RE: "image analysis heuristics" (ISSUE-66)

(NOTE: I have re-added the public-html mailing list to this thread to keep
it "open" - part of the larger issue was the closed process that got us
where we are today. There are many lurkers out there who have concerns
over this topic and process as well, and I believe that they are entitled
to know the discussion first hand - if they are not interested they can
happily delete any message with this subject line)


Sam Ruby wrote:
>
> I do not support a "must read the working group mailing list" as a
> potential modification to the process document.  I actually do believe
> that Ian does read the HTML mailing list, and have seen considerable
> evidence to support that belief.

However, here, we have ample evidence that he did not:

# [02:59] <Hixie> anyone remember what the status is of the "image
analysis" thread is?
# [02:59] <Hixie> was there some conclusion about what i should change the
spec to?
# [03:00] <AryehGregor> I think the conclusion was you should make it
vaguer so it doesn't sound like anyone is doing actual image analysis in
deployed AT.
# [03:00] <AryehGregor> Like "UAs should think up clever stuff to do if
there's no alt text, but any author who relies on this will be held to
judgment and suffer in hell for all eternity".
# [03:00] <AryehGregor> (the latter part is non-normative)
# [03:01] <AryehGregor> (unless you think you can God to follow the spec)

# [03:01] <AryehGregor> s/God/get God/
# [03:04] <TabAtkins> AryehGregor's summary is accurate.
# [03:05] <Hixie> ok
# [03:05] <Hixie> done

(http://krijnhoetmer.nl/irc-logs/whatwg/20100205#l-152 - I believe this is
known as the smoking gun)

In a previous email
(http://lists.w3.org/Archives/Public/public-html/2010Feb/0131.html), I
documented with URLs the entire thread of 'consensus' quotes that was
building around this point.  Voices from the various 'camps' (and whether
you like it or not, that is a reality too) were in concurrence: (Maciej,
Lachlan, Tab) + (Shelley, Laura, Matt, myself) + (Chaals) - all were
essentially agreeing to strike the paragraph and point to UAAG.  However,
that's not what happened, instead Ian posed a question in a real-time fora
that few of us can participate in due to real-world work commitments, and
5 minutes later made the change. That's not listening to all voices,
that's not reading the list, that's being bloody-minded.  All because:

> I think we do a disservice to users by not mentioning explicit
techniques.
(Ian Hickson -
http://lists.w3.org/Archives/Public/public-html/2010Feb/0133.html)

As Shelley noted, if we are to *trust* that this really is being done in
an asynchronous fashion using the mailing list and the issue & bug
trackers (I was able to quickly extract those previous quotes from the
comprehensive list of emails tracked with/to Issue 66 -
http://www.w3.org/html/wg/tracker/issues/66), then changes like this must
simply stop, and it is the responsibility of the chairs to ensure that the
process that all have signed on for is followed - and here, clearly it has
not been done.

>
> Instead, I would prefer that things that people wish to have tracked be
> recorded as bugs.  And, yes, that even includes tracking of changes to
> the Decision Policy itself.
>
> And I don't believe that any of the above is in any way inconsistent
> with the idea of augmenting bugzilla with a mailing list.

Mailing list, sure. IRC chit-chat that took under 5 minutes and resulted
in a unilateral decision? Stop! That's not a process, that is anarchy and
completely inconsistent with the spirit of cooperation and equal voices
being heard. Where were Matt, Lachlan, Maciej, Shelley, Laura, Chaals and
myself - all participants with a vested and committed interest in this
discussion - when the decision to make a change was taken? Not in IRC as
far as I know...

>
> Disagreed, though it is true that we have not reached amicable
> consensus yet.

Sam, you say that, but can you point to anyone (outside of Ian) who was
vehemently opposed (or even mildly opposed) to the emergent thought that
striking the paragraph from the spec and point to UAAG was the best way
forward? URLs or references to emails would be most valuable here please
(and, by the way, further underscores why IRC chat alone is insufficient -
unless we need to start auto-tracking IRC discussions in the Issue Tracker
as well - if that can even be achieved)

>
> Ian made a provisional edit in an
> attempt to reach amicable consensus, and clearly and unambiguously,
> that
> particular edit utterly and completely failed to achieve the goal of
> achieving amicable resolution.  Nothing more.  Nothing less.

