In comp.os.linux.advocacy, Ezekiel
<e@xxxxx>
wrote
on Thu, 31 Jul 2008 21:26:25 -0400
<ed228$48926640$9549@xxxxxxxxxxxxxxxxx>:
>
> "The Ghost In The Machine" <ewill@xxxxxxxxxxxxxxxxxxxxxxx> wrote in message
> news:h5p9m5-kvj.ln1@xxxxxxxxxxxxxxxxxxxxxxxxxx
>> In comp.os.linux.advocacy, Ezekiel
>> <zeke@xxxxx>
>> wrote
>> on Thu, 31 Jul 2008 16:39:14 -0400
>> <ef0d6$489222f2$13171@xxxxxxxxxxxxxxxxx>:
>>>
>>> "Roy Schestowitz" <newsgroups@xxxxxxxxxxxxxxx> wrote in message
>>> news:2361986.DX2ftdLHxT@xxxxxxxxxxxxxxxxxx
>>>> -----BEGIN PGP SIGNED MESSAGE-----
>>>> Hash: SHA1
>>>>
>>>> Firefox market share exceeds 20%, Internet Explorer dips below 70%
>>>
>>> So magically today marketshare of a *FREE* product can be measured... and
>>> even measured accurately.
>>>
>>> Amazing because every other day the 0.x% marketshare of linux is
>>> "impossible" to measure.
>>>
>>> Linux "advocates" are such hypocrites.
>>
>> At least in this case, one might be able to have a more
>> accurate determination of a product's share by using log
>> headers generated by usage thereof, as opposed to the
>> log headers generated by usage of a small subpart.
>
> Why would it be more accurate?
Mostly because the tool is communicating directly with
the server, as opposed to an example tool of the solution
communicating with the server. If you look at Konqueror's
settings, for example, you may notice that one can include
or exclude various information at will, including OS type.
> What about firewalls, proxy servers and
> everything else that supposedly "skews" the results.
None of them will skew the results, unless they do very
hefty header rewrites. Most solutions key on User-Agent,
and proxies aren't really supposed to touch that, and most
have no need to.
> What about Rex's
> diatribe about how all of these companies require that the user install an
> Active-X control in order to be counted on the web stats.
I am not aware of any such requirement, nor have I read
Rex's statements in detail. An ActiveX control is usable
on Linux, though the circumstances are extremely limited.
> Unless they
> install the ActiveX control they don't show up at all in the web stats.
> Suddenly this is no longer the case. Ghost... you can't have it both ways
> depending on which direction the wind is blowing.
I think you can see some of the difficulties of using weblogs
to determine accurate statistics. ;-)
>
>
>> Then again, you're absolutely right; for all we know the
>> IE marketshare is over 99.5%, with Firefox and Safari
>> fighting over the bitter dregs of what's left.
>
> I never made this claim.
I am not saying you did (the "absolutely right" refers
to your claims that "magic marketshare" is wrong). I am
merely saying it's possible.
> Not once, not ever. But it's hypocritical to come
> up with a thousand and one excuses of why web stats are unreliable and can't
> be trusted and how the system is rigged to only count IE browsers with
> special ActiveX controls and then the next day use web stats as proof of
> something.
Webstats are utterly reliable, in any event, for the usage scope
they cover. Basically, we can claim with utter certainty that
Linux boxes comprise at most 0.8% of the visitations for hitslink.
And that may be a small overestimate because of proxies.
I know of very few Windows proxying models.
>
>
>> Somehow, though, I rather doubt it.
>
> You and me both.
>
>> A more (though still not very) likely possibility is
>> that custom user-agents, used for the propagation of
>> certain malware/lameware, might be attempting to infect
>> w3schools.org or other such sites, using carefully crafted
>> packet attacks that look exactly like Mozilla Firefox
>> browse attempts.
>>
>> I rather doubt that, too.
>>
>> [.sigsnip]
>>
>> --
>> #191, ewill3@xxxxxxxxxxxxx
>> /dev/signature/pedantry: Resource temporarily unavailable
>> ** Posted from http://www.teranews.com **
>
>
> ** Posted from http://www.teranews.com **
--
#191, ewill3@xxxxxxxxxxxxx
Useless C++ Programming Idea #889123:
std::vector<...> v; for(int i = 0; i < v.size(); i++) v.erase(v.begin() + i);
** Posted from http://www.teranews.com **
|
|
