W3C home > Mailing lists > Public > public-new-work@w3.org > May 2018

Google's response to 'Call for Review: JSON-LD Working Group Charter' (fwd)

From: Coralie Mercier <coralie@w3.org>
Date: Tue, 1 May 2018 20:44:59 +0200
Message-Id: <74411186-4F48-47A3-96CE-E93F38186AD2@w3.org>
To: public-new-work@w3.org


> Begin forwarded message:
> 
> From: Wendy Seltzer <wseltzer@w3.org <mailto:wseltzer@w3.org>>
> Subject: Fwd: Google response to JSON-LD WG charter proposal
> Date: 1 May 2018 at 20:36:37 CEST
> To: Ivan Herman <ivan@w3.org <mailto:ivan@w3.org>>, Ralph Swick <swick@w3.org <mailto:swick@w3.org>>
> Cc: Comm <w3t-comm@w3.org <mailto:w3t-comm@w3.org>>, Coralie Mercier <coralie@w3.org <mailto:coralie@w3.org>>
> Message-Id: <75b6b993-ea0d-d84b-021f-99211db80137@w3.org <mailto:75b6b993-ea0d-d84b-021f-99211db80137@w3.org>>
> 
>  On 30 Apr 2018, at 19:45, T.V Raman <raman@google.com <mailto:raman@google.com>
>        <mailto:raman@google.com <mailto:raman@google.com>>> wrote:
> 
>        Thanks Dan!
> 
>        W3C Team, I submitted the review on Friday -- but parts of the
>        feedback didn't make it through -- (actually almost none of
>        it made it
>        through) Thanks again to Dan for catching this,  and I hope
>        you can
>        take our input into account.
> 
>        Dan Brickley writes:
>>        Rob, Ivan,
>> 
>>        It seems that our AC rep (TV Raman, Cc:'d) had problems with
>>        the W3C site
>>        and that our JSON-LD review comments (and formal objection) were
>>        inadvertently lost on Friday, despite it seeming that they'd
>>        gone through
>>        successfully. It is unfortunate that we had a weekend
>>        deadline as we have
>>        only just confirmed the situation today, after the deadline.
>> 
>>        TV Raman can confirm that the details below represent
>>        Google's position on
>>        the matter; I hope you can take them into account during your
>>        deliberations, e.g. by updating the form on our behalf if
>>        possible. The
>>        fact that (prior to the deadline) I communicated all the
>>        points below when
>>        Rob and I met recently, might also be relevant.
>> 
>>        Thanks,
>> 
>>        Dan
>> 
>>        ---------- Forwarded message ---------
>> 
>>        For the review form at
>>        https://www.w3.org/2002/09/wbs/33280/jsonld-charter-201803/ <https://www.w3.org/2002/09/wbs/33280/jsonld-charter-201803/> is:
>> 
>>        *1. A Formal Objection, i.e. we suggest changes to this
>>        Charter, and only
>>        supports the proposal if changes are adopted.*
>> 
>>        *2. Form field for "Make my responses": "Public and send
>>        email to
>>        w3c-ac-forum and public-new-work".*
>> 
>>        3. Comments for "4. Support for the Proposal" section of
>>        form are:
>> 
>>        "*We object specifically to the phrasing  "1.1 will
>>        supersede the current",
>>        and more generally to the general tone which suggests that
>>        1.0 (despite
>>        currently being used on millions of Web sites) will be
>>        replaced by a 1.1 or
>>        2.0. Further* considerations about direction that are not
>>        directly
>>        editorial are given below.
>> 
>>        4. For section 5.,  "Participation", it is likely that we
>>        will participate
>>        [checkbox yes]
>> 
>>        5. For section 6., "Support for Deliverables of the group",
>>        "intends to
>>        review drafts as they are published and send comments."
>> 
>>        6. Detailed comments for section 7.,  "Detailed Comments,
>>        Reasons, or
>>        Modifications".
>> 
>>        We are pleased to have a played a leading role in the
>>        success of JSON-LD
>>        1.0, and are supportive of W3C's proposal to continue
>>        maintaining and
>>        improving its formats for structured data representation.
>> 
>>        Unfortunately, the fairly shallow way in which JSON-LD has
>>        seen success for
>>        in-page markup across the Web pulls in a different direction
>>        to that
>>        proposed in the WG charter. The draft charter in its current
>>        form is
>>        problematic in several ways. Some of these can be fixed
>>        through editorial
>>        and scoping changes, but there are also  fundamental
>>        technology problems
>>        that echo difficulties from the XML era (GRDDL, XML Schema
>>        vs RDF Schema,
>>        etc.).
>> 
>>        The success of JSON-LD in the public Web dates from its
>>        adoption by
>>        Schema.org <http://schema.org/> <http://Schema.org <http://schema.org/>> and from implementations by
>>        search engines including Google.
>>        These implementations generally focus on a reliably fixed
>>        subset of
>>        possible JSON-LD, allowing consumption of certain expected
