the exchanges that
> took place last Thursday. While I think we are close to wrapping up our
> work on Assignments 1 & 2, I think it would be constructive to quickly
> level set and make sure we are all on the same page to minimize potential
> future confusion.
>
>
>
> Part of my level setting involved going back to the original GNSO
> Council’s charge to the Scoping Team which asked is there “an agreed
> definition of registration data accuracy and, if not, consider what working
> definitions should be used in the context of the Scoping Team's
> deliberations.” See
> https://community.icann.org/display/AST/2.+Council+Instructions+to+Scoping+Team
>
>
>
> This task at first blush seems simple enough, but as we have learned there
> have been several concerns raised in connection with the use of the term
> “definition” and the meaning of “accuracy.” Therefore, instead of using the
> term “definition” as proposed by the GNSO Council I propose that we use the
> phrase “current contractual requirements and enforcement construct.” I
> believe this should meet the concerns of the RrSG that have repeatedly
> raised concerns about “providing a definition” and the concerns of the GAC
> and others about how a definition might bias future discussions.
>
>
>
> Is there any objection to us using the phrase “current contractual
> requirements and enforcement construct?”  If so please explain your
> objection and proposed alternative suggestion.
>
>
>
> Next we need to tackle what I have deemed the accuracy conundrum. The
> intervention by Stephanie this past week reminded me of some previous
> research that I was doing which I decided to revisit. I think Stephanie hit
> the nail on the head when she talked about how “accuracy” to most people
> conveys a binary choice, e.g. the data is accurate or is the data
> inaccurate.  It is a black or white answer with no room for grey. In fact
> this seemed to align closely with the RrSG proposed “current contractual
> requirements and enforcement construct.” If the data collected meets
> syntactical validation and either the email or phone number was
> operationally verified, then the data provided by the Registrant was
> “accurate” per their interpretation of the 2013 RAA.
>
>
>
> So I decided to spend a couple of hours researching the definition and
> origins of the word “accuracy” online and with an old school trip to the
> local library. I believe this definition of the word “accuracy” best
> describes the conundrum that we as a group find ourselves.
>
>
>
> noun, plural
>
> 1.           the condition or quality of being true, correct, or exact;
> freedom from error or defect; precision or exactness; correctness.
>
> 2.           Chemistry, Physics. the extent to which a given measurement
> agrees with the standard value for that measurement. Compare precision
> (def. 6).
>
> 3.           Mathematics. the degree of correctness of a quantity,
> expression, etc. Compare precision (def. 5).
>
>
>
> Source Dictionary.com
>
>
>
> Now the first definition “being true, correct, or exact; freedom from
> error or defect” is a rather high bar, particularly if you are applying
> this bar to all registration data elements processed like some working
> group members have advocated. However, that bar is substantially lower if
> free from defect simply means that the data collected by the Registrar was
> syntactically correct and a Registrar at a point in time got an affirmative
> response from either telephone number or an email.
>
>
>
> Alternatively, the third definition of a “degree of correctness” suggests
> something other than a binary accurate or inaccurate response.  Therefore
> to help steer our future discussions I would like everyone to answer the
> following question:
>
>
>
> Question #1
>
>
>
> For purposes of our Working Group the term accuracy should be defined as:
>
>
>
> [  ] true, correct and free from error; or
>
>
>
> [  ] degree of correctness;
>
>
>
> (PICK ONE)
>
>
>
> I think once we get clarity and/or agreement on these points, we should
> have a more clearly defined path forward for our post ICANN73 call.
>
>
>
> Best regards,
>
>
>
> Michael
>
>
>
>
>
>
>
>
>
> _______________________________________________
> GNSO-Accuracy-ST mailing list
> GNSO-Accuracy-ST at icann.org
> https://mm.icann.org/mailman/listinfo/gnso-accuracy-st
>
> _______________________________________________
> By submitting your personal data, you consent to the processing of your
> personal data for purposes of subscribing to this mailing list accordance
> with the ICANN Privacy Policy (https://www.icann.org/privacy/policy) and
> the website Terms of Service (https://www.icann.org/privacy/tos). You can
> visit the Mailman link above to change your membership status or
> configuration, including unsubscribing, setting digest-style delivery or
> disabling delivery altogether (e.g., for a vacation), and so on.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://mm.icann.org/pipermail/gnso-accuracy-st/attachments/20220307/21703494/attachment.html>