Talk:Usage share of operating systems/Archive 5

This is an archive of past discussions. Do not edit the contents of this page. If you wish to start a new discussion or revive an old one, please do so on the current talk page.

Archive 1

←

Archive 3

Archive 4

Archive 5

Archive 6

Archive 7

The article has this in the beginning:

This article possibly contains original research. Please improve it by verifying the claims made and adding inline citations. Statements consisting only of original research should be removed. (November 2011) (Learn how and when to remove this message)

AFAICS, the only OR was the Median row, now removed, so maybe we should remove it? - 195.23.92.74 (talk) 11:59, 30 November 2011 (UTC)

Yes definitely. Also Useerup had a graphic somewhere that could be used as a basis for replacing the graphic that was removed, dunno where it went to.

Dmcq (talk

) 12:29, 30 November 2011 (UTC)

Actually more severe OR is multiplying total numbers from one source on desktop/mobile split number from another source. This is definitely improper synthesis and OR, so the tag should stay until that issue is addressed.Wikiolap (talk) 03:22, 1 December 2011 (UTC)

All other similar articles like

) 16:52, 23 December 2011 (UTC)

What do you mean by 'useful' please? Sorry about my paranoia but I've seen people using the excuse of making the articles more useful as a reason to stick in the medians and diagrams using it whereas

Dmcq (talk

) 19:51, 23 December 2011 (UTC)

On what grounds are you assuming any of these things ? Where in the paragraph which i wrote did i mention median or any other things that you have mentioned. I was saying that these articles need general cleanup, figure updates and general maintenance. No need to put any

) 20:44, 23 December 2011 (UTC)

I didn't assume anything. I asked what you meant by 'They need to be made more useful' and said why I was concerned. If you search for 'useful' in this talk page you'll see it is sometimes used as a reason to justify actions and I've been through various requests for comment polls dispute resolution and an aborted mediation on related pages where it kept being bandied around as a reason for doing things I believe are against policy. Anyway thanks for the answer and I fully agree with making things useful if possible, just any more than a small bending of policy isn't what I count as a possible path. Sorry I annoyed you, I should just ignored any worries until something actually happens.

Dmcq (talk

) 22:59, 23 December 2011 (UTC)

It's okay, i truly understand the tough situation you have been through these heated discussions and strong debates related to the 'median' topic here and on other related pages. Misunderstandings can happen, it's alright. But nonetheless thank you very much for your support in helping me improving Wikipedia :). TheGeneralUser (talk) 01:21, 24 December 2011 (UTC)

The tablet section only talks about iPads and Androids. What about Windows Tablets? Also, should there be a section for e-readers?Cavebear42 (talk) 02:02, 25 January 2012 (UTC)

Done. Wikiolap (talk) 17:39, 26 January 2012 (UTC)

Sweet, Thanks. Cavebear42 (talk) 21:03, 10 February 2012 (UTC)

The tablet section only talks about iPads and Androids. What about Windows Tablets? Also, should there be a section for e-readers?Cavebear42 (talk) 02:02, 25 January 2012 (UTC)

Done. Wikiolap (talk) 17:39, 26 January 2012 (UTC)

Sweet, Thanks. Cavebear42 (talk) 21:03, 10 February 2012 (UTC)

Not a major item, but the term usage share is defined in the lede as marketshare perhaps because that is a better term. I suggest a move to "market share of operating systems". History2007 (talk) 22:12, 8 February 2012 (UTC)

All numbers within the table have the % sign attached to them. Highly redundant. And it spoils cut and pasting into a spread sheet.
This means I cannot easily get the median of a column (yes, I know what a median is and why and when I want it) and do it by hand. --77.1.11.143 (talk) 20:18, 4 March 2012 (UTC)

also, while at it, please have numbers right aligned --77.1.11.143 (talk) 20:25, 4 March 2012 (UTC)

This article starts with:

According to the linked discussion page, in particular to this comment, it is my interpretation that the issue is now gone, and as such we should get rid of this notice. Comments? 195.23.92.74 (talk) 19:38, 19 March 2012 (UTC)

How are you sure that those os named linux in data source are gnu/linux?

In fact i use linux but mainly without gnu (i replaced nearly every common gnu part with free (=non-GPL) versions from freebsd and other sources) and so possibly did others. And there are several os != Android using linux but not beeing gnu/linux, often found on embedded systems. — Preceding unsigned comment added by 131.220.197.210 (talk) 12:00, 15 May 2012 (UTC)

Agreed. Perhaps "gratis unix" would be better. — Preceding unsigned comment added by 216.254.210.163 (talk) 22:12, 12 July 2012 (UTC)

Then you use heavily modified FreeBSD with Linux kernel. — Dmitrij D. Czarkoff (talk) 08:36, 13 July 2012 (UTC)

The graph of OS's that access wikimedia is totally inappropriate for this page and should be removed immediately.

