A Measure of the Bias of NetApplications Against GNU/Linux

Another month approaches and we expect to read once again how GNU/Linux is not making it on the desktop as evidenced by NetApplications biased web-sampling:
For the week of April 1, 2012:

Region Share (%)
USA 1.28
USA without Google HQ 0.62

How do they shift the data for a whole country with a few thousand users in California? They count business-use more than anything. Google happens to be a business. All those schools, government offices and individuals count for nothing with NetApplications. Google no more doubles usage of GNU/Linux in USA than they cut usage of that other OS by 5%.

Maybe it’s time people cut out repetition of the “1%” lie.

UPDATE World – (World – Mountain View – Sunnyvale) = 0.95% – 0.89% in April 2012, so the folks at Google are supposedly 6% of the world’s users of GNU/Linux… Yeah, right. Do the maths. 10K/109 is 0.00001 %.

About Robert Pogson

I am a retired teacher in Canada. I taught in the subject areas where I have worked for almost forty years: maths, physics, chemistry and computers. I love hunting, fishing, picking berries and mushrooms, too.
This entry was posted in technology. Bookmark the permalink.

7 Responses to A Measure of the Bias of NetApplications Against GNU/Linux

  1. Clarence Moon says:

    Like we know 25 percent of servers sold by revenue was Linux

    Apparently reading reports is not your long suit, Mr. Oiaohm. Linux share is nowhere near that high.

  2. oiaohm says:

    Clarence Moon this the problem most of the numbers people throw up are horribly bias.

    Like we know 25 percent of servers sold by revenue was Linux. We now 45 percent were windows by revenue.

    But really how many are running Linux or Windows the numbers don’t say from IDC.

    The we get to web numbers and we know those a bias.

    The reality of the problem the 1 percent claim of Linux Market share is very poorly based.

    Same with the numbers of windows or Linux servers.

    Yes there is a chance that 25 percent of revenue for Linux could be the same number as 45 percent revenue of windows.

    With web numbers it could be the sites monitored Linux people don’t visit and windows users visit a lot.

    Really we need some real objective numbers.

    Mobile phones and mobile network connected devices are simple to count because service providers report how many of each phone is connected to there network. They also know if they are duel sim and reporting twice because the phone ID number is unique.

    So one section of the market has perfect numbers.

  3. Clarence Moon says:

    My data clearly show that NetApplications is counting client PCs

    They don’t count PCs at all. They count visits to web sites. They only use a small fraction of the total number of sites that exist. If they seem to exclude schools, it is likely that the students are not very interested in the commercial sites that Net Applications uses in their visitor counting. Which seems reasonable since most of Net Applications’ client sites are business sites that would likely have little or no appeal to students.

  4. Clarence Moon wrote, “Net Applications only counts visitors to sites where they have sold the site owner on paying Net Applications to count visitors.”

    Nope. My data clearly show that NetApplications is counting client PCs from businesses and not client PCs from government and education. If not their numbers for GNU/Linux would be many times higher. How else do you explain that Google, alone, doubles the count for California? We know how many employees, roughly, Google has at HQ and their “unique visits” should not have that effect but they do because NetApplications gives them preference either counting from “business” domains or during office hours or some combination of that. California has 30 million people. Why do ~10K people count so much? Google should just be a tiny blip, not a doubling. There are school divisions in California with 100K GNU/Linux PCs. Why don’t they count?

  5. Clarence Moon says:

    NetApplications has a terribly biased sample counting only a fraction of PCs.

    Net Applications only counts visitors to sites where they have sold the site owner on paying Net Applications to count visitors. That is not Google, MSN, Amazon, eBay, or many other sites that do not subscribe to their counting services. Furthermore it is not Facebook or other social media sites popular with mobile users. Thus the demographics from Net Applications are very limited to a small slice of user types.

  6. Viktor rudely wrote, “You don’t know what NetApplications counts. And you’ve just proven it. Again. Go to their website and get a clue.”

    see NetApplications FAQ
    “What is a daily unique visitor, and why do you count those instead of pageviews?
    Net Market Share data is an aggregation the traffic of all of our HitsLink clients, but instead of counting pageviews we count daily unique visitors. A daily unique visitor is counted only once per day per website we track, regardless of the number of pageviews the visitor has. While this may seem to greatly reduce our sample size from the billions of monthly pageviews we process to only the daily unique visitors, we do so to provide a more accurate picture of market share showing the number of users of a technology instead of the number of clicks. Counting unique visitors also renders bots designed to influence market share harmless. Counting pageviews for market share reports would be susceptible to bot attacks and inexplicable jumps in market share that don’t represent the true nature of the market. “

    They count unique visitors which is people/PCs depending on how many people use one PC. The bias in favour of GNU/Linux because it’s a business with Google is in favour of that other OS because most businesses use that other OS. They clearly don’t count the school divisions in California which outnumber Google’s PCs or users. They clearly don’t count the government offices that use GNU/Linux. The “1%” figure clearly does not apply to all use of PCs. No way. Yet trolls here claim GNU/Linux has ~1% share of PCs when we know of huge roll-outs globally. This year, Russian government began switching offices and schools to GNU/Linux. Not a blip. Same with Munich with 10K PCs. NetApplications has a terribly biased sample counting only a fraction of PCs.

  7. Viktor says:

    You don’t know what NetApplications counts. And you’ve just proven it. Again. Go to their website and get a clue.

Leave a Reply