Summary & Minutes of Monday Part 2 of Silver Virtual Meeting

== Summary ==

We opened with a discussion of the definition of normative and 
informative.  W3C does not appear to have an official definition, and 
different groups (including WCAG 2.0) have their own definition.  We 
decided that we have no special usage of normative and can just use a 
generic definition, so we moved into the session exercise.

For this exercise, we analyzed the various levels of the WCAG 3 
Information Architecture and captured the pros and cons of having that 
level be normative.  Rather than paste it in an email, see Normative 
Informative Pros and Cons 
<https://docs.google.com/document/d/1QF5Olq8880_OmP0Eyr7CmFti5O1VIFUXPhguA6DFzlE/>. 


== Minutes ==

The minutes are a continuation of the morning minutes, so you may have 
already read them when you were reading Monday Part 1.  The Part 2 
session begins with agenda item 3: Normative Informative Pros and Cons

https://www.w3.org/2020/03/09-silver-minutes.html#item03

=== Text of Minutes ===

Normative and Information Pros and Cons

    <Lauriat>
    [21]https://www.w3.org/WAI/GL/task-forces/silver/wiki/2020_Marc
    h_F2F_Meeting_at_CSUN

      [21] https://www.w3.org/WAI/GL/task-forces/silver/wiki/2020_March_F2F_Meeting_at_CSUN

    Lauriat: initially review homework from earlier session, but
    time wise, for minimum conforance, that will be tomorrow
    morning

    later tomorrow agenda will include ACT tests

    topic for today: normative vs informative

    before exercise, recap the definitions

    <Fazio> [22]https://www.w3.org/2013/09/normative-references

      [22] https://www.w3.org/2013/09/normative-references

    <jcraig> ETSI definitions: Normative references are necessary
    for the application of the standard in which they are mentioned
    (they shall be publicly available and in English). Informative
    references assist the user with regard to a particular subject
    area.

    <jeanne>
    [23]https://www.w3.org/TR/qaframe-spec/#norm-informative-gp

      [23] https://www.w3.org/TR/qaframe-spec/#norm-informative-gp

    <jcraig>
    [24]https://portal.etsi.org/Services/editHelp/To-help-you-in-yo
    ur-work/FAQs/Normative-informative-references

      [24] https://portal.etsi.org/Services/editHelp/To-help-you-in-your-work/FAQs/Normative-informative-references

    <alastairc> There is a definition in WCAG, but not sure if that
    will help

    js: Reminder from Tim Boland to Good Practice #2 --js: So, how
    to treat normative references is one thing, but we're hoping to
    find an actual definition of what constitutes normative

    <Chuck> alastair... do you have a link to the wcag def?

    <jcraig> Cerification Kit definition: Normative elements are
    those that are prescriptive, that is they are to be followed in
    order to comply with scheme requirements. Informative elements
    are those that are descriptive, that is they are designed to
    help the reader understand the concepts presented in the
    normative elements.

    <jcraig>
    [25]https://www.certificationkitbag.com/blog/2015/9/30/normativ
    e-vs-informative

      [25] https://www.certificationkitbag.com/blog/2015/9/30/normative-vs-informative

    <jcraig> RFC-2119

    <jeanne>
    [26]https://raw.githack.com/w3c/silver/ED-changes-25Feb-js/guid
    elines/explainers/visualContrast.html

      [26] https://raw.githack.com/w3c/silver/ED-changes-25Feb-js/guidelines/explainers/visualContrast.html


    <charleshall> this debate goes way back. 2002 conversation:
    [27]https://lists.w3.org/Archives/Public/spec-prod/2002JanMar/0
    011.html

      [27] https://lists.w3.org/Archives/Public/spec-prod/2002JanMar/0011.html

    <jeanne> This guideline section is required (normative).

    jeanne: links to visual contrast definition

    <jeanne> The following tabbed sections contain helpful
    information, but are not required (informative).