The page is about usage share, not usage share on wikimedia. It is grossly misleading to show that data when the article is written in a global (i.e. non-wikimedia) context.

I disagree on the following grounds: "Operating system statistics on Wikimedia" is clearly labeled as such. It is on topic. Additionally the paragraph next to it puts it into perspective "[..] difficult to obtain, since [..]". The reader (in the sense of reading) won't get this wrong. Non readers are not the target audience of an encyclopedia. --HeWhoMowedTheLawn (talk) 18:43, 25 August 2012 (UTC)

I disagree entirely with HeWhoMowedTheLawn. We're on a page about operating system statistics and instead it references wikimedia. If there is a wikimedia usage share page, put the information there. Otherwise I don't see how this is anything other than commonplace bias being created as the statistics presented do not reflect on usage share but instead reflect on wikimedia operating system usage share. Edit: also for what purpose or benefit does it somehow make sense to have this specific usage graph at the top of the article? That in itself makes me question the objectiveness of acting as if this is a neutral statistic. Add in addition that wikimedia's usage is in sharp contrast to other similar statistic results from netapplications and other statistical analyis. — Preceding unsigned comment added by 12.32.89.121 (talk) 20:04, 7 September 2012 (UTC)

This page is about usage share of operating system, so one would suppose to get a clue about operating systems usage share when reading it, so graphics in the most visible place is definitely appropriate and required. The choice of Wikimedia stats (as the most neutral and credible source, as well as the source related to Wikipedia) is definitely the most appropriate. This wrongheaded thing rises up nearly every month, so it is probably a good time to place a banner that this question was already discussed million times, which resulted to this chart. — Dmitrij D. Czarkoff (talk•track) 21:07, 7 September 2012 (UTC)

Why would one set of servers counting requests over the web be less reliable than another ? Since both serve a broad --- if self-selecting --- audience, and both are only showing pure results. Actually I would be more inclined to accept Wiki results since there is very little possibility of bias, whether bought or skewed, as there is no profit in Wiki misreporting what their servers list. Claverhouse (talk) 21:32, 20 September 2012 (UTC)

Wikimedia is a great source for statistics. In fact, Wikipedia is the only top 10 site that provides statistics (as far as I know) and I don't think it's biased. Being on Wikipedia, it would be normal, I think, to have its statistics displayed on the top of the page. Anyone can make charts for other sources and display them. A3e6u9 (talk) 15:55, 13 October 2012 (UTC)

I agree that the huge user base of Wikimedia servers, in all languages and across all projects, must provide pretty interesting stats. Other stat providers often have limitations, like a US bias, or an English language bias, a bias due to counted web visitors' specific interests, or even (we'll probably never know for sure) a bought bias as suggested above. No one says Wikimedia's stats are perfect, but no one can deny that they are interesting. They're properly labelled and sourced, and that's just about what encyclopedias do. --Nigelj (talk) 16:35, 13 October 2012 (UTC)

The top chart's bar lengths don't match the percentages. Mac OS X with 7.78% has a shorter bar than iOS, at 7.08%, for example. Also, the Android bar length at 5.11% appears close to half the size of the iOS bar at 7.08%. — Preceding unsigned comment added by 67.169.121.41 (talk) 12:44, 16 October 2012 (UTC)

Fixed (set percentages according to table) --77.1.10.131 (talk) 19:36, 16 October 2012 (UTC)

For mobile and tablet data I suggest using this site, it seems to have the best stats. The site owner is an ex-Nokia executive, who is now a consultant. UrbanTerrorist (talk) 15:27, 14 November 2012 (UTC)

This article seems to mix market share, which is the share of sales (revenue) during a given time interval, with installed base share, which is the share of the installed base (units, probably) at a given point in time. These are very different things, so this ought to be cleared up. Based on the term ‘usage share’, it sounds like it is referring to installed base share rather than market share, but market share is also important (and more widely discussed in the business literature.)

Perhaps the article could be split into two sections, one on market share and the other on installed base. Market share is generally easy to measure when publicly traded firms are involved, since they are required by law to publish revenue data. A product’s installed base is a much more nebulous thing. It is often difficult to measure, and estimating it relies crucially on proper statistical methodology. As a result, mixing the two concepts together is not really a good idea.

