[Date Prev][Date Next]   [Thread Prev][Thread Next]   [Thread Index] [Date Index] [Author Index]

Re: Need a Usage Survey (WAS Re: Fedora extras and the distribution size)



On Wed, 14 Jul 2004 21:06:44 -0700, Max K-A <max_list fedorafaq org> wrote:
>         For example, if I put up: "What MTA do you use in your Fedora systems:
> Sendmail (the default), Postfix, Exim, other" and we see that only 10%
> of people are using one of them, we can actually say, "Let's put it in
> Extras." If the numbers change in the future, we can move it back. None
> of it is permanent. :-)

Flawed methodology... these numbers only make sense if a majority of
the userbase CARE
about specific packages.  I frankly dont give a flip about which mta i
use, i'm not supporting any machines that NEED anything beyond sending
mail from localhost out to the net. I never notice sendmail so i
frankly do not CARE.  The fact that I'm using sendmail is only
evidence that its the default application to fit the function. And I
would dare say, that a lot of people... a majority of users in the
userbase are going to feel the same way about MOST if not all of the
applications and services... they just dont care about the specifics
beyond a couple of pet peeve apps.  What you are going to find with
this survey, are the squeaky wheels, the small percentage of the
userbase not satified with the defaults. A survey that asks
participants to self-select is flawed, its not going to cut across the
full userbase in a representative fashion. Unless you can get an
accurate representation of the total number of installs and have a way
to RANDOMLY survey the userbase, your data wont represent anything
beyond a weak statement like "some people in the userbase don't like
sendmail as the default" becuase you will have no factual or
statistical basis to say that the respondants in your survey represent
a valid random cross-section of the base. Its flawed, dont do it, your
just going to piss people off and give them bad data to base flamewars
on.

-jef"a survey like this might barely meet FoxNews definition of fair
and unbiased...barely"spaleta



[Date Prev][Date Next]   [Thread Prev][Thread Next]   [Thread Index] [Date Index] [Author Index]