In comp.os.linux.advocacy, Rex Ballard
<rex.ballard@xxxxxxxxx>
wrote
on Fri, 6 Jun 2008 11:55:47 -0700 (PDT)
<73ba3de1-cd17-4783-b89b-efcacc379eb9@xxxxxxxxxxxxxxxxxxxxxxxxxxxx>:
> On Jun 1, 11:21 am, bbgruff <bbgr...@xxxxxxxxxxx> wrote:
>> Latest Net application figures:-
>> http://marketshare.hitslink.com/report.aspx?qprid=9
>
> I went to the site, it looks like a very flawed survey, but they don't
> show the methodology they used. I've seen similar surveys and the
> methodologies used were very "Windows centric".
http://marketshare.hitslink.com/
could be slightly clearer but the parsing of the User-Agent: line
might be construed as a trade secret.
>
> Only count whatever is in the first entry of the browser Identity.
> Linux browsers usually have X11 as the first entry and don't include
> the word Linux until almost the end of the entry. These are semicolon
> delimited fields, and so it's easy to "not count" Linux.
Uh, no. Konqueror has an identity that is
Mozilla/5.0 (compatible; Konqueror/3.5; Linux 2.6.23-gentoo-r5; X11;
i686; en_US) KHTML/3.5.9 (like Gecko)
The X11 is quite obviously stuck in the middle.
There are many options: "Operating system name" is
Linux. "Operating system version" is 2.6.23-gentoo-r5.
Platform name is X11. Machine (processor) type is i686.
Language information is en_US.
And the very first option is whether any identification
is sent at all.
Firefox on Linux gives
Mozilla/5.0 (X11; U; Linux i686; en-US; rev:1.8.1.14) Gecko/20080425
Firefox/2.0.0.14
Epiphany is almost identical:
Mozilla/5.0 (X11; U; Linux i686; en; rv:1.8.1.14) Gecko/20080425
Epiphany/2.20 Firefox/2.0.0.14
as is Galeon:
Mozilla/5.0 (X11; U; Linux i686; en-US; rv:1.8.1.14) Gecko/20080425
Galeon/2.0.4 Firefox/2.0.0.14
and Seamonkey:
Mozilla/5.0 (X11; U; Linux i686; en-US; rv:1.8.1.13) Gecko/20080522
SeaMonkey/1.1.9
and Kazehakase:
Mozilla/5.0 (X11; U; Linux i686; rv:1.8.1.14) Gecko/20080425
Kazehakase/0.5.0
Opera verges on the minimalistic:
Opera/9.50 (X11; Linux i686; U; en)
and Dillo verges on the cryptic:
Dillo/0.8.6-i18n-misc
and so does wget, though in all fairness that's not really
a browser (but it *is* a user agent, just one optimized
for automated downloading):
Wget/1.11.1
Java's JEditorPane also verges on cryptic:
Java/1.6.0_05
Lynx throws in some mildly interesting crap:
Lynx/2.8.6rel.4 libwww-FM/2.14 SSL-MM/1.4.1 OpenSSL/0.9.8g
and Links throws in the window size for good measure (no, I haven't
a clue as to why this is a benefit):
Links (2.1pre17; Linux 2.6.23-gentoo-r5 i686; 80x24)
For its part my IE6 installation using Wine uses an ancient OS, or
so the rest of the world might think:
Mozilla/4.0 (compatible; MSIE 6.0; Windows 98)
There's probably a configuration adjustment somewhere in ~/.wine that
might tweak this, but there's probably more important things
to worry about, like how long my ear hairs are.... :-P :-)
All of amaya, arena, and mnemonic appear dead, at least
from a Gentoo standpoint (emerge can't find an ebuild for
any one of them). (Amaya's code was pretty junky C stuff.
Arena handled HTML 3.2 and was never upgraded AFAIK.
Mnemonic was rather ambitious at the time but never got
off the ground.)
>
> Any values you don't recognize - assume that they are not browsers and
> throw them out (90% of this "other" category is Linux workstations
> running Firefox, Konqueror, or Mozilla.
>
> Only count one OS per IP, and let Windows "override" Linux entries.
