In short Jason Evangelho on Forbes states Linux usage is rising, Brian Lunduke claims it is actually falling. I quickly want to mention that neither of them is right, nor are they accurate.

Evangelho's Bad Data

Article

Here Mr. Evangelho uses data from a site that is worse than Distrowatch. NetMarketShare. While Lunduke, who will be linked later, gives a solid couple of ideas as to why the main issue is simple. Ubuntu 20.04 (+ remixes and flavors) released in April. Same time, Fedora 32 launched. Two of the largest distributions releasing at the same time will give bloated data here, where NetMarketShare used idiotic measuring methods.

Lunduke's Bad Data

Article

Here Mr. Lunduke uses Statcounter - which yes states a fall over the year from 1.9% to 1.6% - note that Lunduke's article focuses on only 2020 - which isn't over and a lot of the data is still iffy at times. We have seen lower usage since March according to Statcounter. Statcounter also - when doing a measurement of the past 3 years - has seen no real long-term improvement. We do randomly see random dips as well in Statcounter - however not as extreme as Evangelho's.

The core issues

• OS Market Share is impossible to represent.
• Multiple people use singular installs sometimes.
• Single persons use multiple installs sometimes.
• Many users duel-boot
• Many modify or just remove trackers on installs (mostly on Windows and Mac)

Here is why this is a problem. The data these two argue overall is no better than Distrowatch, which puts a relatively unused distribution at number one. Very much do note that with this in mind, that we have no real data on what is what these two articles are both false.

They cannot be proven true, so they have to be classified as false. Why? Well, how do we calculate market share on something that doesn't track users easily? We can't. Seriously.

Lunduke and Evangelho focus on incomplete and misrepresentative data. Both very much also misrepresenting the data they use.

How it could be possible to guesstimate Linux usage.

While we can never be 100% sure, we can try a couple of things. Some of which are checked by distributions.

• Unique IP downloads
  • Using a package manager, see what IPs are unique and which aren't. This can be negative but overall is used in many distributions. While the IP is collected, most distributions - mainly excluding Zorin - are transparent of the use and are opt-in only.
• Download counts
  • This is a bit less accurate because some people download purely to test, however, it is a possibility.
• Online Account Logins
  • This is not great, as it is done by Google - most likely - and they aren't great with privacy. However, this can help pinpoint usage.

Other than these, there is no solid way to see the Linux market share. So I really dislike Jason and Brian's articles, as the data is most likely inaccurate. Especially with some people taking privacy beyond the next level. Many also make their own distributions, run their distributions in a VM or Docker (even long-term installs), and duel boot.

TL;DR - There are too many factors to look at, plus Linux goes up and down, overall not being measurable by data - as even backpedaling Windows 10 data compared to Linux is still difficult and will never really be accurate. Linux isn't falling or rising. As far as we do know - it's stagnant as always.