W3C home > Mailing lists > Public > www-tag@w3.org > September 2006

Re: Approved TAG finding: Authoritative Metadata

From: Fernando Franco <avoid.spam.account@gmail.com>
Date: Wed, 6 Sep 2006 14:04:29 -0300
Message-ID: <002001c6d1d6$9bf06f70$c38c31c8@enterprise>
To: "W3C-TAG" <www-tag@w3.org>
Cc: "Tim Berners-Lee" <timbl@w3.org>

Hello, timbl.

First, thanks for the answer.
I need to mention, however, that it was not exactly what I expected.

For instance, you asumed that the issue was just if W3C should make soft (in
general), but...I also mentioned (perhaps I was not clear, my fault) the
possibility of making a "browser" (in particular).
While "to make a broiwser" falls within the general category of making soft,
I believe that there ARE very, very specific and very very important reasons
(even more so at this precise point in time) to consider this case in
particular, regardless of whatever is decided for the whole soft category.
I will not develop this here and now, however, cause I'm not quite sure if
such topic belongs here, or somewhere else, or if there is the slightest
interest, at least. I *can* do a strong (and perhaps with some novelty) case
for it, but I'll await counsel on apropiateness. Please advise.

Another problem is that you did not quite say yes or no. This is perhaps
understandable, since there might be different cases: is not the same to
make a browser, for example, that some other things. Perhaps not all moments
might be equally proper, as well.
Instead, you listed a sort of pro and cons list. Perhaps this was part an
invitation for further comments? Perhaps something like thinking aloud?
Perhaps education-via-rationale? Perhaps a mix? I'm not quite sure. In any
case, thanks again, I do value, both it and the effort.

Being not sure, I will however proceed to comment some of the points. If
busy or unwanted, please disregard.

------------------------------------------------

> Arguments against W3C doing open source:
>
> 1. Organizations should do done thing and do it well."  The Mozilla
> foundation, and OSAF, etc were set up as software foundations to make
> software.

I understand the logic of specialization, but...specialization itself is
both good and evil. For example...
When was the last time you (and most people here) discussed this stuff with
non-geeks? When the last time you teached html or css to a grandmother?
How many Grandmas and Grandpas will ever read (let alone code) the whole
acronym soup that W3C has generated these years (xml, xslt, xhtml, css, etc
etc ad nauseam)? Did we manage so that *at least* they could produce perfect
code in a WYSIWG manner? Nope, not even that (other than perhaps amaya,
which  I'm not sure either). I'd go as far as to suggest that perhaps it
can't even be done, as xhtml stands today.
Crazy, isn't it?

And if everybody specializes...who ensures that different organizations work
well together? Who checks for overall coherence? Case in point, Microsoft,
Internet Explorer, millions crying.


> 2. This would compete with our membership. Open source software can
> sometimes (not always) be seen as a threat to commercial companies.
> Also, of course, a w3C Open Source project would compete with other
> open source projects in the same area for mindshare, readership and
> the all-important developer pool.

Agreed. Hence my questions regarding W3C bussiness model in general, and as
applied to soft making in particular.
One weird idea, though...did your members ever had the option of not having
to code the central, spec-related parsers/libraries/engines themselves?
What would happen if you did offer them to do that, so that they could later
include it in full, final apps?
Would they provide programmers? Funding?
Does Microsoft really want to be mantaining an html/css rendering/parser
engine?
And even more...if there IS a standard (at all)...then what on earth for do
20 teams develop 20 engines? (which on top, have very little differences
that drive us all crazy already)
Wouldn't just one do, as long as everyone could use it? (or a couple, if one
needs for different ones for different scales, or anything irreducible)
Granted, there would be more trouible if we talked about a full, production
quality app (as oposed to libraries, engines, parsers, components). But
still...
I'm no expert, granted. But does any of this deserve to at least be
considered?
Don't many (corporative) bussiness models come from standardless times?

As for open source projects...if W3C soft is good, why would they do another
copy? Most would likely be very happy not to have to code their own. If they
still want, though...they can. They would not be worst than today.
It makes much more sense to not split developers and feedback among scores
of projects, though, sure. That, btw, happens even today, however foolish it
might be given than most programs of a given type end up being almost the
same. But if W3C was a player, I reckon most people would be glad to leave
it to the same people who created the specs themselves.

