Representativeness of implementors in HTML WG [Was: Use cases]

On Mon, Jan 3, 2011 at 3:53 AM, John Cowan <>
> Benjamin Hawkes-Lewis scripsit:
>> 2. The implementors involved produce the software employed by most
>> end-users and targeted by most web authors.
> This point, if actually provable, is convincing to me.

Well, I'll give it a go.

This seems like a tangent from the use case discussion, so I've started
a new thread.

The targets for most HTML authoring are email clients, web browsers, and
search engines. I'll also briefly discuss our representation of
authoring tools, conformance checkers, and web servers.

A couple caveats. This survey reflects my incomplete knowledge of our
444-strong membership and how I mentally map the landscape of HTML
consumers and producers:

Just because we have representation from an organization responsible for
implementing a given piece of software does not mean that representation
is particularly active, or (in the case of larger organizations)
guarantee that the people actually working on the software in question
are in the loop. I can say that we have very active input from people
responsible for standards support in popular web engines like Trident,
Gecko, WebKit, and Presto. Whether, say, the IIS, Outlook, and Visual
Web Developer teams are being kept up to date on developments and asked
for feedback by Microsoft I do not know.

It's also worth noting that the HTML WG and the WHATWG with whom it is
partnering to produce the HTML specification are open to new members,
conduct technical discussions in public, run an open bug tracker, and
follow discussion of the HTML drafts in the wider world of blogs,
forums, Twitter, mailing lists, and opinion. For example, Bruce Lawson
of Opera does a lot of outreach work teaching people HTML5. So the
inputs to the spec are by no means limited to a WG echo chamber.

+ Email clients

Campaign Monitor is a site devoted to HTML email marketing. They release
annual surveys of email client marketshare and CSS support.

Microsoft, Apple, Google, IBM, and Mozilla all have representation in
the HTML WG and approximate 80% of the market as measured.  Client
detection was based on image support, so "Those email clients that
aren't capable of displaying images, such as older Blackberry models and
other mobile devices cannot be included in this study." We don't have
representation from Blackberry, but we do have representation from some
other handset vendors (Sony, Nokia, Samsung, HP). So I think it's
plausible that over two thirds of email client usage and testing has
direct representation.

+ Web browsers

Internet Explorer, Firefox, Chrome, and Safari command the largest share
of web developer testing:

Unsurprisingly, this reflects their market share among users:

As a web developer myself, I'd say these browser support policies are
typical of the more ambitious end of the scale:

The only maintained browsers mentioned in these policies that do not
have developer representation in HTML WG are KDE Konqueror and Lynx.
It's worth noting that most minority browsers (e.g. Maxthon, Epiphany,
SeaMonkey, OmniWeb, iCab, Chromium) share web engines with the browsers
that do have representation. As mentioned, we have substantial
representation from handset makers and in any case many of the desktop
engines with representation (WebKit, Gecko, Presto, Trident) have made
the jump to mobile, so I don't think our browser representation is
especially skewed to desktop. With respective to assistive technology
used in combination with browsers we have representation from Apple
(who make the VoiceOver screen reader), Opera (who make a addon for
their desktop browser that makes it self-voicing and voice-controlled),
T. V.  Raman (who develops self-voicing Emacs environment Emacspeak),
and Microsoft (who make voice recognition software), so we have some
representation from the assistive technology world, though not as much
as we'd like. (It would particularly nice if Freedom Scientific, who
make the popular Windows screen reader JAWS, and AI Squared, who make
the popular Windows desktop magnification software ZoomText,
participated.) Browser statistics are far from an exact science, but I
think it's plausible to claim that over nine tenths of actual browser
usage and targeting has direct representation.

+ Search engines

Based on Net MarketShare and Comscore stats for Microsoft (Bing) and
Google (Google Search), it's plausible to claim that over two thirds of
search engine usage has direct representation in HTML WG.

+ Authoring tools

Authoring tools are a much more varied category. It's hard to work out
how much of the market we have represented, but I'll try and give some
idea here of the breadth of input.

It seems plausible that email is actually the biggest category of HTML
authoring, so keep in mind the email client representation already

In terms of like wikis, blogs, markets, and social networks are another
big category. Google and Microsoft both offer popular wiki and blogging
tools. We also have a WikiMedia developer, Aryeh Gregor.  Unfortunate
absences from the HTML WG include WordPress, Twitter, Facebook, and

WYSIWYG design tools are another important category. DreamWeaver and
FrontPage are (now Visual Web Developer) the most famous WYSIWYG web
design environments; Adobe and Microsoft have representation. Daniel
Glazman is the lead developer on FOSS alternatives Nvu now

>From the world of text editors, we're fortunate to have two Emacs add-on
developers on the WG: T. V. Raman (Emacspeak) and Edward O'Connor, who
maintains the (X)HTML5 tools for Emacs Lisp.

+ Conformance chechers

Member Henri Svinonen is writing a validator that is also being used by
W3C.  Remember also that developer authoring tools like DreamWeaver and
Visual Web Developer tend to include their own independent validation

+ Web servers

Apache and Microsoft IIS are the most popular web servers and we have
representation from Microsoft. We've also had input from Apache's Roy
Fielding, though I believe his membership may have ended when Adobe
bought his employers Day. It's worth noting that Google develop an
Apache add-on (mod_pagespeed) and run their own custom Google Web
Server, which represents a growing share of pages served:

+ Conclusion

On balance, I submit that my claim that "The implementors involved
produce the software employed by most end-users and targeted by most web
authors" stands up to scrutiny at least to the nine-tenths level when it
comes to browsers and conformance checkers, and at least to the
two-thirds level more generally. If you'd like more information about a
category I've forgotten to mention, please let me know.

Benjamin Hawkes-Lewis

Received on Monday, 3 January 2011 21:45:30 UTC