PDA

View Full Version : Homebrew linux market share report



J V
June 18th, 2010, 10:28 PM
I'm not going to say I've found the answer, I'm just going to give a good argument that most estimates of Linux market share are far lower than what is actually the case.

There was also little to no original research :) - just pointing out some things...

Wasn't this already done?
Yes, but so far, none of the data adds up. If you google "Linux market share" almost all of the results state Linux has a 1-2% of the current market share.

These claims are made based on a report, (Ironically, partially aided by Microsoft, see point 2) that showed only 1.24% of web browsers ran from Linux. Indeed, even Wikipedia claims this value is accurate.

There are several reasons that these stats are skewed.


The test was done on western websites. Unfortunately, most Linux users live in non-US, western European countries: India, China, Russia, the parts of Africa that actually have computers, Latin-America (Particularly Brazil)
Certain websites that the test was run from wouldn't be particularly Linux friendly. The list of "Sponsors" flicks through, among other things, microsoft.com, and people only end up there if they work for Microsoft or just got a BSOD

Edit for clarification: Microsoft.com was one of the websites they got their statistics from. An incredible abundance of Microsoft.com traffic would be from Windows (Mainly support requests)
Linux users may mask their operating system/user agent because certain websites don't function with Firefox/Chrome

So how do we calculate Linux market share?
We don't. We can't.

Market share is calculated by dividing the amount of software sales running by the amount of hardware sales.

Manufacture of hardware is disclosed, things like processors and motherboards are an accurate description of the amount of desktop computers available in the world (Though you must make some assumptions that people buy a new computer on an average of so-and-so years)

Manufacture of software is not often disclosed...

Apple discloses it's sale of Macs, but Linux doesn't track it's sales... It doesn't work that way.

Of course, we could calculate the Linux+BSD "Sales" by subtracting Apples sales and Microsoft's sales from the total hardware sales.

The problem with that, Microsoft doesn't disclose the amount of Windows sales. Microsoft has a monopoly on accurate market share stats.

Show me the stats!
So, if Microsoft is the only faction that knows how the market share is, how do we figure it out?

In February 2009, Steve Balmer presented a report on market share.

In this report (Fig. 1), Microsoft, states Linux has around a 4% market share.

The world is stunned.

Now we may not be able to tell what the Linux market share is today from skewed browser tests, but we can sure as hell tell what the trend is! Up up and away!

Between 2004 and 2009, Linux on the desktop tripled.

That's an average 25% increase per year.

That would put Linux on the desktop at about 5.5% now, June 2010...

2023 is truly the year of the Linux desktop with 91% penetration (At least at this rate): It will pass the tipping point in 2018 (29.8%) where it will hopefully get the proverbial ball o' vendor support rolling.

Assuming Microsoft wasn't telling the truth (GASP) we can expect to shave 5 years off those numbers...

Discrepancies
Can we trust Microsoft? History says no. But lets see all the ways Microsoft could skew the data from the conference where this is based on.


Activation: These stats could be taken from number of computers sold, or they could be taken from number of computers phoning home in the last month. Don't know about you but I was a windows user before I switched to Linux
Piracy: Can Microsoft count pirate copies of windows? They can count torrent downloads, and add them to their sales no doubt. But they can't really know how much of that gray area is Linux and how much is Pirated. Unless, you know, they are using one of the 3 different back-doors they installed in every version of Windows.
Extra purchases: Microsoft is disliked for it's policies of making people pay for windows twice at the drop of a hat (Hardware upgrades, bad disc etc) this could also inflate stats.
Microsoft may not have accounted for ancient versions of windows, if it counted windows 95 in those stats it would have brought

All we know now is that Linux will have a majority market share some time in 2020. Hopefully, Microsoft was full of **** and we get it sooner :)

Why should we care?
Because there are no games for Linux!

Ok, that may sound stupid, but you get the principle, until we break the vicious cycle, we won't get vendor support. And we like vendor support (Looks at ATI users)

Sources
The only long-term (http://www.berkeleylug.com/?p=43) report on the Linux trend I could find.
Market share research: (http://marketshare.hitslink.com/Default.aspx) hosted on Microsoft IIS ;)
Microsoft accidentally discloses (http://en.wikipedia.org/wiki/NSAKEY) 3 separate "Keys", one probably belonging to the NSA

mikewhatever
June 18th, 2010, 10:55 PM
What exactly is the problem with a market share stats being hosted of MS IIS, and why do you think these stats are inaccurate?

McRat
June 18th, 2010, 11:24 PM
% of market isn't a good way of looking at things.