    <sajkaj> Janina notes RFC6919:
    [28]https://tools.ietf.org/html/rfc6919

      [28] https://tools.ietf.org/html/rfc6919

    jeanne: laughs at "must (but we know you won't)"

    Lauriat: queue

    Chuck: someone mentions silver was trying to get away from
    those defs

    <sajkaj> ca: There was mention of getting away from these, but
    there are arguments against doing that

    <KimD> Kind of plain-language-ish, not sure if this is a good
    source: In information technology standards, normative parts of
    a standard are those that specify what implementors should
    conform to and non-normative parts consist of examples,
    extended explanations, and other matter not dealing directly
    with the specifications.

    could list this as a con in pro/con discussion

    <KimD> [29]https://whatis.techtarget.com/definition/normative

      [29] https://whatis.techtarget.com/definition/normative

    jeanne: will discuss RFC-2119 later

    <charleshall> definition from UUAG:
    [30]https://www.w3.org/TR/UAAG10/glossary.html#def-normative

      [30] https://www.w3.org/TR/UAAG10/glossary.html#def-normative

    <charleshall> What is identified as "normative" is required for
    conformance (noting that one may conform in a variety of
    well-defined ways to this document). What is identified as
    "informative" (sometimes, "non-normative") is never required
    for conformance.

    <Zakim> bruce_bailey, you wanted to mention 508 does not use
    word "normative"

    <sajkaj> bb: Notes "normative" not used in U.S. Federal
    regulatory documents

    <sajkaj> bb: First heard the term in W3C

    bruce_bailey: never seen "normative" used in federal
    regulation, but it is referenced where referencing external
    docs

    <sajkaj> bb: Don't believe we need to define, because we're
    using a standard meaning of the term

    in standards, you don't need to define words in common usage. I
    don't know that we need to define it for Silver

    <sajkaj> sl: Reason is for us all to understand when we discuss
    what should be normative, and what not

    <Chuck> wasn't chuck, was someone else

    <Chuck> unless it's a different Charles.

    Lauriat: discussed here so that the participants agree on the
    vocabulary

    JF: I suggested 2119, but it's on the agenda later

    Lauriat: silver has layers of info

    <Lauriat>
    [31]https://docs.google.com/document/d/1QF5Olq8880_OmP0Eyr7CmFt
    i5O1VIFUXPhguA6DFzlE/edit#heading=h.jycm1yj4w2u2

      [31] https://docs.google.com/document/d/1QF5Olq8880_OmP0Eyr7CmFti5O1VIFUXPhguA6DFzlE/edit#heading=h.jycm1yj4w2u2

    need to list pros/cons of using the normative/informative
    definitions

    users needs for functional support, activity for planning,
    designing, developing

    tests: more concrete determination of whether you met the need

    methods: concrete methods of how to meet user needs

    and finally guidelines

    User needs: in order to best explain the need for guidance

    a non-exhaustive list of similar user needs, example: image
    description, used for alt with SR, or voice activation with
    Speech AT

    <jeanne>
    [32]https://raw.githack.com/w3c/silver/ED-changes-25Feb-js/guid
    elines/explainers/ClearWords.html

      [32] https://raw.githack.com/w3c/silver/ED-changes-25Feb-js/guidelines/explainers/ClearWords.html

    jeanne: user needs are on the get started page under how to

    charleshall: Shawn, in the outline, re: tasks: we haven't
    talked about test categories. we know there are some related to
    implementaiton

    also higher order test to ensure the user need was met

    do we need a section test category in outline?

    <Chuck> +1 to Lauriat

    Lauriat: I don't think so. Lower level tests build up to the
    higher level user need

    e.g. in the example I gave: the label solves multiple user
    needs (SR output and Speech input)

    Fazio: re: example: alt test may not solve the user need for
    visual alt text representation for cognitive

    <sajkaj> jc: Have an example, in Mac OS hover will announce any
    element on screen and changes the image when hovered for size,
    color, etc

    <sajkaj> jc: Called hover Text

    <sajkaj> ca: But there are many such situations, which is why
    user testing is currently recommended in 2.x

    <sajkaj> sl: Want to return to core question --- multiple
    levels of test? General, tech specific, etc

    Lauriat: back to agenda: should we have multiple levels of
    tests. we could talk through both.

