Re: The Community Group to Working Group transition, starting discussion

> On May 19, 2015, at 14:12 , Wayne Carr <wayne.carr@linux.intel.com> wrote:
> 
> 
> 
> On 2015-05-19 08:23, David Singer wrote:
>> One of the issues the AB and the Process CG wanted to look at was the question of what issues there are in taking a document that’s developed in a CG, and continuing work in a WG, presumably to put it on the recommendation track.
>> 
>> Here is a starter list of issues, and some possible responses.  Let’s develop the list, and the ideas, and then we can work out what we might need to do.
>> 
>> 
>> 1.  The CGs and WGs have different protocol for tracking patent commitments. If a CG document arrives in a WG, do we have sufficient tracking of contribution, and sufficient commitment of IPR, to minimize issues?
>> 
>> 2.  Do we have any issues if work was first considered in a WG, and then it was realized that it ought to be given to a CG to incubate?  (Includes the reverse of #1.)
>> 
>> 3.  There is great latitude in the CG process for how groups are run. Do we have a possible issue if a member feels that a CG document was developed in a way that was not ‘fair’ or respected due process?
>> 
>> 
>> Possible answers:
>> 
>> 1.  For those organizations who contributed and are also members of the WG, then the IPR commitment will be cleared on the first exclusion opportunity/commitment in the WG. For others, we need a clear patent commitment either by (a) the entanglement being clearly a result of their own Contribution to the CG, and hence covered under the CLA; or (b) the CG gets signatures from all who might have entangled IPR, on an FSA, before the document is transitioned.
>> 
> 
> The CLA has sections specifically about transition to a WG spec.  
> 
> The key thing is identifying contributors.  GitHub is often used in Community Groups.  However, only a single person is automatically listed in a pull request or commits, and that is not necessarily the contributor.  If there is more than one contributor, it can't automatically reflect more than one of them.  A simple solution is to require that pull requests and commits contain a list of contributors in the associated comment.  That could be harvested automatically to associate contributors to specs and more specifically to commits.
> 
> Here is the text that is in the optional CG Charter template: "For Contributions to Specifications, if someone other than the Contributor makes the pull request, the pull request should indicate who the request was made on behalf of and should provide a reference to the relevant archived GitHub Issue or group mail thread where practical. The information should be specific enough to identify the Contributor easily as well as a clear intention to Contribute to a particular Specification. In the commit, if practical, this information should be in the form {Contributors: ["@githubusername1","@githubusername2"]}."
> 
> That JSON format could be required for CGs that use GitHub and that would simplify tracking contributions.  It could also be used to ensure contributors are actually CG participants who have agreed to the CLA.  Otherwise, it can be anyone with a GitHub account who comments on a GitHub issue.  Something like this that depends on people making pull requests and commits isn't perfect, but it is far better than having to read everything in every Issue, meeting minutes, etc., which would not ever be routinely done.
> 
>> 
>> 2.  I can’t think of any issues right now.
>> 
> 
> Deciding work needs further incubation and trying to send it to a CG cannot happen now if the work has already started in a WG because W3C WG specs are under a license that does not allow modification outside the WG, while the CG license allows modification by anyone.  It is a one way path, from CG to WG with no possibility of return if an error is made and something moves to a WG too soon.  There is a mechanism for relicensing to allow a return to a CG, but that is only available if the specification is abandoned, not if it just needs further incubation.  Because of that, it should be possible for work coming from a CG to retain a permissive copyright license in the WG (e.g. use the W3C Software License or an updated version of that).

ah, true, the contributors would have to re-build in the CG, or we’d need to get an ‘abandoned spec.’ declaration, which is unlikely.

> 
>> 
>> 3.  In a sense, a CG input to a WG has no greater or lesser standing than a member submission. We don’t ask how a member submission was written (usually by one member).  Essentially we have a multiple-member member submission on our hands. The question of whether we think this is a good starting point seems decoupled from its provenance.  So, somewhat to my surprise, I think the formal answer is no.  However, we probably prefer that CGs not operate in ways that people think unfair, etc.
>> 
> 
> Community Groups are W3C creations, sponsored and provided for by W3C (so not the same as a Member submission).  

CGs are just a forum; we just provide the desk space.

> More importantly, there is a significant move towards requiring incubation in Community Groups before work moves to a Working Group.  Given that, the fact that Community Groups do not require fairness is unacceptable.  It should not be that a single company or some small set of companies can dominate the development of specs up to the point where implementation is underway and it moves to a WG.  It may be too late then, due to adoption, to make significant changes. If it is a random CG, they can choose whether or not to be fair.  

If we think that a submission from a CG does not represent a good starting point, is unfairly biased in some way, we don’t have to take it in the WG.  A member submission is a result of an extremely biased ‘process’ (they developed it by themselves) yet we are capable of subjecting it to scrutiny.

I agree that if we *recommend* a particular CG, then that CG should be suitable (in all respects including fairness).  But it’s not exclusive.

> 
> There is an optional charter template on W3C CG website that ensure fairness.  A number of important CGs use that, or earlier versions of it (Web Bluetooth CG, Web NFC CG, Auto CG, Trust and Permissions CG, Web of Things CG).  That charter allows CGs to operate any way they like, but has a fallback when there are objections.  That can mean votes or changing the Chairs.
> https://www.w3.org/community/council/wiki/Templates/CG_Charter
> 
> If a Community Group is created by W3C itself or by a W3C Working Group or Interest Group or W3C Staff member, it should be required to use a charter that ensures fairness. 
> 
> I've been creating variants of the Charter template for different types of Community Groups (e.g. whether they know their deliverables or not, whether they are independent or sponsored by one or more WGs). https://wcarr.github.io/cg-charter/index.html
> 
> 
>> 
>> 
>> David Singer
>> Manager, Software Standards, Apple Inc.
>> 
>> 
> 

David Singer
Manager, Software Standards, Apple Inc.