__/ [ Linonut ] on Thursday 28 December 2006 16:18 \__
> After takin' a swig o' grog, Phil Da Lick! belched out this bit o' wisdom:
>
>> OK wrote:
>>> http://www.thecounter.com/stats/2006/December/browser.php
>>>
>>> Not really a surprise...
>>
>> What's really surprising about that is that even with virtually enforced
>> download of IE7 that it still hasn't overtaken IE6.
>
> Never trust a "counter" site that provides "accounts". Not even a Linux
> site.
>
> http://www.thecounter.com/stats/2006/December/index.php
>
> Mon Dec 4 15:08:15 2006 - Thu Dec 28 10:58:00 2006 23.8 Days
>
> 28129374 Visitors
> 49192 Visitors/hour
>
> Now tell me that a site devoted to webmasters has 49000 human visitors
> per hour.
>
> http://www.thecounter.com/stats/2006/December/domain.php
>
> Domain Stats
>
> 7. SWEDEN (.SE) 632574 (2%)
> 8. BRAZIL (.BR) 608997 (2%)
> 9. TURKEY (.TR) 595259 (2%)
> 10. AUSTRALIA (.AU) 299006 (1%)
> 11. Educational (.EDU) 275607 (0%)
> 12. UNITED STATES (.US) 235497 (0%)
>
> 23. INDIA (.IN) 117758 (0%)
>
>
> Now tell me that Swedish government sites are more popular than US or
> India sites.
>
> The OS stats are just as unbelievable.
>
> The resolution stats? You gotta be fscking kidding me!
>
> This is the funniest one of all:
>
> http://www.thecounter.com/stats/2006/December/weekday.php
>
> Weekday Stats
>
> Sunday 2247795 (7%)
> Monday 4339537 (15%)
> Tuesday 6233347 (22%)
> Wednesday 6079268 (21%)
> Thursday 4142277 (14%)
> Friday 2892307 (10%)
> Saturday 2194843 (7%)
>
> It adds up to 96% (LOL!)
"Unknown" gets ignored (tossed way). That's GNU/Linux. I mentioned this in
the other post. If the stats add up to 100%, then something if fishy. In
busy sites, around 10% of the requests cannot be analysed or determinded
with confidence. Also ask yourself, /which/ Web sites are included in the
survey? Where are they from? What do they cover? What are their relative
sizes? If the same UIP comes up with different O/Ses (dual boot), how is it
counted? Is it being discarded?
You can build studies as such to prove /anything/. Fit your desired
conclusion/hypothesis to your data/method and you'll always be a winner. Are
your data instances representative of the entire available sample? What is
it that you gauge anyway? It's a black art which Microsoft masters.
There are over 100 million registered domains. Are my sites being counted at
all? In one of my sites, 44% of the visitors use Linux ("unknown" aside) and
only 9% use Internet Explorer.
Are we living under some illusion that extends beyond the installed
base/market share chokehold? Are there weapons of mass destruction?
--
~~ Kind greetings and happy holidays!
Roy S. Schestowitz, Ph.D. Candidate (Medical Biophysics)
http://Schestowitz.com | RHAT Linux | PGP-Key: 0x74572E8E
6:05pm up 71 days 4:19, 6 users, load average: 0.19, 0.18, 0.08
http://iuron.com - Open Source knowledge engine project
|
|
