Re: Re-visiting how to charter WGs

This is an interesting topic and worth discussing by the Process CG.  We should start by being clear on what problem we’re trying to solve.

However, in practice the staff writes the charters and decides what can
be proposed to the AC. This is a problem if the staff is unqualified or
has an ulterior motive.

Hmm.  The staff works in the name of the Director, whose motives are presumed to be
Lead the Web to its Full Potential.  If the staff is out of sync with the Director (the literal Sir Tim
or the consensus of the management group to whom most decisions are delegated) that’s
an operational problem for Jeff and Ralph, not a process issue.

If the Director/staff is unqualified to make a decision in a particular area, arguably W3C has
no business trying to write standards in that area.  Maybe it should recruit staff, members, invited experts,
etc. to build expertise,  but it shouldn’t charter a WG until it does.

Here’s the problem I see: team resources and W3C’s credibility are finite, and spreading them too thinly over WGs that are unlikely to create Recommendations that get used in real products by real users is a poor use of those resources and undermines W3C’s credibility.   Historically we’ve started WGs when the team/Director/AC identified a CHALLENGE for which a web standard would be desirable.  Now we’ve invented Community Groups and the WICG incubator to discuss problems, incubate solutions, and a mechanism to take them to standardization once a SOLUTION has some maturity and needs broader discussion.

Explicit criteria to guide the team in deciding when to propose a charter, and for the AC to consider when evaluating a charter proposal might be modeled on https://wicg.github.io/admin/intent-to-migrate.html.  These are for a spec to migrate into an existing WG, but creating a new WG should have AT LEAST as high a bar.
- There should be an unofficial draft in some detail specifying the API/format/protocol to be standardized
- There should be evidence of a community behind it, including interest from those who would need to implement and deploy the spec
- There needs to be a clear value proposition for why a new standard is needed, and how it would improve upon the currently available options
- “Data What data do you have available that indicates that this enhancement will affect many users of the Web. Quantify the fraction of websites that are currently using something similar to this feature. Or, if a new feature, characterize the reason that you expect this to be far reaching.”

For those who have member access, I recommend https://lists.w3.org/Archives/Member/w3c-ac-forum/2015JulSep/0185.html as a useful review of a current charter proposal that lays out some additional criteria for what should (and should not) be in a good charter proposal.

On Aug 19, 2015, at 3:39 PM, Harry Halpin <hhalpin@w3.org<mailto:hhalpin@w3.org>> wrote:



On 08/19/2015 05:59 PM, Wendy Seltzer wrote:
On 08/19/2015 05:33 PM, Harry Halpin wrote:
Currently, charters for new WGs are semi-shrouded in mystery. While the
AC is notified when chartering begins, there is no way for the community
itself to ask for a WG to be chartered without going through the W3C Staff.
In practice, anyone can share a draft charter or ideas for a charter
with team and, if they ask an AC rep, with the Advisory Committee. I'd
also welcome better ways for us to recognize whether these ideas,
wherever they originate, are ripe for chartering standards-track work. I
have to raise questions with some of your implications here, however.

However, in practice the staff writes the charters and decides what can
be proposed to the AC. This is a problem if the staff is unqualified or
has an ulterior motive.

Thus, an objective set of rules, an open process from beginning to end
(such as requiring CGs to draft a charter and submit to the AC for vote
in terms of becoming a WG), and actual data-driven analysis would be
useful (probably one that should be done in terms of numbers of users,
since open-source implementations can be a dime a dozen).

I'd like to see this discussed. If the Credentials CG's attempt to begin
a WG is not suitable, I would suggest we use another example, perhaps a
WG that should never have been formed or another CG that wants to be a
WG. I'm all ears for good examples of both success and failure.


        cheers,
          harry


While sometimes this may be a good thing as the W3C staff successfully
charters WG in the best interest of the Web, but in some domains the W3C
staff is unqualified in terms of the modern Web (such as is often the
case in security, such as the demand for the Credentials CG to be a WG
[1]) or may have some other motivational structure for starting a new
WG:  For example, the current process allows W3C staff to run
'skunkworks' research projects as Working Groups and for WGs that
industry and users are not interested in (or even against) to be
chartered, but a small persistent group of hobbyists (that may include
W3C staff) are pushing for.

Currently we have set a higher-bar at AC voting - but would a new
transparent process help?

I'm not sure of the details, but it seems with the amount of activity in
CGs would provide empirical data, and there should be some objective
threshold involving commitment in terms of implementation and real users.

I would like to see this issue taken up by the CG and AB.

This ask by the Credentials CG to be a WG in this blog post [1] and
their analysis [2] is a pretty good test-case.
These posts were shared with us in draft form, and are not yet public if
you don't have the direct link.

In the case of the Credentials CG and related work, I believe we *are*
doing a data-driven analysis, investigating the needs asserted and the
existing technologies available, to ask where, if anywhere, the Web
needs new standards.

Thanks,
--Wendy

Without the W3C hat on, I
see a good case for standardizing vocabularies around health care or
education. I don't see much of a case here [3] for replicating the work
of OAuth, JOSE, FIDO, and then layering a somewhat incorrect mental
model of GPG with multi-origin key material (obviously a security and
privacy concern) on the top of the Web just because it uses RDF.

         yours,
               harry

[not yet public]
[3] https://na01.safelinks.protection.outlook.com/?url=http%3a%2f%2fopencreds.org%2fspecs%2fsource%2fidentity-credentials%2f&data=01%7c01%7cMichael.Champion%40microsoft.com%7c6d88b39dc96348bf9b2a08d2a8e7287f%7c72f988bf86f141af91ab2d7cd011db47%7c1&sdata=P3YbYpUHTyzw6%2bGf3LpP%2bwKeWH27Av4lyfNewe1PBVw%3d