Hmmm. A question. Was [Re: Debian is losing its users]
-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA512
Hi Everyone,
I have a couple of observations, and a couple of questions to think on.
Monique Y. Mudama wrote:
| On 2008-03-27, Wei Chen penned:
|> Hi,
|>
|> I am somewhat disappointed that when you see that post, what most of
|> you, if not all, reply to argue that the decrease of search volume
|> does not indicate the loss of users, rather than thinking of ideas
|> for something that can be done to help, which was the original
|> motivation of the post.
|
| I just assumed you were trolling with the first post. I'm still not
| sure you're not.
|
| If your intention was to get people interested in generating ideas to
| increase the popularity of debian, I don't think your first post did a
| good job of communicating that intention. And even if that had been
| clearly communicated, another question is, does everyone agree that
| increasing popularity is always a good thing?
First things first. I typed in many popular search terms into that page, and
found that the instance of all were declining. I would also note that this is
only *one* search engine. It could simply indicate that people are using other
search engines more, and that one less.
The questions. Has anyone actually shown that Debian is losing its users? A
loss of search statistics from one search engine really proves nothing -
established users would not use "Debian" in a search engine very often.
To directly reply to the question about popularity, nothing is always a good
thing, and popularity is no exception. The more people using Debian, the
greater the load on their servers would be, and the greater the demands on
their staff would be. The choice would then be to either wait longer between
releases, or add more cooks to the kitchen.
|> Even if the losing of users is not proportional to the decrease of
|> search volume, there is a high probability that they are positively
|> correlated. I hope some of you are willing to think of some
|> constructive ideas for that. And in the meanwhile you may want to
|> stop this meaningless dispute.
Let us say that we can show that GNU/Linux and Debian are not being actively
searched for on *any* search engine. Let's also assume that people are
migrating away from Debian and GNU/Linux, in general. Can we actually prove
that it has anything to do with people not searching for those terms? We could
show a correlation, but *correlation does not prove causation*.
There are a few good logical reasons why a reduction in search statistics would
neither prove a *loss* of users nor provide a cause for it, if it can be shown
to exist. One I have brought up - current users of Debian (or any
distribution) would very seldom use search engines for their distributions.
Instead, they would use the resources available on their distribution's home page.
| "There is a high probability ..." That's a pretty bald statement.
|
| An interesting question would be, how do you actually go about
| establishing usage statistics? The number of distinct IP addresses
| that pull data from all the mirrors? What's a stat one can actually
| collect?
I have some follow-up questions on the, "There is a high probability ..."
statement. How would one go about calculating such a probability, especially
since A (a massive reduction in searches for the term "Debian") has not been
proven, since only one of many search engines was used. B (the loss of Debian
users) is only assumed, and never proven (at least that I can see).
In order to accurately state the A and B are correlated, you must first prove
that both A and B are accurate. Do we know that that data provided by that
search engine is accurate? They do not show a margin for error, or other
little details that would make it seem more accurate to me. Do we have
statistics on the number of Debian users over the same period that we can compare?
Regards,
Chris
-----BEGIN PGP SIGNATURE-----
iD8DBQFH6/EJUx1jS/ORyCsRCqCTAJ9ElvWC+f7+qTAIlUR4kqo1Y9RLpQCfTgWD
Swbolz14bMDJUYBa+dlgGZ0=
=qbQr
-----END PGP SIGNATURE-----
Reply to: