Analysis of Distrowatch data

General Discussion
1

There are two types of data available from Distrowatch

  • Hits per Day , as seen on the home screen
    https://distrowatch.com/
  • Voluntary ratings of distros , as seen in this table
    https://distrowatch.com/dwres.php?resource=ranking&sort=votes

I have attempted to analyse these data to see if we could learn anything from how the hits and ratings are distributed across families of distros.

The result is disappointing. I cant really see anything of interest. Nothing relates to families of distros, and ‘Hits per Day’ and ‘Total Rating’ seem to be the same thing.

The only conclusions of value seem to be

  • whatever is being measured by these data, distro family has nothing to do with it.
  • you may as well only look at hits per day, because ratings seem to measure the same thing.

I did document it. I will not copy it here , because it is long and not properly polished. It is available as a .md file ( ie markdown) on my github repo

https://github.com/nevillejackson/Unix/blob/main/Distrowatch/dwatch.md

If you are interested in using R for graphics, it may be worth a look.

5 Likes
2

The distro maintainers could count the daily downloads of their distro and give it to distrowatch that would be interesting

1 Like
3

Sourceforge counts downloads… for example I can see that those spins of antiX I made with various DE’s have a total of 163 downloads. That is tiny… Debian would have thousands.
Yes, a comprehensive daily download count would be more indicative of real usage of a distro.

2 Likes
4

The really interesting thing is not where all those distro hoppers have been but where they finally stay.

1 Like
5

Yes. Even downloads will not tell you that.
A survey maybe?

1 Like
6

Unfortunately a survey requires participation, and this could be a problem.

1 Like
7

Yes it would be like political polls … not a random sample of users.
If linux gets to be so important that it is included in the census, then we might get an unbiased view of who uses each distro.

1 Like
8

Thanks for posting this. I do know R and will look at this. It’s a good idea to give weight to ratings because the assumption would be that the reviewer took the time to use the distribution (as opposed to visits/clicks on a page).

3 Likes
9

I am a distro sampler (I’d rather call myself), and if Firefox locks up with YouTube, like some arch flavors do, I go back to Kubuntu.

1 Like
10

I do not understand why that fault would only occur is Arch based distros?
I know that some distros do configure firefox in a different manner , and they also may vary in what libraries are present by default. … that is all I can guess at.
It is the same base firefox in all distros.

2 Likes
11

I did a web search and could not find a answer. But when I took the -AI off Gemini jumped in with what looked like a reasonable answer. I won’t post all the details but here was the core thought.

No, Firefox is not completely the same across all Linux distributions and 
Windows.  While the core web-browsing engine and features are identical, there
 are  significant differences in how the browser is packaged, optimized, and 
integrated  into the operating system.
3 Likes
12

This matches with my observations. AFAIK, Linux Mint has tampered its FF with whatever.

1 Like
13

My impression is most distros fiddle with the config of packages … especially with the gui appearance to make it match their theme.
but
I have no evidence that anyone fiddles the source code… it would be a dangerous thing to do , given the complexity of a package like FF.
so
you should always be able to set FF back to its generic form… like you would get if you installed it from source code.

In Gentoo you might set flags to include or exclude some features. If you compiled from source in Mint, you could do the same by setting compiler and loader options, but that requires some knowledge.

2 Likes
14

Only thing I can see different is the home page on ff and chromium points at the mint home page and the default search engine. But i could be wrong.

1 Like