W3C home > Mailing lists > Public > www-archive@w3.org > November 2006

Re: Charters for review

From: Dan Connolly <connolly@w3.org>
Date: Wed, 22 Nov 2006 13:38:55 -0600
To: Dean Jackson <dino@w3.org>
Cc: Maciej Stachowiak <mjs@apple.com>, Ian Hickson <ian@hixie.ch>, Chris Lilley <chris@w3.org>, Hypertext CG <w3c-html-cg@w3.org>, Tim Berners-Lee <timbl@w3.org>, Steve Bratt <steve@w3.org>, www-archive@w3.org, "L. David Baron" <dbaron@dbaron.org>
Message-Id: <1164224335.3997.622.camel@dirk>

On Thu, 2006-11-23 at 02:49 +1100, Dean Jackson wrote:
> 
> On 22/11/2006, at 2:46 PM, Maciej Stachowiak wrote:
> 
> >> TIMETABLE
> >>
> >> The milestones in the charter are somewhat unrealistic. I would  
> >> suggest
> >> the following timetable would be far more likely, based on past  
> >> experience
> >> with HTML4 (which is still not fully implemented by any two UAs),  
> >> DOM, and
> >> CSS2.1 (which is the only large W3C specification to have attempted a
> >> serious disambiguation period):
> >
> > I agree the current timetable is overly agressive - it's hard to  
> > imagine reaching REC in 2009 without major process abuse. But on  
> > the other hand, plotting the future out to 2022 has little  
> > predictive or planning value. I would prefer if the charter simply  
> > did not include a milestone for REC. Let's leave five-year plans to  
> > the soviets.

I'm sympathetic to that position...

> >>
> >>     First Working Draft in October 2007.
> >>     Last Call Working Draft in October 2009.
> >>     Call for contributions for the test suite in 2011.
> >>     Candidate Recommendation in 2012.
> >>     First draft of test suite in 2012.
> >>     Second draft of test suite in 2015.
> >>     Final version of test suite in 2019.
> >>     Reissued Last Call Working Draft in 2020.
> >>     Proposed Recommendation in 2022.
> 
> My experience with documents on the W3C track tells me that the only  
> thing you can (fairly) accurately predict is when you'll *first*  
> enter Last Call. Everything past that is mostly out of the WG's control.

There's nothing magic about Last Call; if the WG isolates itself
from the community for 18 months and suddenly goes "tada! we
think we're done!" then they shouldn't be surprised to find that
they're not. On the other hand, if the WG achieves the sort
of seamless interface with the community that Ian advocates,
Last Call will be barely noticeable. Reality will most
likely be somewhere in between; people naturally clump
into groups of 3 and 10 and 100 and 1000 and so on;
see http://www.w3.org/DesignIssues/Fractal for details.

In general, it's easier to predict the near future than it is to predict
events far in the future. I like sort of logarithmic planning:
if you want to make an appointment with me today, we might
as well agree on the time to the resolution of a minute.
For next week, we we share a timezone, maybe we just
say "Tuesday afternoon". To plan a month in advance, "first week
of December" is often close enough. To plan a group milestone
three months in advance, I'll want a confidence interval of about
a month. To plan a year in advance, more like three months. I don't
like to plan more than 18 months in advance; I tend to want
to start something new after working on one thing for 18 months.

Meanwhile, when an engineer goes to his manager and
asks for travel budget to go to a W3C WG meeting, or just
asks for time to collaborate remotely without having to squeeze
it into skunkworks time, it's handy
to have some estimate of how long it will take to get to REC:
it's no fun to work hard on a spec for 9 months, get all
your requirements met, wander off to do something else,
and then find that the WG changes direction later and
doesn't meet your requirements after all.


> Would it be enough to only have a Last Call on the charter?
> 
> I liked David's suggestion that the group consider deliverables with  
> smaller increments, although I understand what Ian says about this  
> being difficult. It still might be possible to develop an HTML  
> specification with the minimum set of new features.
> 
> The development of the test suite is another matter, but I would  
> expect that the majority of it arrives after CR.

There lies madness.

Start with test cases and story telling* and let the spec(s) fall out
naturally between them.

*story telling, whether it be conference presentations and tutorials,
use cases, requirements, blog articles, whatever.


-- 
Dan Connolly, W3C http://www.w3.org/People/Connolly/
D3C2 887B 0F92 6005 C541  0875 0F91 96DE 6E52 C29E
Received on Wednesday, 22 November 2006 19:39:36 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Wednesday, 7 November 2012 14:18:00 GMT