Posted Nov 7, 2011 20:22 UTC (Mon) by smoogen (subscriber, #97)
[Link]
I see it usually mentioned via Distrowatch statistics. Not sure how useful those numbers are, but it is the one people quote a lot.
A Linux Mint 12 preview
Posted Nov 7, 2011 20:38 UTC (Mon) by jspaleta (subscriber, #50639)
[Link]
I very much doubt Mint is relying on distrowatch as the basis for either of those claims. First distrowatch doesn't try to scale interest relative to Windows or osX..so there's no way you can use those numbers to claim 4th most popular OS. Second, distrowatch stats cannon be scaled to userbase size. So there's no way to claim millions of users from these rankings.
Now the movement of Mint in distrowatch rankiing is indeed very interesting. But distrowatch is not a trendable metric as it stands. The distrowatch numbers are very easily misinterpreted because of the way distrowatch has chosen to represent the average over a long timescale.
For example. Right now Mint is leading on the 6month 3month 1month and 7day averages. This does _NOT_ mean that 6 months ago or 3 months ago or even 1 month ago that Mint was dominant. 3 weeks ago Mint was not leading those long time averages. There is a significant spike in interest ongoing for the last couple of weeks and its big enough to impact even the 6 month average.
Spikes in interest will retroactively impact all the other timescale estimates. People continually misinterpret what those numbers mean. What Distrowatch really needs to do is to take their 7 day average and then provide an historic trending plot of that. There is far more information in the historic trend of the 7day average than the continually updated 6month average.
And even then, distrowatch's 7day average is at best a measure of "interest" or "buzz". Once you are a user of distribution you are less likely to click through a distrowatch page to read about the distro you are already using. People still looking for something different than what they are using are more likely to click on competitors. This makes it very difficult to use distrowatch as a userbase estimate (and they admit as much in their writeup of their stats). It is at best an estimate of "buzz". What is fascinating to me is that distrowatch as a "buzz" metric is completely uncorrelated with google trends metric. The current uptick
in Mint's distrowatch interest level is not mimicked in the Google trends data for "Mint", even though Google trends is itself a different way to measure "buzz."
-jef
A Linux Mint 12 preview
Posted Nov 7, 2011 22:18 UTC (Mon) by Topaz (guest, #60130)
[Link]
Distrowatch numbers are also easy to manipulate.
A Linux Mint 12 preview
Posted Nov 7, 2011 22:21 UTC (Mon) by misc (subscriber, #73730)
[Link]
That's easy :
1st one is Windows, 2nd one is Mac OS X, 3rd one is Ubuntu and they are almost as popular as Ubuntu according to distrowatch, so they are 4th.
While that's a joke from me and I am in no way affiliated to linux mint and completely clueless on that point, I strongly suspect that's the real explanation.
Counting Windows as a single OS would be false IMHO, so I suspect that the 3 first place would me microsoft products ( xp, 7, and 2k8 ). That let OS X as 4th, and likely 5th, or vista. Then we can start to find various Linux distribution and based on my own experience at the lug and various events, Linux Mint is no in the top 5.
What amaze me is the number of comment on the blog, and the fact that no one question neither the stats, nor the whole business plan and the lack of detailed financial report ( cause 3500$ a month is enough to pay one person full time, and that's not negligible ).
A Linux Mint 12 preview
Posted Nov 7, 2011 23:26 UTC (Mon) by Lennie (subscriber, #49641)
[Link]
"1st one is Windows, 2nd one is Mac OS X, 3rd one is Ubuntu and they are almost as popular as Ubuntu according to distrowatch, so they are 4th."
That is exactly what they are thinking.
I don't know what the browser string of the browsers on Linux Mint looks like, but it doesn't show up on these reports:
I guess as Linux Mint has a distro based on Ubuntu and one on Debian it probably has Ubuntu or Debian in the string as well.
So we can't compare.
A Linux Mint 12 preview
Posted Nov 7, 2011 23:43 UTC (Mon) by jspaleta (subscriber, #50639)
[Link]
User Agent string is no longer a valid way to compare distros at all. Neither Chrome nor Firefox (circa Firefox 7) list the linux distribution by name any longer in the user agent string. The list "linux" and some cpu information now. I've personally made the wikimedia stat script maintainer aware of this change in the last couple of months.
Any linux distribution shipping officially branded firefox by default and conforming to the constraints that Mozilla places on trademark usage will no longer be differentiated in these stats. It's already started to impact the Ubuntu stats significantly in the last couple of months, There is an accelerated downward trend in "Ubuntu" over the span of several months that can only be reasonably accounted for by this user agent string policy change. I fear that people will misattribute the drop in Ubuntu wikimedia user agents stats I expect to see after the 12.04 LTS release as a reduction in Ubuntu usage. And that would be a real shame.
-jef
A Linux Mint 12 preview
Posted Nov 7, 2011 23:48 UTC (Mon) by Lennie (subscriber, #49641)
[Link]
I guess that means distrowatch will in the future also not be able to differential. So those are also useless.
A Linux Mint 12 preview
Posted Nov 7, 2011 23:57 UTC (Mon) by jspaleta (subscriber, #50639)
[Link]
Distrowatch doesn't use client user agent strings. Distrowatch makes no effort as far as I am aware to tell what you are currently running as you view the site. It's strictly about which pages on their site are viewed by the most ip addresses every day.