On Oct 22, 10:51 pm, big.momma.thorn...@gmail.com wrote:
> http://news.yahoo.com/s/nf/20071017/bs_nf/56073_ylt=Ag8nNmACRlCEbAit...
> Oh excuse me, Linux market share can't be measured.
The key word being measured reliably.
> Except of course when a Linux zealot discovers some survey, usually on
> a pro linux site that shows a good number and then of course Linux's
> market share CAN be measured.
It depends on how the measurements are taken.
The most common form of survey measures IP addresses and assigns each
address as either a Linux, or a Windows device, based on which is used
most often, or first, or last.
The problem with this approach is that a much higher percentage of
Windows machines use DHCP addresses which are measured as the actual
IP address, while most Linux machines access the internet through
dedicated connections or Network Address Translation (NAT) connections
which make dozens, even hundreds, of devices appear to be coming from
a single IP address.
public assigned DHCP addresses are most frequently found on services
such as AOL, MSN, Cellular modems, and public WiFi networks such as
Starbucks.
NAT addresses are most often used by corporate networks, Hotel
networks, hotel WiFi networks, WiFi hubs connected to cable-modems.
This can skew the number significantly. A single dial-up MSN user for
example, dialing in every day for 2 hours could have 30 different IP
addresses. If the ISP has nice big address pools, such as a Class A
or Class B address, they could significantly skew the numbers in favor
of Windows home computers.
Conversely, the corporate network could have 100,000 computers in it,
and all of them would look like a single public IP address.
The even bigger problem is in how the results are filtered. Most
browsers used on Linux systems are not "branded" as Linux. They are
usually "generic" such as UNIX/X11 or X11. OS/X and Microsoft Windows
require that the proper branding be used in the signature as a
condition of using the required libraries for things like graphics and
file-I/O.
About the only Linux-specific browser is Netscape 7 and later.
Firefox, Konquerer, and others not only don't identify the system as
Linux, but allow the user to manually select a signature indicating
that the browser is IE (fooling web sites that refuse to serve
anything but IE).
> Face the facts, Linux is has and always will be a dud on the desktop
> until it gets support from major hardware and software vendors.
Linux has been "under the radar". It's a bit hard to measure a PC
that doesn't properly identify itself. It's about like trying to
measure black cars on black pavement from a low resolution satellite
in a geostationary orbit by looking at a picture of an entire city.
Another tactic is to use cookies, attempt to collect Windows UID tags,
and assume that any other access that does not identify itself
explicitly is Linux. In the DOJ case, Microsoft claimed that Linux
had grown to 14%, and in the EU trial, Microsoft claimed that it was
17%.
> Until Linux can offer seamless integration with what are 99 percent
> Windows shops, Linux is a dead end.
The other problem is that Linux is not an "either/or" solution. Since
End-Users do the Linux installation, it's possible for a Linux user to
ADD Linux to a machine that is licensed for Windows. The End User can
configure Dual-Boot, Live-CD, USB-Drive, or Virtual Machines to run
Linux on Windows, Windows on Linux, or Windows and Linux concurrently
on ESX or ESXi.
> The facts, and no matter where you look excepting Linux whoring sites
> of course, the numbers are all in the same area and they are miserable.
Most Linux distributors will tell you that even they figure that 90%
or more of their customers, end users, use Linux and Windows rather
than Linux exclusively. This means that for every PC shown in that 1%
figure (actually around 6-7% if you count the "other" catagory as
well), there are about 10 Linux machines. Furthermore, the Linux
percentage has been increasing, while Linux has almost 10 years of
"base". If you figure that Windows has grown from 950 million to 1
billion in the last year, that's only 50 million new users. On the
other hand, if Linux has increased from 4% to 7% and Mac has grown
from 3% to 5%, that means that Linux has added 30 million users and
Mac has added 20 million users, even if you don't figure in the NAT vs
DHCP addresses, that is still giving Mac almost 20% of the market for
the year, and Linux getting 30% of the market for the year.
Ironically, the companies most likely to actually know how many of
which kinds of computers are out there, are very tight lipped about
the numbers. Microsoft seems to be very concerned about Linux, and
they have a number of user authenticated Web Sites. Google has been a
big Linux fan since it's inception, and has still tried to support
Linux users on most of it's sites. Yahoo, E-Trade, and Amazon also
have detailed demographics of millions of users. They also seem to be
very careful about supporting Linux.
There are some demographics available that are more reliable, but they
are sold at $5,000 PER VIEWER, and attempting to publish them on this
newsgroup would probably result in either one very nasty lawsuit, or
15 years in federal prison.
There are several web site surveys available that give a wide range of
results, but when you look at the total Linux (Linux plus "other") you
can see that the Linux market is growing. You can also see that the
Mac market is growing. You can see that Firefox is gaining popularity
as well.