No, I am sorry - it is much more. It once again illustrates Ian's contempt
for the consensus process, as well as his and other's contempt for the
concerns of a community:

# [03:00] <AryehGregor> Like "UAs should think up clever stuff to do if
there's no alt text, but any author who relies on this will be held to
judgment and suffer in hell for all eternity".
# [03:00] <AryehGregor> (the latter part is non-normative)
# [03:01] <AryehGregor> (unless you think you can God to follow the spec)

# [03:01] <AryehGregor> s/God/get God/
# [03:04] <TabAtkins> AryehGregor's summary is accurate.

Both Aryeh and Tab are entitled to be jerks in private, but making changes
to the specification based upon back-room banter is unacceptable (as well,
here Tab appears to contradict himself from his previous public statement
that removing the reference was acceptable:
http://lists.w3.org/Archives/Public/public-html/2010Jan/1203.html)

>
> It is equally true that, at least as of the present point in time, the
> change proposal that Matt has edited has also utterly and completely
> failed to achieve the goal of achieving amicable resolution.

I beg your pardon? It seemed to all involved in the email thread that
amicable resolution was indeed emerging. Where exactly are these
discussions supposed to take place? If Ian had issues then, he should have
jumped in (you too BTW), else we must TRUST that those who have a horse in
any given race have had an opportunity to speak and be heard. This is not
the case here - the discussion appears to have ended in IRC, between 3
people almost 2 weeks after the active thread seemed to be in a state of
agreement. The final decision was not made based upon weighing the various
points and counter-points that were discussed in the thread, but rather
based upon a smart-ass comment in IRC.

I ask you to point to any documented response that counters the emergent
consensus that I outlined in my previous email.

> This is a
> true statement, but in no way implies that Matt did not follow the
> procedures both in letter and spirit, or that he at any time acted in
> bad faith.

Again, please document the 'truth' in this assertion. Nothing that I read
in the bugzilla or issue tracker (nor the thread in the W3C email
archives) leads me to believe Matt failed in moving this issue towards
conclusion - he in fact modified his position slightly throughout the
discussion - or that he was failing in reaching 'amicable resolution' - it
is only when we get to the backroom antics of the IRC channel that
dissention and derision emerges and the entire process goes off the rails.

>
> Nor, by Matt producing a document which contains the text "Editor: Matt
> May (mattmay@adobe.com)", has Matt made any decisions, unilateral or
> otherwise, on behalf of the Working Group.
>
> All that we have at the present point in time is that text appears in a
> document that Ian is editing that Matt (and others) disagree with, and
> text appears in a document that Matt is editing that Ian (and others
> disagree with).
>
> Something needs to change.

Yes! The 'something' is the IRC channel as a decision tool. IRC is useful
for immediate discussion, but it should be removed as a vehicle for making
contentions decisions to the specification language - which is what
happened here. Had Ian truly read the emails (as he claims he does)
instead of tossing off a comment in IRC and then making an editorial
change none of this acrimony would have emerged. If you, Paul, Maciej and
Ian cannot see this then we have an even larger problem.

Since we are currently making 'provisional' edits to this part of the
specification, I request at this time that the chairs *directly* instruct
the editor to modify the text to more closely reflect the majority
position of those who have commented to date on this topic, which is, by
my count:

 - Strike the text completely and point to UAAG: Maciej, Lachlan, Tab,
Shelley, Laura, Matt, Chaals, myself

 - Leave references to heuristics (OCR) and techniques inside the
specification: Ian, Sam(?), Krzysztof Maczyński

If we require more discussion here to reach consensus, then let's have
that discussion in the forum that the process requires - the open,
asynchronous email list, and not the semi-exclusive IRC channel that most
cannot participate in (due to work commitments, etc.)

Anything else simply favors the current status quo, and continues to
foster the impression that there is an inequity at play here.

JF

Received on Saturday, 6 February 2010 19:40:35 UTC