For a definition of market share, see: http://www.investopedia.com/terms/m/marketshare.asp Faagel (talk) 14:08, 24 July 2012 (UTC)

Home vs workplace would also be an interesting differentiator. Game companies often cite the 8-9% of Mac installs as the reason why they won't make mac clients for their games, however I imagine if 50+% of windows installs are workplace installs that would never see a game installed on it, that that could make a difference for whether or not companies consider Macs worthy of game clients. — Preceding unsigned comment added by 173.247.202.120 (talk) 03:41, 17 December 2012 (UTC)

Gartner reported on December 23, 2008 that Linux on System z was used on approximately 28% of the "customer z base" …

First, the cited link doesn't work, either for me or for the Internet Archive Wayback Machine (since 2011); that Web page is either dead or restricted. Might anyone have another source of that information?

Second, the idea that 28% of mainframes use Linux prompted an author of the

”) which can run different operating systems simultaneously. The actual division of operating system usage by machine is closer to 28% Linux + ~100% z/OS, although the existance of all-Linux mainframes means that the share of z/OS will never be 100%. I'd like to update that 72% value to a much more correct percentage if a good source can be found.

206.205.52.162 (talk) 20:22, 18 January 2013 (UTC)

according to Net Applications. And here in the summary section. I'm stunned by these numbers. For smartphones alone others (idc, gartner) determine percentages above 70% for Android. With smart phone shipments in China (and China probably not being IOS dominated) outgrowing those in US (canalys.com) it's hard to see how IOS could push down Android share to 25%. --HeWhoMowedTheLawn (talk) 11:06, 19 January 2013 (UTC)

The center of gravity is shifting from desktops to tablets. The chart in the lead currently only shows desktop share. I suggest expanding the chart to include sections for: desktops, laptops, tablets, phones, servers, etc. Or maybe just take the data from the table in the summary section and turn it into a chart for the lead. Sparkie82 (t•c) 18:40, 4 February 2013 (UTC)

Many of the stats in this article are incorrect, as an example IOS 60.56% and Android 24.51% most agree that these figures should be reversed, Net Applications are the only ones reporting IOS as having the lead in the market.

Other problems are in the Summary section only ubuntu is counted for total Linux Desktops. Distrowatch http://distrowatch.com/ has been the the Linux community's goto for determaning os popularity for many years, being that most of the others distrobutions do not count how many users they have do to privacy concerns Distrowatch is used to reflect popularity. Ubuntu is ranked at #3, Linux Mint is the current leader #1 and Mageia at #2. Note: Thee is 5000+ Linux distrobutions they should be taken into consideration also.

In Summary "Server (web)" is also incorrect the stats that are provided by W3Techs, are not reliable there is no break down that differentiates Linux from Unix. Note whoever placed this entry needs to learn the differance between Unix and BSD. Unix is a trademark I only know of three Operating Systems that legaly retain the title of Unix.

Also TRON or ITRON the most widely distributed Operating System in the world isn't even listed, and its numbers alone dwarf every os listed, to the extent that an accurate count would be next to impossable, how do you count every Digital Clock, Microwave, .... on the planit.

TRON is an open real-time operating system kernel design, and is an acronym for "The Real-time Operating system Nucleus". The project was started by Prof. Dr. Ken Sakamura of the University of Tokyo in 1984.

The Industrial TRON (ITRON) derivative is one of the world's most used operating systems, being present in billions of electronic devices, if not most digital devices with the exception to personal computers (Deskop's, Laptop's, Tablet's, Mobile Phones and Servers).

This page is about "Usage_share_of_operating_systems" not personal computers, and there are people here skewing the stats to reflect there own personal choice in operating systems (Windows and OSX).

Being that (Windows, OSX, Linux, BSD) are highly unlikely to ever come to terms that one os is more popular than another, this page serves no other purpose than to discredit others and be ammo for Trolls. It's hurting Wikipedias reputation. The errors I see only cast doubt that anything in Wikipedia is factual.

— Preceding unsigned comment added by Qinux (talk • contribs) 07:41, 20 February 2013 (UTC)

1. IOS and Android marketshare is disputable, all known commpanys that Tracks the Market Share of Mobile Platforms with the exception to Net Applications identify Android as the Mobile Marketshare Leader.

2. Android is a Dalvik SDK Layer ontop of the Linux kernel its not its own Operating System. The correction would be Linux(Android) just like how Ubuntu, Fedora are not there own operating systems there "Linux Distrobutions".

