Browser Distribution?

[Hywel Jenkins]

I'm guessing that somebody, somewhere has done a code-based survey and come up
with what percent of people are using this or that browser. Anybody know where
to find something like that?

http://www.thecounter.com/ has/had global statistics for this sort of
stuff. Of course, it's rather inaccurate as I'm sure you already know,
hence the information may be useless to you.

 

[David Dorward]

I'm guessing that somebody, somewhere has done a code-based survey and
come up with what percent of people are using this or that browser.

What with caches and faked user agent strings, you could only get a very bad
estimate of people who browse specific sites.

http://thecounter.com/ has stats for the group "People who visit sites
written by people who care about stats but don't care enough to get half
decent hosting (i.e. access to server logs)".

 

[(PeteCresswell)]

I'm guessing that somebody, somewhere has done a code-based survey and come up
with what percent of people are using this or that browser. Anybody know where
to find something like that?

 

[Safalra]

http://www.thecounter.com/ has/had global statistics for this sort of
stuff. Of course, it's rather inaccurate as I'm sure you already know,
hence the information may be useless to you.

Hmm... 'Netscape 5.x - 11886'. It's hard to believe a browser that was only
released in very buggy source code form would have even that many hits with
any demographic.

Anyway, permit me to advertise my own highly skewed statistics:

http://www.safalra.com/website/browsermarket/index.html

They're probably totally unrepresentative of the global situation (not
least because over 10% of my visits seem to come from the UK), but at least
they make nice reading for Firefox users. Plus they inlcude some of the
more obscure browsers usually omitted from these surveys (IBrowse, AWeb,
Voyager, various mobile devices, and so on).

 

[brucie]

I'm guessing that somebody, somewhere has done a code-based survey and come up
with what percent of people are using this or that browser.

whatever survey you find remember to take into account:

a) the small sample size.
b) the type of sites the counter is used on.
c) the type of people who visit those sites.
d) how the "stats" are collected.
e) use of filtering software.
f) spoofing
g) section 4 of analog 6

 

[cwdjrxyz]

I'm guessing that somebody, somewhere has done a code-based survey and come up
with what percent of people are using this or that browser. Anybody know where
to find something like that?

Look at http://www.w3schools.com/browsers/browsers_stats.asp . These
stats have been collected for several years from users of w3schools.
They likely are not typical of the average computer user, because most
users of the w3schools likely know more about computers than the
average casual user. This likely explains the very high showing for
Firefox compared with some other stats. The hits I get on my domain
have more IE6 users than do the statistics quoted by w3schools. Every
now and then I get a hit from a really old relic. Some users of the
original WebTV(classic) box are still around, and that box is not even
up to html 4.01 standards and has long been replaced by updated boxes.

 

[Alan J. Flavell]

Per (e-mail address removed):

I think that's a central point. TB meaningful for most uses,

But someone designing a specific web site doesn't need to know the
population of browsers averaged over all kinds of user and over the
whole world.

*If* that information was of use to them, it would need to be for the
kind of users who need the content that they're offering.

But looking at the statistics of their own site doesn't help, either.
Suppose their site just happens to be hostile to browser B without
them realising it. Then they'll be under-estimating the number of
visitors who use browser B, and they'd be liable to make their site
even more hostile to browser B - supposing it to be of little
importance to their users - when they really ought to be making their
site more welcoming to it.

I'd think the stats would need to come from someplace like Google -
that is hit by all sorts of people.

Even *they* can't overcome the fundamental limitations of the
principle.

While I wouldn't suggest ignoring the numbers entirely, I *would*
counsel designing for browser-independence, so that the detailed
figures (which in any case are going to be hopelessly inaccurate, for
all the reasons that are usually discussed) are of little importance.

And don't forget that the actual browsing situation isn't just a
browser version. What *appears* to be MSIE could just as well be IBM
Home Page Reader; or it could be some other minority browser
pretending to be MSIE to avoid being locked-out by certain idiot site
designers. Even some indexing robots fake themselves as an MSIE
version.

 

[(PeteCresswell)]

Per (e-mail address removed):

These
stats have been collected for several years from users of w3schools.
They likely are not typical of the average computer user, because most
users of the w3schools likely know more about computers....

I think that's a central point. TB meaningful for most uses, I'd think the
stats would need to come from someplace like Google - that is hit by all sorts
of people.