> 3. Within the W3C process, the one W3C selected open source project
> would be felt  have have unfair position if there were other projects
> with other ideas about how the standards should go.

If anyone can get it for free...then what would the problem exactly be? If
the functionality is covered completely and perfectly...then why should
there remain any competitor? They can go to the beach and thank you, in
fact. or better yet, they can help you.
(once again, we should consider as different the case for components than
for full apps, but...you likely get the idea)

> 4. The choices of languages to use for projects could be awkward, and
> whether to add support to a one particular platform or another, when
> there are competing open source projects which could be extended.

Ahh, the hands on details!
Granted, here there could be some troubles. The concept of "platform
independent programming" already exist, the situation is better (much) than
some years ago, but yes, there could still be problems.
Furthermore, there could be other problem which you, delicately, do not
mention: if things go wrong, W3C "prestige" could suffer. I'm honest: it
should be considered.
A side idea: do only the central, platform independent programming, and let
a community from each platform to adapt it for them (connect to native GUI,
etc). Is just fair, I think. Logical.
I have a couple more ideas, but this is already long.

> 5. The staff of W3C are stretched thinner and thinner as the number
> of groups and documents rises. To coordinate open source software
> takes time.

For starters, to *make* soft, is not the same as to *manage*, and is not the
same as to *coordinate*, and is not the same as to *supervise*. There can be
degrees, I mean.
But true, it means time and effort. And personnel is stretched, true.

My above idea however perhaps stands: would companies, even members, provide
programmers? Funding? If not...why not? Did they ever had even the chance?
Could arrangements be made?

[A digression: as you read me once, I'd rather have a true internet
governance body, international, paid for by everyone through national taxes.
If so, I see no reason why timbl could not move the whole W3C there. Direct
it, even. Just different funders: grandma, instead of corporations.
Financing soft should then be much easier.]

> Arguments for W3C doing open source:
>
> 1. Open source allows developers everywhere to get up to speed
> rapidly, so that once a standard has ben produced the industry and
> researchers can move on enhancements building on that standard

Perfect, particularly for components/libraries/parsers etc.

> 2. Coordination by W3C could provide continuity where individual
> champions from the volunteer community would hand over after leading
> a given project for a few years.

Since *you* make the specs, and *you* implement them, I doubt you'd be often
(or ever) tempted to use work done by others, though. I'm not talking tools,
such as programming languages, etc, of course, but code related to the task
at hand.
An exception could be, for instance, to use a good chunk of mozilla code, if
you wanted to do a browser, and accepted xul as platform. However, even
then...is tricky. Somebody else's code many times is very tricky.

> 3. There would be assurance that the software does indeed implement
> the standard.

This is the heart of the matter.
Zero doubt.
But furthermore, once more and more soft gets authored/supervised/whatever
by the same spec makers, the process to ensure (or at least to
evaluate/improve) overall coherence of whole systems improves much.
At the least, overall interspec coherence. At best, whole-picture/conceptual
systems coherence (for example, that the whole semantic web level works fine
once all togheter).

> W3C *does* try to make sure that for every specification there are
> typically several interoperable implementations, of which at least
> one is open source.

Exchange in channel #web in freenode, some months back:

<Dorward> Biblio: Today, the W3C require two independent implementations of
a standard before it becomes a recomendation. That proves it can be
implemented from the spec.
<JibberJim> no they don't!
<Dorward> JibberJim: I thought that was a requirement to go from candidate
recomendation to recomendation?
<JibberJim> somewhat less than that Dorward, the W3C only require 1
implementation of every feature, most groups up that to 2 implementations
<JibberJim> but it's still always of each feature, and not of the entire
spec
<Dorward> JibberJim: Eugh.

These people know more than me, for sure, but I can not put my hands in fire
for what they say. If they are right, however, we have an "overall
coherence" sort of problem again.
At any rate, timbl, I *do* know (and value) that W3C does a lot. And pretty
good, if you ask me.
I even dare say: "As good as possible within capitalism". The problem is
that I'm not capitalist.
My motto: "I *think* we can do still better".

> So, all that said, we do look at this issue every now and again.

Should I write about the particular case for making a "browser"? Here?
Somewhere else? Never?

Fernando Franco
Received on Wednesday, 6 September 2006 17:05:20 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Thursday, 26 April 2012 12:47:42 GMT