    Charles Hall: I raised this a few weeks ago. whether the
    guidelines are the only normative part.

    or if tests needed to measure them are also normative

    Fazio: ex: normative for cognitive block in addition to the
    other formats: SR, Speech. etc

    <Zakim> mbgower, you wanted to say that tests are not normative
    in WCAG; they are associated with techniques

    Chuck: could separate by machine automatable and manual?

    MG: tests are non-normative at the moment

    <Fazio> Good point MG

    <JF> +1 to Mike: expected outcomes are what should be normative

    using test technique to show you've met a normative req

    Lauriat: I will separate into automation (machine automatible)
    tests and higher level tests

    we did not discuss normative, but should not represent these as
    an exhaustive list b/c we would alsways leave someone out. that
    would be a CON in terms of calling the techniques normative vs
    /informative

    JF: suggest calling it functional outcomes because it address
    all user needs collectively

    this would make additional techniques additive vs pitting one
    user against another

    Lauriat: concern with that is that it changes the format of
    what we've already done

    jeanne: recommend we address it when we talk about tests.
    (scribe help jeanne?)

    david-macdonald_: discussion in WCAG about what the user needs
    to experience or what they would need to do... we had trouble
    defining how the user would experience it.. that info is too
    broad to list.

    <JF> +1 to David

    <Zakim> sajkaj, you wanted to ask whether JF is arguing to make
    user needs / functional outcomes normative?

    Lauriat: the is the reason we can't call this a definitive
    list. these are examples of the needs we are trying to fulfill

    sajkaj: John, was the reason for your name change suggestion
    because you think this should be normative?

    JF: when we focus on "needs" we'll never be able to list all
    need.

    sajkaj: are what you are proposing as "functional outcomes"...
    would it be normative?

    JF: yes, because we need to know we've met the need of X user

    <jeanne> Meeting the needs is what we are trying to do, so
    meeting the need should be normative. More of a functional
    outcome.

    scribe missed some of JF explanation

    jeanne: I've tried to capture that in the document

    I will screen share

    jeanne: John's "functional outcome" is in between the user need
    and the method

    are we in agreement that user needs are non-normative?

    Lauriat: was not trying to gain consensus in this discussion...
    I wanted the group to list pros/cons so we gain a shared
    understanding

    for example, some of these becoming normative would make them
    more difficult to maintain, and therefore less achievable to
    complete/update wcag3

    <Lauriat>
    [33]https://docs.google.com/document/d/1QF5Olq8880_OmP0Eyr7CmFt
    i5O1VIFUXPhguA6DFzlE/edit#heading=h.jycm1yj4w2u2

      [33] https://docs.google.com/document/d/1QF5Olq8880_OmP0Eyr7CmFti5O1VIFUXPhguA6DFzlE/edit#heading=h.jycm1yj4w2u2

    sajkaj: you are performing an overview of our existing
    wireframe

    charleshall: John, would it be fair to say "user needs" cannot
    be made normative?

    JF: if we make user needs normative, it will always be
    incomplete

    <charleshall> ^that nic = Chuck

    jeanne: let's move on... captured.

    <JF> individual needs should not be normative, but the
    collection of needs - the functional outcome - should

    <charleshall> +1 to sentiment of JF

    <Fazio> SL "Roles" became "Planning Activities" because of size
    variations in companies not having certain roles

    <scribe> scribe: Fazio

    Call if generalized statement says "required" that would make
    it normative

    Chall not call

    Alastair: some value in normative statements in activities

    SL: what parts of activity section have positive aspect of
    being normative?

    Alastair: Tables aspect maybe. Varies depending on guideline

    Chuck: Alastairs observation is pro and con

    dog barks

    lol

    SL: likely different ways of doing activities and meeting user
    needs. Normative would make it more restrictive

    Chuck sees it as a con

    LG: user outcome is what's accomplished and only that should be
    normative

    SL: 2 types of tests: Tech specific and Task Specific.

