begin oe_protect.scr
Roy Schestowitz <newsgroups@xxxxxxxxxxxxxxx> espoused:
> __/ [ Sinister Midget ] on Monday 03 April 2006 10:33 \__
>
>> On 2006-04-03, Handover Phist <jason@xxxxxxxxxxxxxxxxxxxxxx> posted
>> something concerning:
>>> Roy Culley :
>>>> begin risky.vbs
>>>> <1f9nbyadqzv08.dlg@xxxxxxxxxxxxxxx>,
>>>> Erik Funkenbusch <erik@xxxxxxxxxxxxxxxxxxxxxx> writes:
>>>>> On Sun, 02 Apr 2006 18:27:17 -0400, rapskat wrote:
>>>>>
>>>>>> This isn't the way it goes. The change is subtle, gradual. Windows
>>>>>> wasn't an overnight success. M$ didn't become the most powerful
>>>>>> software
>>>>>> company in the world over the course of a year. Rome wasn't conquered
>>>>>> in a day.
>>>>>
>>>>> If by subtle and gradual, you mean not even on the radar.
>>>>>
>>>>> I have access to a lot of web site data for sites that have a wide
>>>>> spectrum of interest, and at best Linux could be counted in the 3%
>>>>> "unknown". The actual stats that Identify themselves as Linux are
>>>>> less than 1%.
>>>>
>>>> So back this unsubstantiated with verifiable data or shut up.
>>>
>>> There's an interesting thought. From a gaming website that I run (my
>>> stats arent counted or there would be MUCH more Linux traffic) over
>>> the course of a single day (yes, it very new and has little traffic):
>>>
>>> Windows XP 229 78.97%
>>> Windows 2000 34 11.72%
>>> MacOS X 11 3.79%
>>> Linux i686 6 2.07%
>>> Windows 98 6 2.07%
>>> Windows 2003 2 0.69%
>>> Windows ME 2 0.69%
These are interesting, but the absolute totals are probably not large
enough to form a useful picture. Also, the nature of the site might
tend to encourage one type of user more than another. Personally,
although I still have some gaming interests, my kids are interested in
gaming, and I tend to look at Linux-centric gaming sites. I suspect
that if you looked at happy penguin you'd see a very different (but not
necessarily more meaningful) picture. Samething would apply to a
distro's site - how useful would Debian's stats be?
Larger absolute numbers from general purpose sites might be more use?
>>>
>>>
>>> So there we have it, a market segment (gamers) that prefer Windows over
>>> Linux to a huge degree. I would love to see Slashdot browser stats.
>
>
> FWIW, here are last month's stats from my main site. The stats exclude robots
> and crawlers, which sum up to similar figures:
>
> Windows 573642 71.3 %
> Unknown 98424 12.2 %
> Linux 85656 10.6 %
> Macintosh 42126 5.2 %
> Unknown Unix system 1864 0.2 %
> FreeBSD 1009 0.1 %
> Sun Solaris 519 0 %
> NetBSD 202 0 %
> OS/2 163 0 %
> OpenBSD 55 0 %
> Others 122 0 %
These numbers are large enough to give a good picture for the top 4, but
not enough after that. You'd want a /really/ big set of stats to get
the rest, I think. perhaps google ;-)
>
> The numbers correspond to hits. bear in mind that unknowns are most likely
> Linuxes. Windows /does/ know how to identify itself uniformly.
>
>
>>> I dont think looking at those stats shows a fair representation since
>>> it's a Windows-centric site. So how would we go about gather a fair
>>> representation? Suggestions? I did have a poll on the site asking if
>>> people wanted to see what was offered for games on linux recently and
>>> there was a %100 yes reply from the users, but they generally know I've
>>> been using Linux for ages and am a reliable source for this sort of
>>> information, so is that a fair representation? I dont think browser
>>> stats are a good approach and I cant think of a good one off the top of
>>> my head.
>>
>> Take a poll and ask them what /they/ use. Allow multiple selections.
>
>
> A poll would be no good. The selection of the sites is imporatnt too. Just as
> Microsoft selected 400 businesses which they claim preferred moving from
> UNIX to Windows rather than Linux, they could intervene with Web-based
> statistics. I wouldn't trust a word that comes from studies as long as money
> can be exchanged under the table and a certain company has plenty of FUD
> money to spare, as well as a motive.
Quite.
>
>
>> Still not perfectly accurate, but better than counting user agent
>> strings. It would certainly be _more_ accurate if they're allowed to
>> show they use more than one thing.
>
>
> Frankly, my user-agent string does not include "Linux". I spoof it. Many
> others do the same, knowingly or unknowingly.
>
> Best wishes,
>
> Roy
>
--
| Mark Kent -- mark at ellandroad dot demon dot co dot uk |
Flirting is the gentle art of making a man feel pleased with himself.
-- Helen Rowland
|
|