 — Preceding unsigned comment added by Qinux (talk • contribs) 18:04, 20 February 2013 (UTC) 

Smartphone sales grew 38 percent last quarter to reach 217 million units worldwide, and over 700 million units for the entire year, according to a new report from Strategy Analytics. Of those 700 million-plus smartphones, 68.4% smartphones ran Android as the operating system, while only 19.4 percent ran iOS, Apple’s mobile operating system. Read more at http://venturebeat.com/2013/01/28/android-captured-almost-70-global-smartphone-market-share-in-2012-apple-just-under-20/#McUAxppTQ2Ped1Fa.99 — Preceding unsigned comment added by 74.45.230.127 (talk) 01:08, 23 March 2013 (UTC)

1. IOS and OSX are listed as "Other Unix" thats incorrect there both BSD derivative's

2. BSD is listed under "Other Unix" thats incorrect BSD is its own OS

3. Under Super Computers there is the entry 4.2% UNIX that should be changed to (HP-UX, Aix, Solaris) these are they only UNIX OS's that still exist.

4. PS3 is listed as "Other Unix" this is incorrect PS3 is a NetBSD derivative and should be listed under a BSD column

5. QNX is listed as "Other Unix" this is incorrect QNX is its own RT Operating System, it neather UNIX or a BSD derivative, and should have its own column, the only thing QNX has in common with a UNIX OS like Solaris or a BSD OS like FreeBSD is it's Userland Tools but those tools also work on Windows(Cygwin) so calling it UNIX is like calling Windows UNIX or Linux based on Userland Tool support.

OS X is the only BSD with a Unix certification.163.178.208.6 (talk) 17:05, 20 May 2013 (UTC)

Once again, in these edits, Deepen03 (talk · contribs) has replaced the data in the lede summary graph from Wikimedia to Net Market Share. Once again these changes are without edit summary and with no comment on this talk page. Are other editors happy with this change? The differences are significant. Looking at the latest Wikimedia stats (March), we now say 44.73% for Windows 7, whereas Wikimedia says 34.72%; we show 61.41% (of mobile OSs) for iOS while Wikimedia shows 15.52% for iPhone and 8.23% (of overall total) for iPad. This should be something agreed by consensus here on the talk page, not left for people to work on making a new graph, only for one user regularly to replace it with their own preference again. --Nigelj (talk) 20:20, 12 April 2013 (UTC)

Given that Wikimedia stats are so different from the most often cited sourced I find it unacceptable that self-selected Wikimedia is being used in the prominent lede section. I recognize that all sources are probably biased, but we cannot even begin to address that problem with Wikimedia without venturing into

WP:OR. Every time I see netmarketshare or statcounter or some other source cited in reputable medias I am reminded how off Wikipedia is by presenting raw data as information. I totally agree that netmarketshare or some other reputable source should be in the lede. At least for those sources we can find secondary sources which can be used to warn about possible bias. Useerup (talk

) 21:33, 12 April 2013 (UTC)

Note: This isn't the first time Deepen03 (talk · contribs) likes to remove anything not favoring his/her own personal favorites. Other company’s have placed there own graph's with there own data, interestingly favoring the rest of the worlds data. And being its not what the company Net Market Share, shows has been promptly removed. — Preceding unsigned comment added by 74.43.197.233 (talk • contribs) 15:32, 16 May 2013

The section above explicitly says "statistics related to one web site only are excluded" so W3Schools should probably be removed. Plus it is bound to have a web developer bias. --89.12.49.35 (talk) 20:20, 2 July 2013 (UTC)

With the 24th of December edit ^[1] the source for the bar chart data was changed to NetMarketShare. There was no prior discussion here. No source for the claim that was written in the name of the reference "name=NetMarketShare (Please do not change to wikimedia, NetMarketShare is the most accurate site" was given. The editing comment is empty.

NetMarketShare says about its methodology "We collect data from the browsers of site visitors to our exclusive on-demand network of HitsLink Analytics and SharePost clients" which seems a moderately good start.

Going back to the wikipedia article one year ago^[2] shows that NetMarketShare has the smallest Windows 7, the smallest Vista and the highest XP usage share reported by any of the 9 providers. Its XP usage share is more than 6% higher than the second highest reported share. Likewise it has the second smallest reported share for OSX and the second highest for all Windows. NetMarketShare was not representative (for the providers) at that time. Going back another year^[3] shows that NetMarketShare likewise had reported more than 5% higher for XP than any of its peers.

On Dec 25th NetMarketShare was renamed to Net Applications by an anonymous edit (Net Applications links to NetMarketShare). The editing comment is empty again. I neither like the change to NetMarketShare/Net Applications nor how it was arrived at and suggest to revert. --HeWhoMowedTheLawn (talk) 21:57, 2 January 2013 (UTC)

