On Aug 6, 2:50 pm, The Ghost In The Machine
<ew...@xxxxxxxxxxxxxxxxxxxxxxx> wrote:
> In comp.os.linux.advocacy, Rex Ballard
> <rex.ball...@xxxxxxxxx>
> > Inhttp://www.w3schools.com/browsers/browsers_os.asp
> > I get
> > Vista 11.5% 1 in 10 users after 18 months (XP had 33% after 18
> > months).
> > Linux+Other is about 9%
> > Mac is 4.8%
>
> All of these figures are extremely suspect -- although I
> can't say how to make them better.
Actually, they are suspect. They only count IP addresses, which means
that Windows users who use DHCP assigned public addresses will be
overcounted, and Linux users who are more security concious and use a
NAT router that can retain the same IP address for months, are
undercounted.
> The problem is that www.w3schools.comis a self-selected sample, and as any
> serious pollster can tell you, one can draw no conclusions
> from such regarding the superset of that sample.
Ironically, it's a bigger sample than you might think. W3schools is
the site that provides the tutorials for pretty much every aspect of
the web from sending http to writing HTML to PHP to XML and Ajax
tutorials. Of course, one of their most popular links is the browser
survey, which is one of the first listed in the google target "Browser
Statistics".
> For all we know bots are hitting w3schools.
It's possible, however, since this only counts IP addresses, the bot
would only need to send one message per week to show up in the count.
Not exactly a swarm the extra hits from the same IP address wouldn't
change the count at all.
> The same problem presumably plagues hitslink.
Hitslink, which is so popular among the wintrolls, requires that the
browser load and execute an ActiveX control. One can see why the
Linux count might be a bit low.
> >> Linux is an enormous failure as a consumer product, and would have been
> >> killed off years ago if a return on investment was required.
>
> > Linux on the DESKTOP has faced the full force of Microsoft's $40
> > billion in leveragaged advertizing revenue,
> > $40 billion in OEM and Corporate revenue which fund a $4 billion/year
> > legal and settlements fund.
>
> True, and it's still doing remarkably well, regardless.
> I also suspect that many Windows-side desktops are in
> fact dual-boots (certainly three of mine are, although
> I do not boot into Windows that often).
The big trend these days is Virtualization. Furthermore, VMWare has
made it much easier to generate a Windows "appliance" from your
existing PC configuration, then install Linux as the Native OS, and
then run Windows as the VM. That way, you only need the dog-slow
Windows system when you absolutely have to run some Windows
application or utility that isn't available on Linux.
> > According tohttp://www.w3schools.com/browsers/browsers_stats.asp
> > FireFox has 42.6% of the market, which could be as many as 800 million
> > PCs (since many FF users also use IE and both would be counted in this
> > survey).
> Same problem. Still, Firefox does seem to have carved out a
> hefty niche; it might even best IE8. Microsoft might have
> a slight problem here....
Open Office is not doing too shabby either. The thing that makes
these numbers interesting is that if you figure that all of the FF
users are also using IE, that could mean that Linux could be on 80% of
the machines. Assuming 1.2 billion machines, if IE was installed on
all 1.2 billion, and FF was installed on 800 million machines, that
would make a total "count" of 2 billion combinations. FF would be on
40% and IE would be on 60% of the 2 billion. So in reality, FF would
be on 800 million of 1.2 billion machines, or just over 60%.
Going back to the w3schools survey again, the similar "double
counting" occurs in that survey. If you have 2 billion combinations,
and 220 million of those combinations are Linux (11%), but there are
really only 1 billion machines out there, thout would mean 22% of the
actual machines were running Linux at least some portion of the time.
Of course, since the Linux NAT addresses are hiding more machines, it
could be 33% or more.
|
|
