[thelist] Drawing the line was Tip o'the day (x2)

aardvark roselli at earthlink.net
Sun Feb 18 16:44:36 CST 2001

> From: "Philippe J" <philippejadin2 at swing.be>
> > and where's the definitive users browser preference stats for a
> > given demographic.
> You'll never know that. The only reliable way would be to make a first
> site totally compatible with any browser, then analyze the stats**,
> and then see if your site can be built in a more high-tech way*. I

agreed, with comments on your footnotes...

> **no one ever speaks about the fact that stats are not totally
> reliable, because some older clients won't even be detected by the
> stat software. In the "100%" browsers, you can count a few % using a
> browser so old that it can't be "detected", either because the
> detection is based on a non working javascript in older borwsers
> (which is basically stupid), or because the browser doesn't send any
> "client" information. It will be placed in the "others" category.

i can say that i get a pretty darn good cross-section of my users... 
however, i don't rely on client-side browser checking... that's just 
completely unreliable...

instead, i use server-side to gather the info from the HTTP 
headers... of course, some browsers allow you to modify it, others 
may not send it (still trying to find one of those), and some people 
hack it... in each of those cases, that data is bad data no matter 
what since the user clearly wants a different experience and isn't 
interested in being tracked... the advantage is that the user has 
either entered a bad record into the logs (making sure i don't track 
it), or mimicked another browser (throwing support in with another, 
thereby making some kind of choice)...

as for screen size, well... i'm glad you (didn't) ask(ed):

Real-World Browser Size Stats, Part II

More information about the thelist mailing list