Re: Grounds for Formal Objections

I’ve been reminded that the TAG does indeed document some of the design principles it uses in spec evaluation, e.g.  https://w3ctag.github.io/design-principles/ , https://www.w3.org/2001/tag/doc/promises-guide.  I do think WG’s SHOULD respect the TAG’s judgment on such things and AC members should consider voting against advancing specs that don’t follow their guidance. I’m just not convinced we need the formal objection mechanism in the process to make this a MUST.

> as disagreeing with fantasai is highly correlated with being wrong :)

Indeed, but taking that risk:

> . A qualified Director, in consultation with the TAG and other
> experts, should be able to adjudicate such issues on technical grounds: it's not
> always and only about process.

Can anyone think of an example where this has happened in W3C?    What specific technical grounds could be invoked to over-ride  a WG’s “rough consensus and running code?”   I can think of *values*, especially around privacy rights, that I could imagine a particular WG coming down on one side of and the TAG/Director on the other.  Or I can imagine an accessibility or security concern that the WG believes is a corner case that would be too expensive (in WG time, implementation effort, overhead for users, disruption to the existing ecosystem, whatever) to mitigate but the TAG/Director disagree.  Are these the sort of “technical grounds” we’re talking of here?

It would be great if the TAG had maintained the Architecture of the Worldwide Web or some other fundamental principles document that we could use to evaluate proposed specs against, but it has chosen the (more pragmatic) approach of evaluating specs on a case by case basis.  Likewise, it would be great if W3C had a fully engaged Director who was an expert on current web technologies.  But without one or both, I’m not sure technical formal objections to spec features are worth the trouble they are sure to cause.

I think FOs do make sense if a WG hasn’t followed their defined process, or at the WG creation phase when there is a values-based decision to make about whether a proposed spec would improve the web platform or be worth the resource investment from the community….

> And if they brush [TAG]  feedback off (or feedback from other horizontal review groups), now there are grounds for procedural objections.
The way I see it, the scenario where a WG brushes off feedback from the TAG or some other horizontal group is why we have an AC ballot at the PR stage:  There is a judgment call to make here about the tradeoffs a WG has made between the various constraints it is trying to optimize for.   I don’t think there’s any getting around this being an ultimately humanistic rather than technical decision.

From: Florian Rivoal <florian@rivoal.net>
Date: Tuesday, March 19, 2019 at 2:50 AM
To: Michael Champion <Michael.Champion@microsoft.com>
Cc: "Siegman, Tzviya" <tsiegman@wiley.com>, Chris Wilson <cwilso@google.com>, fantasai <fantasai.lists@inkedblade.net>, public-w3process <public-w3process@w3.org>
Subject: Grounds for Formal Objections

This is veering off topic, so I'm changing the title, but I think it is an interesting topic.



On Mar 16, 2019, at 2:46, Michael Champion <Michael.Champion@microsoft.com<mailto:Michael.Champion@microsoft.com>> wrote:

 Likewise I think W3C could learn from the WHATWG’s extremely limited ability for contributors to file “formal objections” to the Steering Group if they think their expert opinion is being buried in groupthink.  (It exists on paper, but has never been used, and the SG cannot make a technical judgment but only a procedural decision that the Editor hasn’t respected the various WHATWG policies.)

In another mail, fantasai was arguing for the opposite, and supporting the existence of Formal Objections on non procedural grounds

Quoting here here for convenience:

I think there's also cases where the WG might be intentionally disregarding the FO
because its choice of design principles disagrees with the FO, and that's not an
issue of process but of technical architecture. In such cases a review of the issue
by a higher technical authority representing W3C's technical values and expertise as
a whole is warranted. A qualified Director, in consultation with the TAG and other
experts, should be able to adjudicate such issues on technical grounds: it's not
always and only about process.

I'm leaning towards agreeing with Mike rather than fantasai, but I'd like to talk through it some more, as disagreeing with fantasai is highly correlated with being wrong :)

Currently, the process allows for that sort of FO. Do they happen often (or at all)?

Could this be handled differently? For example, people can always ping the TAG (which includes the Director so long as we have one) on any issue they want feedback on, and we could include the TAG  in the mandatory horizontal reviews (which is de-facto the case already, given that intent-to-ship requires it). Then, if the TAG tells a WG they're going against some web-wide architectural principle, that doesn't give them the authority to order the group to change their spec in a particular way, but the group nonetheless needs to handle that feedback one way or another. And if they brush that feedback off (or feedback from other horizontal review groups), now there are grounds for procedural objections.

The advantages I see to that sort of approach is that it keeps the discussion and consensus wrangling closes to the group that owns the spec, does make sure that we can insist on web-wide higher principles, but doesn't invoke some higher authority who can disregard consensus (and who may not be as well versed in the details of the topic at hand).

—Florian