    <Zakim> bruce_bailey, you wanted to ask about method vs test

    Bruce: Whats difference between tests and methods

    SL: Tests are what's needed Method what you do to pass test
    ... Methods ensure it's there

    D MacDonald: WCAG 1 was tech specific. Went Tech agnostic for
    WCAG 2 for longevity. With current 18 month cycle Tech agnostic
    is ideal. Brings normative tech specific more relevant

    Alastair: reluctant to pin low level test validations to
    normative, because fast iteration on informative docs not
    really on normative docs

    <Zakim> bruce_bailey, you wanted to ask about normative in
    context of test versus normative in context of standard

    Bruce: need to distinguish requirements from test method and
    conformance

    <Zakim> jcraig, you wanted to list some more cons of making
    technology-specific tests normative

    <jcraig> cons: maintenance (in review and process), slow to
    update, and ARIA AAM examples

    J Craig: Maintenance, timelines for iteration are tedious. Adds
    a lot of time, responsibility to make normative docs. Can't
    update easily

    SL: Silver will have Widder scope of tech and guidance, Poll
    requests, file bugs, process will be extremely helpful

    <Zakim> MichaelC, you wanted to summarize current W3C thinking
    on living documents

    LG: Non-Normative more searchable, more understandable

    <MichaelC> [34]https://www.w3.org/wiki/Evergreen_Standards

      [34] https://www.w3.org/wiki/Evergreen_Standards

    <MichaelC> [35]https://www.w3.org/wiki/Process2020

      [35] https://www.w3.org/wiki/Process2020

    J Craig: Non-normative docs are more flexible like continuous
    delivery

    SL: Higher level task tests - some may be tech specific. Mostly
    around interactions or platform

    <jcraig> I was not suggesting user tests be "Living standard"
    track. I used AAM as an example to talk about the downside of
    having frequently changed/updated documents on Rec Track or
    other normative track. For the record, I think the
    technology-specific techniques should be non-normative
    examples. Otherwise, I believe technique contributions will be
    fewer and slower.

    <Zakim> alastairc, you wanted to ask if there were would be
    some tech specific needs?

    SL: Alastair: Specific tech that's is particularly strong or
    weak may be included or excluded easier if normative tests were
    implemented

    <Zakim> bruce_bailey, you wanted to say FPC statements might be
    evergreen

    SL: could have Bruce: functional criteria can be written in an
    evergreen way

    <alastairc> Note that I'm not in favour of this level being
    normative, had to stretch for a 'pro'

    <Lauriat> We'll get to "Cons" as well, no worries.

    LG: People might opt out by saying these tasks don't apply to
    our tech

    <alastairc> I think most of the cons from the last one apply?

    JS: normative tests might restrict innovation

    LG JF: functional outcomes should be normative

    <alastairc> in some cases user need = functional outcome...

    C Hall: Higher tests should include cognitive walk throughs for
    entire tasks

    LG: Con If it's normative it's not cumulative

    D MacDonald: Guidelines can provide catch all accessibility
    supported method/technique/way of achieving it

    D MacDonald Guidelines are so general that it might be better
    to make the methods normative

    <Zakim> alastairc, you wanted to ask if we are considering it a
    continuum? E.g. Guideline > Statement of Requirement > Method >
    Test, with pyramid of content.

    <Zakim> bruce_bailey, you wanted to say that normative methods
    (or maybe normative test) covers for current Silver GL being
    too open to interpretation

    Alastair: pull apart testable statements and functional
    outcomes maybe

    <alastairc> (Self scribing) Normative does not mean testable
    statement, we could have another concept / statement /
    paragraph below guideline but above method.

    <jcraig>
    [36]https://docs.google.com/document/d/1QF5Olq8880_OmP0Eyr7CmFt
    i5O1VIFUXPhguA6DFzlE/edit#heading=h.jycm1yj4w2u2

      [36] https://docs.google.com/document/d/1QF5Olq8880_OmP0Eyr7CmFti5O1VIFUXPhguA6DFzlE/edit#heading=h.jycm1yj4w2u2


    [End of minutes]
      __________________________________________________________