It's an ongoing argument, that we'll probably never get to the bottom of. Here's a statement by StatCounter about what they see as the shortcomings of Net Applications' data. When I used to take a keen interest in the stats at Usage share of web browsers, and we used to calculate a monthly median, I often found that the single set of figures closest to the median was usually Wikimedia's. It seems like all the others have fish to fry, and Wikimedia just collects dumb figures and publishes them for free. The Wikimedia usage totals are also huge, and there is no real user bias - Wikipedia and other WM projects' links just keep appearing in the top few of almost every Google search, no matter who the user is or what they search for. I don't think there's a better way to gather representative statistics, IMHO. So, yes, my opinion would be to revert. But I can't prove it. --Nigelj (talk) 23:34, 2 January 2013 (UTC)

I agree too. Revert it.Michael9422 (talk) 14:26, 5 January 2013 (UTC)

The NetMarketShare data is NOT a summary graph, it is a desktop usage graph, and as such belongs in that section. Wikimedia's data, although flawed like most of the data on this page, is a more general summary of web usage. 108.176.134.72 (talk) 20:37, 6 January 2013 (UTC)

"The Wikimedia usage totals are also huge, and there is no real user bias"... While Wikimedia clearly has not some of the commercial bias we see elsewhere there are still problems of the weakness of User_Agents and unique IP addresses in the data. Also Wikipedia has a serious bias to English, thus not being representative of much of the world. No website that I know could overcome that problem except search engines but they are not talking.Pogson (talk) 17:16, 27 January 2013 (UTC)

Someone reverted back to NetApplications and made it exclusive to desktop systems. This is a summary chart, so it should be the summary, not desktops. 1980's Mainframe is to 1990's PC as 2000's PC is to 2010's Mobile. In short, desktop market share will become a grosser distortion of actual usage over time. Let's use the real numbers. 69.193.188.146 (talk) 15:35, 8 February 2013 (UTC)

The change back to NetApplications was made in this edit by Deepen03 (talk · contribs) on 2 Feb without any edit summary. Deepen03 has not joined in this discussion, and the consensus here seems to be to use Wikimedia stats for the lede graphic. I would support re-introducing Wikimedia stats, but the figures previously used are now looking fairly out of date, and we should use the most current available. An hour later, two anonymous edits were made from 69.125.196.139 (talk · contribs) that added unsourced comments of personal opinion that supported Deepen03. That IP also has not commented here so I'll revert those straight away. --Nigelj (talk) 18:32, 8 February 2013 (UTC)

Just... get rid of this chart! It has nothing to do in the summary section. Having a single source of stats displayed up there turns the article into an advert. Net Applications itself isn't especially renown for its unbiased position. I'm not spending a lot of time analysing all of this but it feels wrong. I leave the decision to those who follow this more closely, but if I see no decision emerging in a couple weeks I'll take action myself. :) Ovocean (talk) 17:12, 2 July 2013 (UTC)

The fact that the graph is still here is pathetic... It's so obviously skewed data it's embarrassing. ALL of the other data points towards dominance by Android in terms of market share, and the only data show in the summary goes against this? The fact that the NetApplications page also has a warning pending says it all. Apple are sticking their sticky fingers in Wikipedias pie. JosephAshburner (talk) 17:51, 11 August 2013 (UTC)

Moved the bar charts from the introductory section (where they are off topic an prominently visible) to the corresponding subsections. As a side effect the bar charts can be more easily compared to numbers of other vendors... I still think the data of NetMarketShare/Net Applications should not be used for the bar charts. --HeWhoMowedTheLawn (talk) 19:15, 14 August 2013 (UTC)

I guess this article isn't updated regularly. the Tablet number were from 2011! I updated the Tablet Number to reflect the current Market share. Number Android Tablet lead the market share. Source is here: http://www.idc.com/getdoc.jsp?containerId=prUS24253413

I also remove the speculation quotes from a source of what Apple plans to do with iPad shipments and just added he facts.

Android is now the Largest Tablet Operating System. As of Q2 2013 — Preceding unsigned comment added by 98.203.36.3 (talk) 02:22, 25 August 2013 (UTC)

I see XML comments like "Please keep Net Applications as the source for this chart!" in two places in this article. This is not the way that individuals get to enforce their POV in Wikipedia editing disputes. I see the thinnest sliver of consensus in two discussions above started in January and September 2013. Experienced editors know that on Wikipedia

) 15:01, 22 October 2013 (UTC)

Rounding?

Better rounding off to one decimal: 40.39 becomes 40.4. The accuracy now looks like over-accuracy. Maybe you editors agree! Good luck.Super48paul (talk) 08:22, 3 November 2013 (UTC)

I am not the creator of this article but, please don't delete it. This is also a useful article knowing which OS has the highest-sales and that is a fun fact. Leave it blank? Seems like the article is still updated... — Preceding unsigned comment added by EncycloAgainstWiki (talk • contribs) 13:23, 14 November 2013 (UTC)

On November 3 2013 the netbook section was deleted by a user at IP 68.191.211.99 without discussion on the talk page. Although certain operating systems do not fair well on hardware in the $100-$300 dollar range, the market is certainly not dead. "Chromebook sales have grown to around 25% share of the sub-$299 sales" ^[5] . Although a lot of the market was absorbed by tablet devices, it is expected to be a significant growth sector in a declining PC market.

