On Sun, 09 Dec 2007 22:50:30 +0000, Roy Schestowitz
<newsgroups@xxxxxxxxxxxxxxx> wrote:
>Let's begin by considering how Web statistics get gathered for studies which
>strive to show browser/O/S installed base. Here you have some companies that
>receive the log files from various sites/Webmasters that they have an
>agreement with. In other cases, with site meters and certain type of
>XSS 'spyware' (Google AdSense and Google Analytics, for example), you have
>something that collects data from PCs that allow JavaScript to run, don't
>forge HTTP headers, and don't have AdBlock. In the former case, the sites that
>are willing to give away their logs are, in my opinion, site whose visitors
>care less about privacy. Linux users are, on average, more concerned about
>such issues (Groklaw won't touch Google AdSense for that reason alone, let
>alone track usage). In the latter case, rarely will you see a company like
>Google giving away the statistics it collects (which suffer from many problems
>anyway). So, all in all, the nature of statistics and the fact that
>certain /types/ of populations choose to visit certain /types/ of Web sites
>(no equal distribution) are factors that leave you with lies and damn lies.
>It's easier to measure growth, but not to obtain hard numbers.
>
>The statistics are as valid as you wish to believe they are. I have some
>academic background in statistics, so I've seen people cooking 'studies' or
>presenting results where they are aware of weaknesses. Just look at IDC and
>Gartner any day. Even financial analysts...
Ok Roy, let's talk....
In simple terms:
Let's say I'm a Linux advocate for Redhat.
So I use MIT. CalTech, Digg, Propeller, groklaw and other sites which
are going to typically attract nerdy people, me included, who possibly
and maybe even likely, might be running Linux.
Ok the deck is loaded, Linux wins.
Now from the Microsoft point:
I use Tucows, MSN, freesoftware.com etc.
Of course Windows will show up on top
Deck loaded again, but this time in Microsoft's favor.
However, how can you explain sites like w3schools, CNN, Foxnews,
Reuters etc and other sites that are pretty much neutral?
They favor Windows by a large margin and Linux isn't even on the
radar.
Even considering sampling errors, Linux is so far down in the pack
it's not funny.
Also, the "fake the browser id string" argument is flawed because if
you look at the entire string it is trivial to parse the OS from it.
So in conclusion, how do you explain the fact that just about all of
these studies show Linux well below the radar?
Are they ALL wrong or biased?
|
|
