Number of Mageia Installs

[marcpare]

Does anyone have an estimated number of Mageia installs worldwide? Are there numbers that we can all quote and feel confident in quoting these estimations?

It would also be good to know how any number would have been estimated, and by which method.

Marc

[doktor5000]

No.
There have been multiple discussions and attempts to add some kind of anonymized counter or metrics but none of those went through because of tracking / privacy concerns.

[marcpare]

Is the https://www.Mageia.org site the only one where people get to download the official ISO's? IMO, the Mageia.org site should be the only site where people should be downloading the official ISO's.

Would there not be, at the very least, a way whereby an estimated number of downloads could be gathered, just by the sheer number of ISO downloads from the official Mageia.org site? At least, this would give us an "estimated" number of downloads for which could be claimed in publicity. Would it be difficult to gather this data from the point where Mageia9 was released to today's date? One could gather the same data for the preceding Mageia versions for the sake of comparison as well.

I doubt that this kind of data would be cause of any privacy worries.

[sturmvogel]

There seems to be a lack of understanding how the distribution works. Each linux distro provides their software via mirrors which are placed all over the world. Each mirror serves as provider for multiple distros/projects/companys and more.

Example mirror which serves multiple distros/projects including Mageia:
http://mirror.accum.se/mirror/

Mageia description:
https://wiki.mageia.org/en/Mirror_List
https://mirrors.mageia.org/


If you use https://www.mageia.org/en-gb/downloads/, one of the many mirrors is used to download (depending on the location and speed/quality of the mirror). This can also easily be seen when you click down to the actual ISO. There you can see which mirror is choosen for you. There you can also decide to use another mirror.

[marcpare]

I don't believe how the ISO's are served up from different mirrors is an issue; I would think that most people on this forum are aware of this. However, it should not be impossible to get download stats from each mirrors and arrive to a certain estmate of overall numbers.

I am part of a large open source group where the ISO's are also delivered through their official website and thereafter various mirrors. They manage their stats and have a good enough estimation of how many installs have been made worldwide on any given month/day etc..

I am not sure why this could not be accomplished from Mageia. I am not sure how Mageia can exist without supporting stats for its marketing/publicity team, or, are these stats already available internally to team members?

[benmc]

just my $0.02.

lets say I download 1 .iso.

what do I do with it?

Do I just leave it in my downloads directory, never to try it because something newer / shinier catches my eye?

Do I install it once, decide I don't like it, overwrite the install with another distro?
do these this count as one installation?

or

I install it once , decide I do like it, and proceed to install it into 100's of computers, friends, family, strangers off the street!
does this count as one installation?

My point: unless there is a way to motivate those private individuals that have installed and continue to use a Linux installation to put their collective hands up to be counted, and they are unlikely too, as one of the likely reasons they are using any Linux distribution is due to privacy concerns, it is near impossible to use just one metric ( .iso downloads) to calculate the installed base.

Then you have another metric: updates.
Some people never update after install - the system is offline and if it ain't broke, why fix it?
Some choose never to update for "reasons", and continue to use it online complete with security risks.
Some people run a private mirror service, and may have many connected systems, but only ever download 1 copy of an update package.
Then there are those that blacklist updates due to hardware related breakage. The system is still installed, running and in use.

So, again, a different metric, a different problem trying to calculate the actual installed base.

[marcpare]

benmc et al,
If we are to put up all kinds of barriers on collecting data rather than just going ahead and collecting it, then the Mageia project will always find itself in a quandry with never having any kind of stats to count on.

IMO, better to just collect the number of downloads from the various mirrors and start collecting proven stats. At least the marketing team would have some kind of indicator as to how well the project is doing ... some kind of barometer. Second guessing as to one download were to be used for a second or third or ... is just guessing. Getting the stats from the mirrors is not guessing, but factual; these are the data points that we should be collecting and working with.

At some point, the Mageia project has to overcome this "hand wringing" over how to interpret possible "guesstimations" of download stats and start working with the real data that can be counted and proven. Perhaps thereafter, some of the leads on the marketing team could make a more reliable guess as to how many estimated downloads/installs could be quoted either publicly or in-house. But, working with real stats should come first, and the membership should be aware of the numbers. Transparency will encourage more people to participate on the project.

[sturmvogel]

If you do a small search, you will find some traces from 13/14 years ago, when Mageia tried to introduce such statistics. Due to lack of manpower, needed filtering of private informations, required change of the Mageia privacy policy and some more points, nothing happened.

As already mentioned, download statistics are kind of (useless) unprecise. They do not reflect if 100 systems where setup from 1 ISO, the ownload was partial, the download ISO was not used, robots and internet archives made a copy, ...you need to find other ways to meassure the usage.

its marketing/publicity team

That made me kind of chuckle.

I started my linux journey with the first Mandrake release, Mandriva and Mageia (up to MGA9). I followed the developement, did bugtracing, worked on the wiki and some more stuff. As long time users who knows from where Mageia came and how it evolved, how the internal works, what the buildsystem is, how the packaging process is, how many active contributers are there (and left over the past years), i can tell you that download statistics won't change anything.

I can't count the amount of reasons, which, after so many years, lead to my decission to no longer actively use MGA on my machines. I have only one physical machine left (from former several) where MGA9 is still installed (resides as secondary OS). I do not want to go to deep into details for my decission, but every other OS has overtaken MGA developement wise, maintenance wise, security wise, feature wise, and active developers wise. As example the exchange of the build hardware won't change anything on the antique building process and packager application process. All modern distributions are way ahead in the packaging process on modern build clusters and infastructures, and easy ways of getting a permanent or part time maintainer.

[marcpare]

Here is an example of ZorinOS that is doing it right with stats. Have a read ... and yes, on its website pages it does commit to privacy.

We should be doing the same, even if the stats are to be kept internally for the Atelier - marketing/publicity/advocacy team.

https://linuxiac.com/zorin-os-18-downlo ... -48-hours/

[doktor5000]

We should be doing the same

WHY should we be doing the same?
It doesn't matter if we would currently have 10 or 1.000.000 downloads per month, no contributors would magically spawn because of a random number that appeared.
You also seem to forget that this feature also needs to be implemented, and we're already starved for developer time ...

You're comparing apples to oranges. And that example you put up - ZorinOS by default also collects the city and country the user is from, so much for privacy. And they're not even making the numbers fully public.
https://www.reddit.com/r/zorinos/commen ... t/idlpedq/