Also, the deleted article text claimed Microsoft windows dominated the market, but only cited projected figures from 2009 that Microsoft obtained from a pivate analysit. While it's true Microsoft embraced the market by extending sales of Windows XP into 2009, they were forced to extended the definition of the hardware and price point for later versions of the operating system. Although the sub $300 dollar laptop was almost extinguished it is far from dead. The market "is expected to increase more than 10% in 2013. By comparison, the PC industry as a whole is forecasted to see negative growth of -7.8% this year." ^[6] 24.90.208.131 (talk) 05:14, 18 November 2013 (UTC)

The offending edit specifics can be found by comparing: 22:02, 3 November 2013‎ 68.191.211.99 (talk)‎ . . (33,677 bytes) (-2,360)‎ . . (→‎Netbooks: Section deleted as outdated; 'netbook' is not a separate (or relavent) category from laptops) (undo) (Tag: section blanking)

Noted reference to Windows Mobile in the mobile bar chart as WP meaning Windows Phone from 08:57, 3 November 2013‎ 82.232.24.147 (talk)‎. This obfuscates Windows market share on this WP article. The text also listed example mobile operating systems in a strange order. That is, windows Mobile, 7 and 8 with 0.5-3% market share were listed before Android. Changed both with comments. 24.90.208.131 (talk) 23:41, 17 November 2013 (UTC)

Column was widened by by Nigelj at 23:54, 17 November 2013 to prevent it going to two lines‎, Thank you! 24.90.208.131 (talk) 05:21, 18 November 2013 (UTC)

2013-11-18 It may be time to put the summary table in the article summary

It would appear that the Summary table is being maintained consistently, whilst living at the bottom of the page for almost a year. It has also escaped the attention of the PC Net Applications reverters long enough to mature. I would like to propose:

1. Put the "Market Share by Category" table to the top of the page. It was formerly called "Summary Table".
2. Replace the summary figure with a stacked vertical bar graph based on device sales ^[7] where available.

A reliable source for total embedded and real-time devices seems unlikely, sales not tracked or "NA" seems appropriate. After market installs should be mentioned.

Obviously, the PC market share estimate graph should go down with the PC section where it belongs. Since a consensus on source seems unlikely, a dual bar graph showing Net Applications next to Wikimedia for each OS seems very appropriate. With a good color scheme the Net Applications figures can be juxtaposed with a half dozen usage sources. 24.90.208.131 (talk) 06:27, 18 November 2013 (UTC)

2013-11-18 Article states bar chart is both mobile and desktop, however chart has been reverted to desktop only. 24.90.208.131 (talk) 00:22, 18 November 2013 (UTC)

2013-09-02 the bar charts with Net Applications data have been re-added. This should be reverted:

1. the data itself is controversial,
2. the data deviates from data given on the very page (by more than 40% (forty! percent) for Android),
3. the contributor ignores discussion on this talk page,
4. desktop data is shown on overview (first) section (clearly off-topic there),
5. the commit itself is flawed (e.g. replaces 2013 data with 2011 data)

--HeWhoMowedTheLawn (talk) 20:10, 7 September 2013 (UTC)

No revert. Wikimedia is (at best) a primary source. Net Applications is *the most* often cited source by reliable secondary sources.

1. You need to back up your claims with reliable sources. Who says the data is controversial? citations needed. Remember that WP must present all relevant and significant viewpoints. It is NOT for WP editors to decide on suppressing a source obviously trusted by so many secondary sources.
2. That the data deviates is no reason to distrust it. It could just as well be that the other data is wrong. More likely, though, that Net Applications data is biased differently from the bias of the other sources. All of the sources has bias. Net Applications is the only source trying to correct for that bias.
3. Not material to this discussion
4. No, the lede should summarize or exemplify the article. Choosing one of the most prominent sources as the lede seems reasonable. Certainly *much* reasonable than biased, self-referencing Wikimedia statistics.
5. Then please correct it. Useerup (talk) 20:51, 7 September 2013 (UTC)

You misunderstood me. I'm against re-adding. This means: rather show no charts than misleading charts.

