Looking at his methodology

Story: New site ranks Linux and BSD popularityTotal Replies: 7
Author Content

Feb 15, 2013
11:14 AM EDT
Looking at his methodology it's hard to see how his rankings are any more valid than, say, the DistroWatch page hit rankings. He's attempting to collect web data which, by definition, grossly undercounts enterprise machines, both servers and desktops, which are often behind firewalls and may not even have internet access. Where I work a large percentage of desktop users are intranet only. In addition, he's chosen 20 distros that he feels are popular and left off others that, IMHO, may outrank some on his list. Where's Mageia, for example? Some of his numbers make sense to me, some just don't. Count me as unconvinced, at least so far. Maybe, as he works on refining his methodology and expanding his list the results will be more convincing.

Feb 15, 2013
2:23 PM EDT
Mageia's there. He's got it at #18, which I believe is way too low. DistroWatch has it at #2 for the last 6 months. Anyway, I agree his metrics don't make much sense.

Feb 15, 2013
5:33 PM EDT
Thank you very much for the feedback, guys. Rest assured, I'm not purely counting web data. As I state in my "how I'm doing it" page, there are a number of data sources used, including netcraft data. Yes, a lot of the data is about networked machines, but not all.

Whereas distrowatch is merely a "who clicked on me today" type of ranking, I believe my method will be more accurate over time. I'm absolutely sure there are bugs to be worked out in my methodology, but I'm very determined in my attempt to rank them accurately and fairly. For example, RHEL. I'm showing it's very popular, which includes a lot of enterprise machines behind firewalls, for sure.

My ranking resists the volatile bumps that the flavor-of-the-week might get, like Mageia (no offense to those guys). They just haven't reached the market penetration or popularity threshold that an Ubuntu or Red Hat has, and my list reflects that.

So far, I'm fairly happy with the results but it's a work in progress. Thanks again for your honest feedback.

Feb 15, 2013
5:44 PM EDT
Quoting:n addition, he's chosen 20 distros that he feels are popular and left off others...

The way I read it, it appears to me that based on the algorythm he's using those are the top 20, not so much his choice as what the results were.

based on my personal experience (a pretty poor metric to be sure) those results feel about right.

Feb 15, 2013
6:37 PM EDT
Quoting:those results feel about right.

Same here and I agree with @TA. I believe, compared to DistroWatch, the various major sources of data will give much more accurate results.

One thing would be of interest to some is to also publish the data compiled.


Feb 15, 2013
7:05 PM EDT
This seems a whole lot more solid than Distrowatch, not that Distrowatch claims to show anything more than what Distrowatch visitors click on when they visit that site.

Feb 21, 2013
11:53 AM EDT
distrorank: do your datasources include http://linuxcounter.net/distributions/stats.html ?

greetings, eMBee.

Feb 21, 2013
12:06 PM EDT
Greetings back to you, eMBee. Good question - yes, I do take the linuxcounter stats into consideration.

Posting in this forum is limited to members of the group: [ForumMods, SITEADMINS, MEMBERS.]

Becoming a member of LXer is easy and free. Join Us!