Total active seats, a number really hard to quantify, is what would drive support by 3rd parties.

If you have 1,000,000 potential customers, you have far more than developers did in 1980, yet development proceeded anyhow.

Saab might sell very few cars compared to Ford, but that doesn't mean that making seat-covers for Saab models would be a losing business model.

J V
June 18th, 2010, 11:39 PM
@mike: Please read next time:




The test was done on western websites. Unfortunately, most Linux users live in non-US, western European countries: India, China, Russia, the parts of Africa that actually have computers, Latin-America (Particularly Brazil)
Certain websites that the test was run from wouldn't be particularly Linux friendly. The list of "Sponsors" flicks through, among other things, microsoft.com, and people only end up there if they work for Microsoft or just got a BSOD
Linux users may mask their operating system/user agent because certain websites don't function with Firefox/Chrome

@McRat: True, but the companies were much smaller then... Now we have giant monopolies creating all the hardware/software, and if we want hardware support we'd have to get them to support Linux (Most of the drivers were written by corporations, not hobbyists)

These companies don't see the point in doubling their software work for what they see as a mere 1 or 2% profit increase...

McRat
June 18th, 2010, 11:55 PM
@mike: Please read next time:

@McRat: True, but the companies were much smaller then... Now we have giant monopolies creating all the hardware/software, and if we want hardware support we'd have to get them to support Linux (Most of the drivers were written by corporations, not hobbyists)

These companies don't see the point in doubling their software work for what they see as a mere 1 or 2% profit increase...

Even if you make $100 billion a year, if it takes $1 million to make $2 million, they will do it. It's still a million dollars. Unless that is, if that $2 million will **** off a $100 million dollar customer.

J V
June 19th, 2010, 12:18 AM
The point is, the people who have to make that decision, make 50 million a year (About a thousandth of the total)

Someone making 50 million a year wouldn't go to the effort of convincing everyone, hiring new people, doing all the work, just to get an extra 50k a year... It may seem like a lot to you and me but not to the people calling the shots...

But thats not really disputed, otherwise there would be no hardware issues, and lots of Linux games.

There are more windows games than there are packets in your distro repository. Bit of a problem.

ubunterooster
June 19th, 2010, 12:28 AM
Reasoning solid; wow, me like.

BuffaloX
June 19th, 2010, 12:34 AM
We don't know how MS arrived at such a high number, and we don't know what interest MS might have in presenting it.
I suspect that MS knows more about Linux market share than anyone, but I'm not sure if it's in their interest to show accurate numbers.

There are a number of sites tracking OS usage, all show rising market share in a growing market, proving that Linux usage is rising, but by how much exactly is hard to tell, estimates vary from 1% to 4.5%.

Personally I estimate Linux market share to be about 2% of 1.5 billion computers connected to the Internet worldwide, putting Linux at about 30 million Internet users.

This does not include Android, WebOS, Chrome OS, MeGoo.

The stats I follow are these:
http://www.w3schools.com/browsers/browsers_stats.asp
http://marketshare.hitslink.com/operating-system-market-share.aspx?qprid=8
http://www.w3counter.com/globalstats.php
http://gs.statcounter.com/

It seems that the current market share is already tipping the balance a bit, I suspect if we hit 10%, the market response would be dramatic.
But reaching 10% will probably take at least 4 years. The market is gigantic, and it moves extremely slowly.

t0p
June 19th, 2010, 12:42 AM
This is an interesting analysis and I urge the OP to do the necessary research to back up his thesis. If manufacturers could be convinced that present-day Linux "market share" is 5.5% and the tipping point will be reached by 2018, we'd see a lot more "supports Linux" labels on packaging and an explosion of Linux versions of software.

J V
June 19th, 2010, 12:46 AM
Unfortunately, all browser-based shares counters fall prey to problems 1 and 3, and only google would be (partially) exempt to number 2.

As for Microsoft, of course it's not in their interest to show accurate numbers. But then it's definitely not in their interest to over-state Linux market-share, so that makes 4% seem a lot bigger.

@t0p: I'll endeavour to get some more "Hard" facts tomorrow, 2 AM atm :)

Btw, talking about the tipping point that way made me realise, at a certain market share, it will be more beneficial to make Linux drivers than Windows drivers (Cause Windows needs a new version every release) hardware manufacturers would have a better interest in Linux by about 30% (Assuming a 9 year/3 Windows release hardware lifecycle or so)

mikewhatever
June 19th, 2010, 01:00 AM
@mike: Please read next time:

I didn't want to refer to your arguments as they seemed rather week and ungrounded, but if you insist, let's do it.;)

1. The test was done on western websites. Unfortunately, most Linux users live in non-US, western European countries: India, China, Russia, the parts of Africa that actually have computers, Latin-America (Particularly Brazil)

Ho do you know that. Can you show the proof?