1. I did back up my claims. The paragraph above addresses and criticizes NA's methodology. It addresses the history of the Net Applications data (having a track record of extreme data). And a reference to an open letter against NA by a competing company was given by Nigelj (interesting read, link again given in paragraph above). Also see the paragraphs in archive 5 beginning with Mobile Operating System .. or Android captured .. or Android 25% (that one by myself). A note to your allegation of suppression, 2011 the article cited 9 sources (with NA at the extreme [2011 Web_clients] for Windows 7, XP and Vista), nowadays just 4 are left. And one of the survivors (naturally speaking Net Applications) then being prominently shown with two charts!
2. it deviates markedly [Mobile devices] from other data. This means it ought not be the summary. Not in a section, not in the lede.
3. given.
4. desktop data is off-topic there. There is a separate desktop section where desktop data should live in. See also [archive 3].
5. No. That would be the third time I remove that very data (previously 2013-05-14 and 2013-08-15). HeWhoMowedTheLawn (talk) 16:43, 9 September 2013 (UTC)

I may have misunderstood that you did not want any graph in the lede. I apologize for that. And disagree. It is perfectly reasonable to use a graph in the lede, even if that graph is only representative of (a big) part of the article. The graphic is an illustration. And the graph is not misleading - at least you have not shown any substantial opinion by reliable sources that NA is misleading.

1. You did not back up your claims with reliable sources. An opinion piece in an open letter (by a competitor, no less) is not a reliable source for calling a statistic misleading. At best it could be used for a section on differences in philosophy and measurement techniques between stat providers. In the meantime, many real
   reliable sources
   quote NA (and other statistics providers as well). WP assumes that reliable sources have meaningful editorial oversight and that they wouldn't quote sources if they didn't trust them.
2. It deviates, so what? The stat providers do not even claim themselves that they measure the same metrics. They are aware that they measure different demographics. NA tries to correct for skew caused by geographic bias, statcounter does not (instead preferring to quote raw data). Differences are expected(!) and certainly not cause for exclusion, nor for deeming one source more reliable than another. Suggestion: Should we create a column in the table where we summarize the measurement technique for each provider ("raw page impressions of html pages on Wikipedia and associated sites", "Based on unique visitors to 40000+ commercial sites, corrected for each country according CIA population numbers" etc)?
3. .
4. No, it is an illustration. A given that NA is easily the most often quoted stats in reliable sources it is certainly not off-topic.
5. Didn't ask you to remove it. You claimed it was wrong - so I suggested that you used that knowledge to correct it. Useerup (talk) 19:40, 10 September 2013 (UTC)

This chart is not representative of this wiki page's subject: the usage share of operating systems. This page is not about browsing habits of various mobile device users, which is exactly what the Net Applications data shows. The chart must be removed all together. Web traffic data may be related, but certainly is not proof of percent market share of installed OS's. Sales and shipped unit data, which differs greatly from the Net Applications chart, would be much more representative of this page's subject matter. Psyden (talk) 17:08, 26 September 2013 (UTC)

yesterday the numbers for supercomputers were updated to November data. But when viewing the diff https://en.wikipedia.org/w/index.php?title=Usage_share_of_operating_systems&action=historysubmit&diff=582267416&oldid=582159485 the data does not show up. (Revision as of 21:21, 18 November 2013) Later diffs (Revision as of 22:47, 18 November 2013 (currently the latest)) do not show removal of November data, so it is supposed to be there yet it is not there. Am I mistaken or is there a glitch somewhere? --HeWhoMowedTheLawn (talk) 18:47, 19 November 2013 (UTC)

If I'd think about private computing (although some home users may have servers, I don't know, if it would make sense to incorporate it in the basic OS "market share" comparison), how would I divide it? Maybe (is this by usage?) into handheld (tablets, phones, etc.?) and non handheld (desktops, towers, notebooks, subnotebooks, maybe even netbooks etc.?) ? Although the sentence starts with "private", of course I'd include the enterprise machines too (except the servers, as stated above: in the article they already are a dedicated specific paragraph by itself too).

The article already seems to follow a little bit along this path; would it make sence to make it even clearer in this direction? --Alien4 (talk) 11:24, 3 December 2013 (UTC)

Since there only numbers for the market share of web servers and which OS used with Apache calculating the numbers for the OS-Share is a little bit difficult and not exact. So I gave only ranges. May the whole line should be removed. I just updated the numbers. – Was not my idea to use them.