>
> Use activeX controls or other Windows specific elements before the
> count. The result is that the browser can't send back the
> information, or the browser is disqualified because it doesn't support
> functions such as PDF or Flash or ActiveX.
>
>> Over the past 10 months:-
>> Windows: From 93.28% to 91.13% of User base
>> Mac: From 5.99% to 7.83% of User base
>> Linux: From 0.46% to 0.68% of User base
>
> If we figure that 1% is roughly 10 million users, that would indicate
> that even with this screwy survey, Linux grew from 4.6 million to 6.8
> million users - an increase of 2.2 million users.
>
>> Hence, that puts Market Shares (meaning "% of new machines sold with" - and
>> perhaps a bit conceptual re. LinuX) at about:-
>>
>> Windows: 81%
>> Mac: 17%
>> Linux: 1.7%
>
> Here is another survey that is also flawed, but less flawed than the
> one above.
>
> http://www.w3schools.com/browsers/browsers_os.asp
> Windows XP 73.3%
> Windows 2000 3.3%
> Win98 .05%
> Vista 8.8%
> W2003 1.9%
> Linux 3.7%
> Mac at 4.6%
> And "other" = 3.9%
>
> That would put Linux at Roughly 76 million users.
An optimistic assessment IMO, though heartening.
>
> Vista seems to gave gone up to 88 million users (even though 200
> million PCs were sold this year).
>
> Keep in mind that this survey, like the other, only counts IP
> addresses, not cookies, and therefore miscounts corporate desktop and
> laptop machines as well as ISPs who use NAT routing.
Not to mention emulators, rotating proxies, and other such miscellania.
> Most Linux users
> are behind NAT firewalls (because they tend to be more security
> conscious)..
Well, in my case it's almost by accident now (though I
did buy a NAT switch/router at one point as well, since I
needed a switch); Earthlink's DSL service includes a little
free black box. (It's definitely little, black, and boxy.)
> Microsoft can skew IP surveys in it's favor by
> increasing the number of IP addresses used it it's DHCP pool, this
> includes users who come in via MSN, AOL, and cellular modems, which
> has been increasing the number of users.
And then there's GoDaddy, which I get the feeling they included
solely because they wanted to make their server numbers look good.
>
>> For Mac to be so high (and for that matter, Linux so low), I can only think
>> that the Net applications sampling is very biased towards U.S, sites - an
>> observation which is borne out by their browser stats.
>
> Not just US based, but probably a very limited demographic base, based
> on Microsoft oriented content. I don't know if this is the case or
> not, but I do remember a similar survey that was only supported on
> Windows servers and required a front page extension to actually get
> the "counter" to work. There could have been millions of users
> hitting the site but didn't get counted because they were using Linux
> and "bounced off".
The main problem is that Apple's sopping up the
Vista malcontents. MacOSX Leopard is a slick,
skillfully-implemented affair which appears almost seamless
(more properly, the seams are in areas where they don't
get in the way).
Plus, the iPhone and the Air make life interesting,
though I'm not sure as to the Air's usefulness; my nx9010
is dog slow in comparison but has a lot more capability
built in, as it has a 3 1/2" floppy, a CD reader, a
Firewire connection, 3 USB ports, a D25 for a printer,
and a combination mouse/keyboard connector (just the one,
strangely enough, but it works nicely for both), plus a
Windows-friendly Broadcomm WiFi modem (I need ndiswrapper
to make it work). And of course it can harbor a 2" hard
drive in the front -- 160 GB is now nicely ensconsed there.
Sexy? Probably not. Capable? For now, yes, though my
Athlon has better graphics capability (though the nx9010
can do OpenGL...just not as well).
If one wants a car analogy (well, OK, "wants" is way too strong
a word):
Linux: kit car, made out of standard spare parts and very robust,
or maybe a Chevy or maybe even a '68 Dodge Charger [a la
"them good ole Duke Boys"]
MacOSX: Cadillac, with comfortable everything and lots of power
Windows: Either a Trabant or a '74 Pinto with a glittery paint job
but not much else.
--
#191, ewill3@xxxxxxxxxxxxx
People think that libraries are safe. They're wrong. They have ideas.
(Also occasionally ectoplasmic slime and cute librarians.)
** Posted from http://www.teranews.com **
|
|
