RE: Followup to "Supergroups" message to AC Forum



> -----Original Message-----
> From: Florian Rivoal [mailto:florian@rivoal.net]
> Sent: Monday, June 20, 2016 8:03 PM
> To: Michael Champion <Michael.Champion@microsoft.com>
> Cc: daniel.glazman@disruptive-innovations.com; Carr, Wayne
> <wayne.carr@intel.com>; Revising W3C Process Community Group <public-
> w3process@w3.org>
> Subject: Re: Followup to "Supergroups" message to AC Forum
> 
> 
> > On Jun 21, 2016, at 06:27, Michael Champion
> <Michael.Champion@microsoft.com> wrote:
> >
> >> Again, there's the theory and there's the WG practice. The practice
> >> is that WGs work around the blockers in the Process, on Membership's
> >> firm request.
> >
> > As others have pointed out, the “blockers in the process” are there to make
> sure that the WG is operating within the boundaries of its charter, which is
> necessary under the patent policy.
> >
> > As I see it, we’ve already worked around the problem by creating the
> Community Group IPR policy which allows people to collaborate on new things
> so long as they commit to RF licensing any patents that cover their own
> contributions to a discussion. So the way forward isn’t to change the Process or
> Patent Policy to match the practice of mixing standardization and incubation in a
> WG, but to change the practice so that WGs have one or more affiliated CGs
> where incubation takes place under an IPR policy designed for that purpose,
> then specs can move to a WG where broad patent commitments are made for
> work that is in the scope of its charter.  That’s what Web Platform / WICG have
> done.
> 
> Let's see if we can bring CGs into this conversation without going off the rails :)
> 
> I agree that the CG's IPR policy is better than the kind of limbo that not properly
> chartered EDs live in.
> 
> However, I think there remains very strong doubts about whether CGs
> themselves are a good solution to the problem we're discussing now.
> 
> Before I dive into why I think that's a bad idea in this situation (because that's
> somewhat long), here's an alternative proposal:
> 
> Explicitly charter WGs so that they can take on **as ED** any topic that falls
> within their scope without rechartering, and apply CG-like IPR rules to such
> drafts. Recharter (with a deliverable addition) only when you want to publish
> FPWD, and switch to the old fashioned WG IPR rules at that point. This isn't that
> far off what we do in practice already, but would get rid of the gray area for
> not-yet-chartered EDs.
> 
> I think incubation CGs are a good fairly good way to start working on something
> for which no existing WG would be a good home. They are also a good place to
> work on things that members of a WG have rejected when you want to try and
> prove them wrong.
> 
> However, I think that pushing work out of WGs into GCs when the membership
> of the WG is not a priori opposed to that topic (as evaluated by the usual
> informal consensus process rather than by rechartering) is risky.
> 
>  * There's no guarantee that the membership of the WG and that
>    of the CG will stay synchronized, and given the different
>    conditions for becoming a member of either, they are actually
>    quite likely to diverge. If a WG and its CG do have
>    significantly different memberships, moving a spec from the
>    CG to the WG when several (or all) of its key contributors
>    or editors are not part of the WG is going to be a pain. It
>    is not clear that there will be much motivation for the CG
>    to hand off its work early to the WG when clear signs of
>    traction appear.
> 
>    - If the WG takes it over early anyway, this introduces a
>      significant risk of fork, and history has shown we're not
>      particularly good a dealing with forks with TR track on
>      the one side and live standards on the other.
> 
>    - If the WG waits too long, there won't be much else
>      to do other than dotting Is, crossing Ts, and rubber
>      stamping. This would still be useful from an IPR point
>      of view, but it is not clear how many people would be
>      motivated to work in such a WG, and even if that would
>      work out, we'd have thrown out the consensus approach to
>      spec building by moving most of it out of the WG.
> 
>  * If the WG and its CG do have largely the same membership,
>    then the argument that moving exploratory work out to CGs
>    saves the membership time and resources is void. People
>    who do not care about a particular topic which is discussed
>    in a group they are part of can ignore it just as well in
>    a WG as in a CG. Having started the draft in a CG
>    actually creates more administrative work: all the overhead
>    needed to publish the FPWD in the WG is still there, but
>    in addition, you now need to document that the spec fulfills
>    the the CG->WG graduation criteria. This may or may not be
>    a lot, but it is a non-zero addition. This additional hurdle,
>    even if small, also increases the risk that CG specs do not
>    graduate into WGs early and spend a large part of their life
>    outside of the consensus process, leave the WG mostly rubber
>    stamping, or never make it to TR at all.
> 
> And if we alleviate all that by making sure that incubating CGs have the same
> membership as WGs, the same consensus process, the same editors, and a zero-
> effort transition between CG and WG, then why have a CG in the first place?
> We'd be better off applying CG-like IPR rules in the WGs to EDs that do not yet
> have a FPWD.

I agree with Florian's analysis of starting new work in a WG. If this new work does not fit within the charter of the WG, then it does not come with a Patent Policy defined IPR commitment. That should be clear. That should not, however, keep the work from being documented. I think his suggestion to establish IPR commitments that are the same as the best CG commitments for this new work is a good one. This allows suitable incubation within the WG under basically the same conditions that a CG would provide.  It the incubation (within the WG) has been successful by the time of the next charter, then it should  be included in that charter as a work item and, if the charter is approved with that inclusion, a First Public Working Draft should be published to put the work under the Patent Policy.

It is, of course, up to the WG to decide whether to do the incubation within itself or to request those who want to do the work to create a separate CG. Depending on the proposed work it may be better to do it within the WG or in a separate CG. Work that needs to fit with the other work of the WG is best done within the WG. Work that is pretty independent of the WG is best done in a separate CG. If there is sufficient support in the W3C for the work, whether or not it is best done in the WG, then the WG cannot prevent the work from moving forward in a separate CG. 

Steve Z
> 
>  - Florian

Received on Tuesday, 21 June 2016 05:18:11 UTC