W3C home > Mailing lists > Public > public-json-ld-wg@w3.org > April 2021

JSON-LD 1.0 - "superseded" UI on site is overly harsh and pushy

From: Dan Brickley <danbri@google.com>
Date: Mon, 26 Apr 2021 19:40:15 +0100
Message-ID: <CAK-qy=7gfw3AqHn-jqvWYOYQJxCd4iJCdBhFX=yX4AJX18TZ2A@mail.gmail.com>
To: Gregg Kellogg <gregg@greggkellogg.com>, Ivan Herman <ivan@w3.org>, W3C JSON-LD Working Group <public-json-ld-wg@w3.org>, Ralph Swick <swick@w3.org>, Benjamin Young <byoung@bigbluehat.com>
Hi folks,

There is a problem with the citability of the JSON-LD 1.0 specs as
currently published at W3C.

In Google's original objection to the JSON-LD 1.1 charter (which I largely
wrote), we went to verbose pains to urge W3C not to undermine what it had
achieved with JSON-LD 1.0's adoption. I feel like that's happened anyway.
https://lists.w3.org/Archives/Public/public-new-work/2018May/0000.html

I have just gone to W3C to look something up to send to a colleague -
https://www.w3.org/TR/2014/REC-json-ld-20140116/#identifying-blank-nodes -
and this (screenshot below) is the user experience - it basically tells
people to use 1.1 instead. As you know 1.1 contains a bunch of new stuff,
and the strong push towards the 1.1 spec with the warning loses the
specificity of what is being pointed to. While I can understand the desire
to draw attention to the existence of related later specs, many readers are
going to be there to understand implementation constraints being documented
by others, so pointing at a literally different spec is not helping them,
W3C, or the implementor.

Context - this from Ivan,
https://lists.w3.org/Archives/Public/public-json-ld-wg/2020Nov/0000.html
-

> FYI: the old, 1.0 versions of JSON-LD 1.0 and JSON-LD API 1.0 are now
officially superseded. (This should have been done when we published the
1.1 version, but it fell between the cracks :-(

Currently - apart from the context-discovery mechanism using HTTP headers,
Schema.org's very substantive deployment across the web is based on JSON-LD
1.0, and getting to that stage was not easy. We need to simply and clearly
be able to cite the JSON-LD 1.0 specs without confusing implementers (data
consumers) and publishers by making it look like the approach is obsolete.
Right now, visitors to the those documents get upsold to go learn about a
different, later technology instead.

It would be misleading to publishers and to data consumers at this moment
to say that Schema.org uses or endorses JSON-LD 1.1, although of course the
project wishes the 1.1 endeavour every success. I haven't dug out the
specific assurances made around the launch of the 1.1 WG but I thought the
spirit of our understanding was a mutual "let's not undermine the success
we've only just achieved with 1.0". For example, it isn't even clear in
2021 which search engines understand JSON-LD 1.0 (or some subset) for
Schema.org use (as opposed to say Microdata in HTML5). I don't believe the
new additions in https://www.w3.org/TR/json-ld/#changes-from-10 has got
much attention yet, for the public web markup usecases focussed on at
Schema.org. I thought 1.1 was understood to be targeted at those usecases
for which new expressivity in JSON-LD's context mechanism was really
useful, and that those of us who were happy enough with 1.0 would be left
in peace to continue using it.

I appreciate that the "processing mode" machinery allows for situations in
which solely 1.0 content can be reliably discovered and correctly
interpreted, ... but by making the 1.0 specs almost impossible to be
sensibly linked to, it becomes very hard to say "write your documents per
the 1.0 specs", and point to the stable W3C specs we're supposedly relying
on. When someone cites a 1.0 spec, having the spec urge the reader to learn
about 1.1 instead is both counter-productive and confusing. What's the
point of having a standard that can't be cited?

The world has larger problems right now, but this does feel to me like
undermining one of the bigger success stories in terms of a widely deployed
W3C RDF format.

According to https://www.w3.org/2020/Process-20200915/#rec-rescind

"6.2.12.3. Abandoning a W3C Recommendation"

... JSON-LD 1.0 is abandoned.

Should I feel like a schmuck for talking 10s of millions of sites into
using it, or happy because they're also in some tenuous sense equally users
of JSON-LD 1.1 too? Should we really be pointing webmasters and publishers
at the 1.1 spec when there was never any expectation it would target
Schema.org's usecases?

My main concern here is with the ability to document Schema.org by saying
"For use in search engines, Schema.org can be written in JSON-LD 1.0" and
having something reliable to point to. Maybe sometime it'll be possible to
say that about 1.1 too. Should we be looking at requesting the superseded
status to be restored, or could the popup warning shown on rescinded
specifications be made less pushy and upselly?

Thanks for any thoughts,

Dan





Attached screenshot:

ALT description: it shows JSON-1.0 spec, greyed out with a big red warning
saying "This version is outdated! For the latest version, please look at
https://www.w3.org/TR/json-jd/" (UI has a collapse button which seems to
hide the warning).

[image: Screen Shot 2021-04-26 at 6.26.42 PM.png]

Screen_Shot_2021-04-26_at_6.26.42_PM.png
(image/png attachment: Screen_Shot_2021-04-26_at_6.26.42_PM.png)

Received on Monday, 26 April 2021 18:42:08 UTC

This archive was generated by hypermail 2.4.0 : Monday, 26 April 2021 18:42:09 UTC