2. Certain websites that the test was run from wouldn't be particularly Linux friendly. The list of "Sponsors" flicks through, among other things, microsoft.com, and people only end up there if they work for Microsoft or just got a BSOD

Edit for clarification: Microsoft.com was one of the websites they got their statistics from. An incredible abundance of Microsoft.com traffic would be from Windows (Mainly support requests)

Which ones? The one you linked to, http://www.netmarketshare.com is Linux/Firefox friendly, according to my experience.

3. Linux users may mask their operating system/user agent because certain websites don't function with Firefox/Chrome

And so may Windows users, or really anyone.

undecim
June 19th, 2010, 01:04 AM
We should all start visiting Microsoft.com at least once per day. That way, we skew the stats back into our favor.

ubunterooster
June 19th, 2010, 01:07 AM
We should all start visiting Microsoft.com at least once per day. That way, we skew the stats back into our favor.
That's "intentional alteration of marketing studies"...it does sound fun though.

J V
June 19th, 2010, 01:47 AM
1: First of all, goverments and schools are a prime dissemination of Linux, hidden behind single IPs which diminishes the count, and, notably, the US and all of europe but Germany seem to be ignoring it.

Comes v MS 9644 (http://groklaw.net/pdf/iowa/www.iowaconsumercase.org/011607/9000/PX09644.pdf) refers to "Linux infestation" among government machines
Very very long page of low-IP, high PC goverment desktop linux usage (http://www.linux.org/info/linux_govt.html) (For the purpose of hardware support, business PCs like this do count towards market share)

Top 10 results of "brazil linux" on google are very positive (4 of them say brazil 'loves' linux)

http://www.google.com/search?client=ubuntu&channel=fs&q=brazil+linux&ie=utf-8&oe=utf-8
http://duartes.org/gustavo/blog/post/why-brazil-loves-linux
http://news.cnet.com/Brazils-love-of-Linux/2009-1042_3-6245409.html
http://news.bbc.co.uk/2/hi/business/3445805.stm
http://www.theregister.co.uk/2004/03/05/brazil_loves_linux_true/

"india linux" shows they've had it in goverments since '06, Digg says it was made mandatory in 2007 but the website migrated, can't find the article (767 diggs though)

http://www.linux-watch.com/news/NS6504473525.html
http://www.rediff.com/money/2006/apr/08spec.htm
http://www.businessweek.com/globalbiz/content/sep2006/gb20060921_463452.htm

Same story in russia, talk of installing in all schools, advertising (Countries with economical problems same millions switching their people to linux, so the advertising pays for itsself...)

http://linux.slashdot.org/article.pl?sid=07/09/22/1648209
http://profy.com/2009/08/20/russian-teachers-will-learn-linux-by-2010/

In case you hadn't noticed, north Korea created an official national Linux OS (Red star), south korea reviews are positive (I'd still be worried about spyware though lol)

China has been pushing it hard, but it's sitting at 15% (They actually census them there) cause people prefer to just pirate windows like the rest of the country :)

These numbers are very high some places, very low others... But 10% in countries with hundreds of millions of people (Market share per person or per PC? Interesting question...) gives a significant boost to the global.

2: No, what I mean is, that website has to get it's sources from multiple places on the internet. Microsoft.com is one of those (Or at least was) - most people on microsoft.com aren't linux users. Only very large websites such as google would be valid for this (And shutting out all of china isn't a good idea) - but google wouldn't (To my knowledge) disclose logs on every search they ever got.

3: Windows users would mask their user agent, sure, but why their OS? The only times someone has to mask their OS is when they are trying to access a website that is OS specific, and there are pretty much no Linux specific websites (Would be too hard to build anyway with all the different distros), but quite a lot of windows specific ones... Besides, the user agent switcher masks the user as IE/Windows, so anyone using that for compatibility gives a false positive... For example, there was a 30% linux user spike after the FF2.0 alpha was released... Probably cause user agent stopped working...

Also, wikipedia says IE users can be overestimated because "Extra files like CSS (http://en.wikipedia.org/wiki/Cascading_Style_Sheets) hacks and JavaScript (http://en.wikipedia.org/wiki/JavaScript) hacks are often sent to Internet Explorer."

@undecim: LOL - give microsoft a heart attack when they check their stats :P

Realy gotta go now, 3AM xD

edit: First draft of more "Clean" version below, whadaya think?