Louigi Verona's Workshop


Linux Adoption on the Desktop and the myth about Linux security

Louigi Verona, July 2016

Researching material for my upcoming article “On the justification of the four freedoms”, which challenges the validity of many of Richard Stallman’s writings, allows me to see just how many bad arguments are leveled to advance notions about Linux-based systems that on closer inspection turn out to be questionable.

In the Wikipedia article on Linux adoption, "free" software activists challenge numbers of 1-2% adoption, obtained through browser counters, claiming that they are not accurate enough.

What is their counter-claim? Carla Schroder of Linux Today is being quoted as saying:

"I don't believe that the desktop Linux market share is barely 1%. I think it is a lot higher. I have no good data to share; I base my assessment on experience and knowing the industry. There is something else that is even more persuasive, and that is how Microsoft behaves. If Linux is so insignificant, why do they pay so much attention to it?"Tiny 1% Linux = Big Microsoft Terror

A somewhat honest admittance of lack of evidence, mixed in-between two fallacies: the argument from personal experience and a "more persuasive" informal fallacy called "appeal to motive".

End result: she has no evidence. She simply wants to believe desktop Linux share to be higher.

Caitlyn Martin is then quoted to say that 1% number is bogus, since it relies on web counters, and those are unreliable because "they generally only include websites that have paid to be counted. That pretty much guarantees that Windows will be overcounted ... ...most Linux and technical websites do not pay to be counted by the web counter companies, which in turn skews numbers wildly in favor of Windows."

She offers no evidence for this. The argument itself is sloppy anyway. It assumes that Linux users visit mostly "Linux and technical websites". This assumption is questionable and requires statistical evidence in order to be believed.

And if we do take that assumption onboard, then it can probably be used to disprove significantly higher claims to Linux adoption, as these sites are likely to have very modest amounts of traffic. As of the moment of writing, Linux Journal holds Alexa Rating of 36,172. And this is one of the most visited Linux websites, most others having ranking beyond 300,000.

But neither does she explain why we have to be focusing on counter compaines that have to be paid. There are notable web counters that don't have to be paid for.

One of the examples is a Wikimedia counter. With Wikipedia having an Alexa Ranking of 6, it can be reasonably argued that most Internet users visit Wikipedia regularly. Their counter, which has been collecting data up to 2015, shows Linux Desktop share well below 2%. This number is fully in line with other widely used counters, the comparison can be examined here. Note, that on the page of the Wikimedia report Linux is being estimated at 16.36%, but at least 15.01% of it is Android, with an additional 0,84% being "Linux Other", leaving desktop systems even below 1%.

In the comparison cited above, W3Counter shows Linux desktop share of 2.44%, highest among other counters. As of the moment of writing, their updated Linux desktop share is 2.48%.

So, although we can and should acknowledge that this method of estimating operating system share is far from perfect, it definitely cannot be considered completely unreliable, and the numbers pertaining to Linux adoption and coming from numerous independent sources seem to be in good enough agreement with each other.

Caitlyn Martin then says:
"If we talk about actual usage there really is no way to get an accurate measure. Educated guesswork probably puts Linux at close to 10%, just about even with MacOS."

Unfortunately, she does not expand on her "educated guesswork", as this statement happens to be the last statement in her blog post. Either she prefers to keep her reasoning a secret or this guesswork is simply a gut feeling based on no objective data.

What's interesting about this is that "free" software activists claim 1-2% number to be thrown around for no good reason. But web counters, although far from perfect, give us at least some data. In contrast, the 10% number seems to be arrived at through wishful thinking and mysterious "educated guesswork". In other words, 10% is the number that is based on less data than the number based on web counters!

Another Linux feature, which is revered by many Linux users, is its lack of malware. Although it is clear that Unix systems (including OS X) benefit from their structure and safer method of software delivery, adoption plays a huge role in the matter. Arguing against it in the Wikipedia article is a blogger, an "experienced Linux system administrator" Rick Moen. According to him:

"[That argument] ignores Unix's dominance in a number of non-desktop specialties, including Web servers and scientific workstations. A virus/trojan/worm author who successfully targeted specifically Apache httpd Linux/x86 Web servers would both have an extremely target-rich environment and instantly earn lasting fame, and yet it doesn't happen."Rick's Rants

But of course, this is not true. Attacks on Apache do happen and happen all the time. Secondly, web servers are environments with tightened security. It is entirely possible that Windows servers also have good statistics at preventing viruses, trojans and worms from getting inside. We, unfortunately, could not find any such statistics neither on Apache nor on Windows Server. Statistics on security vulnerabilities are a different matter and are not directly relevant, but several old comparisons rated linux server software to be less prone to critical bugs than Windows, but again - no recent data.

Usually the context is the desktop environment. And this is where the argument from adoption actually does work and seems to be supported by the steady rate of growth of Linux malware.

In conclusion, current level of dialogue on the topic is still very primitive and riddled with too many bad arguments. If there is a case for Linux to be made, it has to be made with proper evidence, not with logical fallacies and personal opinions.