>>        data structures
>>        indicated by e.g. "@context": "https://schema.org <https://schema.org/>", without
>>        that context
>>        needing to be fetched/interpreted at runtime. No search
>>        engine (to our
>>        limited knowledge) fully implements the "context file"
>>        processing that
>>        fully general JSON-LD 1.0 implementations are expected to
>>        handle. We
>>        believe JSON-LD 1.0's processing API and Framing
>>        specifications are
>>        similarly under-implemented in a wide-scale Web search
>>        environment, raising
>>        questions about their moving to version 1.1 if the new
>>        standards work is
>>        justified using search-based deployment of the version 1.0
>>        data format.
>> 
>>        Search engines at all scales are still engaging with their
>>        JSON-LD 1.0
>>        implementations and the draft charter language (specifically
>>        that "1.1 will
>>        supersede the current") risks undermining implementation and
>>        deployment of
>>        the current approach. Browser implementations are almost
>>        non-existent,
>>        although we have some preliminary explorations around
>>        Chrome. Implementors
>>        working on 1.0 implementations  are likely to be tempted to
>>        stop work the
>>        moment that W3C announces a 1.1 WG;  this seems unfortunate,
>>        and unhealthy
>>        for the wider ecosystem.
>> 
>>        We urge a much stronger commitment to backwards
>>        compatibility, and an
>>        explicit phase of the WG's activity to engage with current
>>        and potential
>>        implementors beyond the core JSON-LD specification and tool
>>        development
>>        community. Some of this might be within a new WG, some
>>        within W3C's various
>>        relevant CGs.
>> 
>>        There are serious usability issues with JSON-LD 1.0 and
>>        W3C's commitment to
>>        improving the format is welcome. In our experience the
>>        biggest usability
>>        problem with JSON-LD stems from its combination of an
>>        RDF-based unordered
>>        graph data model with the JSON specs unreliability regarding
>>        repeated
>>        fields. While there may be potential for improving the use
>>        of JSON's array
>>        notation, any WG will need to balance the temptation to fix
>>        such things
>>        with the usability lessons from RDFa, where a number of
>>        clever notational
>>        tricks combined to create a standard whose rules were nearly
>>        impossible to
>>        remember. The best solution might turn out to be the
>>        existing design.
>> 
>>>        From personal communication with the charter authors we
>>>        understand that one
>>        usecase driving the proposal was a desire to allow syntactic
>>        schemas (e.g.
>>        JSON Schema) to work more effectively with JSON-LD data.
>>        This deserves
>>        exploration in the charter and WG, including an account of
>>        how it relates
>>        to mechanisms that operate over the RDF abstraction (e.g.
>>        Shex and SHACL).
>> 
>>        The charter should also give a more substantive treatment of
>>        JSON-LD's
>>        known privacy/security issues, namely that any consumption
>>        of general
>>        purpose JSON-LD data using RDF-oriented tooling typically
>>        requires fetching
>>        and interpreting so-called  "context files", e.g. revealing
>>        (to remote
>>        service or by network traffic) the broad topic of the
>>        content. While this
>>        is less of a concern for the simple schema.org <http://schema.org/>
>>        <http://schema.org <http://schema.org/>>-oriented use of JSON-LD in
>>        the public Web, the charter is broader in scope and the
>>        issue deserves
>>        attention at this early planning stage. It is possible that
>>        an approach
>>        that documented the popular technique of consuming only
>>        known, fixed
>>        contexts/schemas that don't need runtime fetching might both
>>        address some
>>        privacy issues, as well as fit better with the Web Search
>>        usage of JSON-LD
>>        that resulted in its widespread use.

--
Coralie Mercier  -  W3C Marketing & Communications -  https://www.w3.org <https://www.w3.org/>
mailto:coralie@w3.org <mailto:coralie@w3.org> +337 810 795 22 https://www.w3.org/People/CMercier/ <https://www.w3.org/People/CMercier/>







--
Coralie Mercier  -  W3C Marketing & Communications -  https://www.w3.org
mailto:coralie@w3.org +337 810 795 22 https://www.w3.org/People/CMercier/
Received on Tuesday, 1 May 2018 18:45:03 UTC

This archive was generated by hypermail 2.3.1 : Tuesday, 1 May 2018 18:45:03 UTC