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

Re: Estimating number of existing Debian users

On 5/9/07, Junichi Uekawa <dancer@netfort.gr.jp> wrote:

I've tried to craft an hypothesis about number of Debian users, from
apt-listbugs logs. I've analysed merkel.debian.org access logs. Since
etch was released, 30000 unique IPs were used to access
merkel.debian.org using apt-listbugs. According to popcon, 3% of all
Debian users install apt-listbugs. Let me add three simplifications to
the equation.

(A) popcon models Debian user installation behavior reasonably well.

(B) A unique IP represents a single system (Even though DHCP may
change IP address over a month's period, NAT may share IP addresses).

(C) All users are using Debian etch, lenny, or sid.

I can assume there are 1,000,000 Debian users worldwide. Does this
number sound big or small to you?

Hi Junichi,

I'm sorry, but the number is wrong as I replied to your blog post[0].

This is a wild extrapolation, so the number is pretty unreliable,
especially the usage of popcon is too small to be significant;
however, I do not know of a better way to estimate the number of
Debian users.

Exactly as I wrote, and there's a big difference between Debian
installations and users. We've more users than installations, estimate
(not measure) installations is easier than estimate number of users
(direct or indirect). With direct I meant 'getent'able users and
indirect somebody browsing a website running Debian.

Btw, are you trying to estimate the number of installations (not
users, no way) or reach the exact number? Reading again the two
paragraphs above it isn't clear. Based only on your blog post content,
the answer would be exact number and that's also impossible with the
data on the table.


[0] = http://stratusandtheswirl.blogspot.com/2007/05/re-hypothesis-about-number-of-debian.html

-- stratus

Reply to: