Lies, Damn Lies and Linux Market Share Statistics
Linux market share ... how does one even attempt to measure such a figure? Commercial software can be tabulated by units sold, but when so many copies of Linux operating systems are tossed around the Net for free, who's counting? A few outfits tried recently, and the results caused quite the commotion.
05/11/09 4:00 AM PT
Market share, market share, what's Linux's true market share? That, in essence, has been the question du jour on the Linux blogs in recent days.
It all started when NetApplications' Hitslink.com released some statistics for April indicating that Linux just passed 1 percent for the first time.
Around the same time, however, W3Counter published figures for the same month indicating that it had just passed 2 percent.
Many FOSS aficionados, meanwhile, argue that it could be 6 percent or higher.
The result? You guesstimated it: nothing short of chaos and confusion.
'100 Percent Dishonesty'
Bloggers at Linux Today, Slashdot (not just once, but twice), Computerworld, Digg ( one, two and three times), LXer ( once and then again) and the Linux Loop jumped on the topic in short order, leaving a trail of broken pencils, empty coffee cups and smoking keyboards in their wake.
"1% Linux Market Share = 100% Dishonesty" was the title of Carla Schroder's post on Linux Today, for example, placing much of the blame on the tech journalists who report on such studies. (We're innocent, we swear!)
Then again: "Depending on your outlook, this figure is cause for cautious celebration, or equally cautious dismissal of the operating system's prospects," wrote Bruce Byfield on Datamation. "Alternatively, you might prefer -- as I do -- to question the statistic's accuracy, and look at other ways to estimate GNU/Linux's presence."
'I Can't Wait!'
Somewhat less seriously: "THIS IS TRULY THE YEAR OF THE LINUX DESKTOP!" quipped redaphid on Digg.
An alternative view: "I can't wait! At this rate, 2024 will be the year of Linux on the Desktop!" Slashdot's BadAnalogyGuy quipped back.
Is it all just a case of lies, damn lies and statistics? Or is there some substance in these conflicting reports? We here at LinuxInsider felt compelled to investigate.
"Measuring market share of open source software is extremely difficult," Chris Travers, a Slashdot blogger who works on the LedgerSMB project, told LinuxInsider.
"The basic problem is that one can use sales data as a close proxy for market share when one is selling a tangible and restricted resource, but for something like Linux, actual product sales probably account for a very small portion of installed systems," Travers explained. "In the end, it is reasonably impossible to estimate market share in this area with any accuracy. I don't think anyone has a solid idea of what the actual Linux desktop market share is."
The Trouble With Being Free
"Tracking Linux is difficult in the current climate," Montreal consultant and Slashdot blogger Gerhard Mack concurred.
"My current laptop came with XP simply because I got a better deal on the hardware; how can Dell know I opened it up and swapped the Broadcom wireless adapter for a more Linux-friendly Atheros that I got on eBay and reformatted?" he explained. "They can't. Unless someone wants to make an automated Linux counter for the distros to install, we're out of luck. And even then it will count low thanks to Linux having a large number of privacy-conscious users who will disable that feature if it exists."
Indeed, while Linux does have some sales data for evidence of market share, the fact that it's generally free makes tracking it difficult, Slashdot blogger yagu told LinuxInsider.
Consider, for example, how many Linux servers are running without management's knowledge, he pointed out -- "I know they're out there; I've worked in places where this is humorously so," he said.
'No Way to Figure the Numbers'
Also relevant -- but difficult to know -- are how many Linux servers are the primary use for a Microsoft host -- "Windows merely serving as a surrogate host for virtual machines," he noted.
Then, too, "how many Linux instances would there be if those who could were permitted?" he asked. "How artificially limited is Linux's market share because of Microsoft's (allegedly) unfair business tactics?"
Yagu breaks down the percentages as follows:
- "For general market percentages by sales and 'instances of,' I would guess Linux is probably, by the numbers, right around 1 or 2 percent. When Linux is free, there's no way to figure the numbers and, even when you do, people ignore them."
- "For live real-world, mission-critical use, I would guess Linux is in the 10 to 30 percent range. I see lots of .Net and Microsoft applications, yes, but when I see and talk with people about their mission-critical applications, it's very likely some Unix variant, with Linux increasing its Unix-market share steadily."
'It Doesn't Really Matter'
Whether the actual numbers even matter is another question, Mack asserted. "Linux has finally got the base OS (operating system) roughly polished, and now we need to spend more time cleaning up the apps and writing software that matches the needs of new markets," he said.
Slashdot blogger hairyfeet agreed that the numbers don't matter -- but for different reasons.
"Whether it is 1 percent or 2 percent or 5 percent doesn't really matter, because for most of the world Linux simply doesn't exist," hairyfeet told LinuxInsider by email. "Sorry, but it's true. And while Ubuntu has made great strides, there is MUCH more work to be done."
'Free as in Worthless'
A good example of how far Linux has to go to be accepted is Vista, he asserted. "That OS is truly hated. Even when the totally awful WinME was being sold, I didn't have as many customers beating down my door to get rid of it," he noted. "At this moment I have a dead Vista box sitting in the corner, and the customer said, 'If you can't get XP to work on it then toss it' -- that's how much Vista is hated."
Given that, why hasn't Linux shot through the roof in the meantime? "While it works fine on enterprise-class hardware -- the ones being paid to develop Linux work for large corps that care about enterprise and server -- on consumer hardware, frankly, it is awful."
Most consumer hardware in stores "simply will not work at all out of the box with Linux, and if you do manage to get it working after hours or even days of majorly complex CLI commands, it simply won't work half as good or have even half the functions that it would in Windows," hairyfeet added. "I can tell you from 15 years of experience in PC retail the customers don't care WHY it doesn't work, they simply know it is 'broken' and therefore your 'free as in beer' OS quickly becomes 'free as in worthless' to them."
'Unrepresentative of Reality'
Yet whatever number is put to Linux's market share, for some, the OS itself -- like the proverbial rose -- will always smell as sweet.
"The numbers from NetApplications are clearly unrepresentative of reality," blogger Robert Pogson told LinuxInsider via email. "Around 2003/4, IDC determined by survey that GNU/Linux was ahead of Mac OS at about 3 percent. Since then GNU/Linux has had growth numbers from 20 to 50 percent in various places.
"That would put GNU/Linux at 7 to 9 percent," he asserted. "Apple publishes its numbers for production of units in reports to the SEC and they are around 3 percent of PC production."
'Around 10 Percent'
Numbers are likely to be much higher in places like Russia, India, China and Brazil -- "where government actively promotes the use of GNU/Linux," Pogson added -- but "NetApplications samples very little outside USA/Europe."
In short, "I believe 2009 is the year of GNU/Linux and the share is around 10 percent," Pogson opined.
Of course, "the only thing of which we are sure is that GNU/Linux is growing rapidly," he concluded. "All the stats, even from NetApplications, show that."
Microsoft knows, too -- "that is why they cut prices in China and on netbooks," he said. "They can no longer afford to charge what they want because the world knows there is a good alternative. M$ cannot put the genie back in the bottle."