Re: Obsoleting a Recommendation, round four

On 2016-05-06 07:49, Stephen Zilles wrote:
>
>
>
>
> Sent via the Samsung Galaxy S7 edge, an AT&T 4G LTE smartphone
>
> -------- Original message --------
> From: David Singer <singer@apple.com>
> Date: 5/6/16 11:44 AM (GMT+03:30)
> To: public-w3process@w3.org
> Cc: Daniel Appelquist <appelquist@gmail.com>, Peter Linss 
> <peter.linss@hp.com>
> Subject: Obsoleting a Recommendation, round four
>
> [4] Added Wayne’s text to allow the AC to override the TAG (using the 
> same threshold, 5%, as an appeal) and appeal the Director’s decision.  
> Added to say that if there are obvious alternative technologies, they 
> should be documented in a note beside the Obsolete declaration in the 
> document itself. Added to say that anyone can notiuce the TAG has 
> timed out and that we move directly to ballot (i.e. it’s not a 
> specific person’s job to set a timer).
>
> I previously had  "If there was any dissent in Advisory Committee 
> review, the Advisory Committee may appeal the Director’s decision.”, 
> copied from somewhere else in the process. This is wrong.
> There is a perverse corner case (which I think affects at least one 
> other appeal); what if the AC votes, without dissent, “yes, obsolete 
> it!” and the Director decides “no”? No dissent, no appeal allowed. 
> Strange.  I removed the condition, it now reads "The Advisory 
> Committee may appeal the Director’s decision.”

I think that's what we should always do.  If everyone says yes and the 
director says yes, it's pretty unlikely that 5% would agree to appeal - 
so we don't need to specifically ban that.

The perverse case can happen if everyone says yes, someone makes a 
comment that is not a formal objection, and the Director changes what is 
to be done -- like alters a Charter - and then says yes to that version 
no one had seen.  That doesn't apply to obsoleting a spec but does to 
things like Charter approvals.  So, I think we should remove the 
restrictions on when appeals can happen everywhere.  But, that isn't 
this topic :)

>
>
> We still have pending how the TAG ‘announces to the other groups’ that 
> they are considering this.  AC (not technical).  In their agenda (who 
> reads it?). Chairs list (relies on the chairs forwarding to relevant 
> groups)?  I tend to think both of the last two: put it in their agenda 
> and notify the chairs.  Actually, thinking about it, maybe the TAG 
> agenda should always go to the chairs?
>
> SZ: I think requiring the chairs to read an agenda to find out 
> anything is a bad idea. There should be an announcement of the 
> "Proposal to Obsolete a Recommendation" that includes the REC name in 
> the subject line to the Public Announcement list that the AB had 
> created and to the chairs list. That allows the reader to quickly tell 
> whether to open the message or not.
>
>
> * * * *
>
> [3] Added SZ’s point, and Wayne’s about being clear about Appeals, and 
> added “if it exists” to the need to consult with the WG.  I think we 
> need to be clearer about announcing “to other W3C Groups” — like, 
> which ones? How? Maybe the Chairs list? The AC?
>
> * * * *
>
> [2] After offline discussion with some AB members, and the call today, 
> I offer the following.
>
> * * * *
>
> Accumulated text:
>
> 1) A new page, or section of a page, that defines what an Obsoleted 
> Recommendation is.
>
> An Obsoleted Recommendation is a Recommendation that the W3C 
> membership no longer actively recommends be implemented; however, its 
> formal status as a Recommendation (including its licensing status) 
> remains.  (This is in contrast to a Rescinded Recommendation.)
>
> A Recommendation may be considered obsolete if it is neither widely 
> implemented nor expected to be. It may represent a technical direction 
> that was not pursued further, or an architectural direction that is no 
> longer in alignment with best practices in the industry. There may be 
> alternative technologies better aligned with other parts of the Web 
> Platform, or more in line with best practices. There may be technical 
> drawbacks or even flaws associated with the Recommendation, but not so 
> serious as to cause it to be Rescinded.
>
> The W3C marks these as Obsolete to give guidance to the industry that 
> new implementation is not sought or expected.
>
>
> 2) A new section of the Process Document, 6.X (6.10 if existing 
> sections are not re-numbered, but it probably belongs before rescinded 
> in logical order).
>
> 6.X Obsoleting a Recommendation
>
> Anyone may request of the TAG that a Recommendation be considered for 
> Obsoletion. The request to the TAG MUST identify the Recommendation 
> and give reasons why it should be considered Obsolete; for example, 
> that the Recommendation has not been implemented, and no new 
> implementations are expected; that there are better alternative 
> specifications; that the Recommendation in question is not in 
> alignment with best design practices, and so on.
>
> The TAG MUST announce its intent to consider the Request to Obsolete 
> the Recommendation [[to other W3C groups]] and to the public and 
> SHOULD consult with any pertinent working groups, especially the 
> Working Group that developed the Recommendation, if it exists, or any 
> obvious successor WG. The TAG MUST make the decision to proceed, by 
> formal decision of the TAG.
>
> The proposal to obsolete a Recommendation follows the process for a 
> Proposed Edited Recommendation as defined in 6.7.2 and 6.5 for changes 
> to a Recommendation that are Editorial only. The proposal is placed 
> before the AC as a ballot if any of the following occur:
> a) The TAG decides to proceed;
> b) The TAG decides not to proceed, but 5% of the Advisory Committee 
> file a request for the ballot;
> c) The TAG makes no decision within 90 days of the receipt of the 
> request; anyone (e.g. the initiator) may then request that the AC 
> ballot proceed.
>
> If there is dissent in the Advisory Committee (votes against, or 
> formal objections) the usual process to find consensus will be 
> followed. Objections SHOULD include evidence that the proposal is 
> flawed; for example, that the Recommendation is widely implemented, or 
> it is reasonably expected that it will soon be widely implemented.
>
> Considering the advice of the Advisory Committee, the Director 
> approves or denies the decision to obsolete. The Advisory Committee 
> may appeal the Director’s decision.
>
> An obsoleted Recommendation is marked as such (a) in the document 
> itself and (b) on the TR page. The status ‘Obsoleted’ links to a 
> standing page which explains the meaning of the term. If there are 
> obvious alternative technologies, they should be documented in a note 
> beside the Obsolete declaration in the document itself.

looks good

>
>
> David Singer
> Manager, Software Standards, Apple Inc.
>
>