Philip Chalmers

Mark,

I bookmarked your site (http://www.howtocreate.co.uk/) ages ago because the
content's so useful.

But I disagree with the tone of your "No stats" page
(http://www.howtocreate.co.uk/nostats.html). OK, browser stats are subject
to all the distortions you described. But:
*	It's impossible to support literally every browser -
[URL] lists so many that the page takes forever to load,
let alone read.
*	Only a mega-corp can afford all the platforms you'd need in order to test
for every browser - Win (PC), Mac (Mac OS and OS X), Linux (KDE and Gnome),
Solaris, Win CE, Palm, BEOS, Atari ST, Amiga, .........
*	If you tried to support every browser you'd be reduced to the lowest
common denominator of facilities, probably HTML 2.0. Goodbye CSS, JS, etc.
That's unfair on users of good browsers, who are the vast majority.
*	Even "good" modern browsers are imperfect- you can't just write
standards-compliant HTML / CSS / JS and assume it will work.

So developers have to draw a line somewhere, particularly on which browsers'
quirks they will work round, and they need some guidance on this. Browser
stats are imperfect, but they're the best info available.

Best wishes,

Philip Chalmers

Mark,

I don't think I've  missed the point, but perhaps I wasn't clear enough.

Let's look at some cases:
*	Netscape 4 was so bad that it didn't just mess up CSS-based page layouts,
it crashed whenever it saw some valid HTML / CSS combinations and sometimes
took Windows with it. So graceful degradation was difficult, and often
wasn't possible without scripting (server and / or client). And adding code
to work round NN4's defects made pages load more slowly and sometimes
limited them what they could do in terms of facilities and / or aesthetics -
which was unfair to users of better browsers (the majority). The question
for developers was "When I can I stop worrying about NN4?" Only browser
stats could give some sort of answer.
*	One of these years we may get a standards-compliant Internet Explorer.
Then developers will need stats to tell them when they can stop worrying
about old, non-compliant versions of IE.

I also think you give the game away by saying, "...you should make sure your
page works on the
lowest common denominator...". What's the lowest common denominator? From a
theoretical point of view, probably HTML 2.0, w/o CSS or JS. But then you'd
have to limit pages to the capabilities of HTML 2.0 and treat all more
advanced facilities as mere window-dressing. The only realistic alternative
is to find out what is the actual lowest common denominator, i.e. the least
capable browser with a market share worth considering (which will vary
depending on the site's objectives). For that you need stats.

Best wishes,

Philip Chalmers