I calculated:
apache=0.6539 (Market share of apache)
linux=apache*(27.76+27.75+13.32+10.44+6.64+3.24+0.14+0.13+0.1) Apache market share * Linux Distributions with market share over 0.1%
ounix=1.12 (Zeus is used only with UNIX) + apache*(6.05+0.18) Darvin and FreeBSD are UNIX-like
windows=15.68 (IIS is only running on windows) + apache*3.74
unknown=100-(ounix+windows+linux) what i don't know and have to add to all values to get the possible maximum market share

windows
18.125586
windows+unknown
37.269075
linux
58.537128
linux+unknown
77.680617
ounix+linux
62.730925
ounix+linux+unknown
81.874414

--Fabiwanne (talk) 12:22, 8 December 2013 (UTC)

Reference 26 counts only operating systems which are sold with Hardware – and it counts by revenue of the hardware. I think this is really misleading since z/OS typically soled with a Mainframe with a price of a view hundred thousand $, while linux support is often soled seperatly to the hardware. So it says nothing about the market share of z/OS or linux. I like to remove this statistic. --Fabiwanne (talk) 13:07, 8 December 2013 (UTC)

I like the table, but was wondering if we could add a heading above all the windows os's, Apple, Linux, and other to include a link to the NT Kernel, Unix and Unix-Like Operating Systems, and other as all the Windows OS's are subsets of NT and all the Apple and Linux stuff are subsets of Unix/Unix-Like. NT and Unix and Unix-like are clearly the majority of everything simply divided into subsets thereof, and adding this to the top of the table would express this more clearly. talk 07:49, 13 April 2014 (UTC)

[Update: iOS (all versions combined) is not far behind XP, confused the gray line for "other" with the gray line for iOS:]

According to Statcounter May statistics Android is now the "second" most popular OS after Windows 7. Android is for sure many versions and fragmentation was at one point a problem. Still as a whole, more popular (for web use) than XP and iOS also and as people on iOS upgrade frequently iOS 7 could soon be second or at least above XP. As an application platform, I'm not sure how fragmented Android is any more, there are different API levels (but not so important? Some compatibility layers to minimize that).

Android is the most popular

) 10:47, 19 May 2014 (UTC)

Note the .csv file gives a better picture of this mysterious "other" (11.81%). Seems less than half of it might be Win 8.1 (4.29%), not included in the graphic unlike Win8 (5.31%). Still Microsoft Windows "other", Linux based other, general other or all of them combined give less than this 11.81%. The table seems wrong.. but might be only out-of-date. comp.arch (talk) 11:36, 19 May 2014 (UTC)

The ref includes Cisco (5%) I usually do not associate with operating systems (Linux or Unix). But they sell servers. Right to assume these numbers reflect those only?

And about Windows "50.3% of overall quarterly factory revenue"? "factory revenue" implies only about the cost of hardware? Or would Windows servers include Windows OS and be more expensive (than say Linux servers)? comp.arch (talk) 12:04, 19 May 2014 (UTC)

NetMarket share seems to count traffic from certain countries multiple times, giving the popular browser and OS in those countries added weightage...

The Net Market Share data is weighted by country. We compare our traffic to the CIA Internet Traffic by Country table, and weight our data accordingly. For example, if our global data shows that Brazil represents 2% of our traffic, and the CIA table shows Brazil to represent 4% of global Internet traffic, we will count each unique visitor from Brazil twice. This is done to balance out our global data. All regions have differing markets, and if our traffic were concentrated in one or more regions, our global data would be inappropriately affected by those regions. Country level weighting removes any bias by region.

--Ne0 (talk) 10:08, 11 June 2014 (UTC)

Blog criticizing the accuracy of this page:

• http://therealmadhatter.wordpress.com/2010/11/06/server-operating-system-market-share-lies-lies-and-more-lies/

--Guy Macon (talk) 16:03, 27 June 2014 (UTC)

we should have something like this in the article...

Source	Date	Microsoft Windows: 61.37%								Apple: 16.28%		Linux kernel based: 17.08%					Others
		8	7	Vista	XP	WP	2003	2000	98	OS X	iOS	Linux	Android	Chrome	Bada	Symbian
StatCounter Global Stats	June-11[update]	9.43%	37.71%	2.42%	11.07%	0.61%	0.1%	0.02%	0.01%	6.06%	10.22%	1.35%	14.78%	0.13%	0.05%	0.77%	3.7%

I agree - it would be very interesting to count mobile in (Microsoft really this low - and not lower? Numbers made up? Or this old?).. Note however that Symbian is not Linux-based, they say

RTOS, might be Unix-like, not the same thing. I would say just skip Symbian: "Official website symbian.nokia.com (defunct as of May 2014)". Have any Symbian phones been release for years now? And Bada is also dead/replaced with Tizen (those are Linux-based). For now might also be counted under other? comp.arch (talk

) 20:57, 12 June 2014 (UTC)

re-calculated the stats, including Symbian there are actually 6.05% other Operating systems browsing the web. Linux Kernel is now at 16.31%, just above Apple. [1] Data from gs.StatCounter.com --Ne0 (talk) 06:05, 13 June 2014 (UTC)

If you are counting bada as Linux kernel based you are not counting LG (also 0.05%) as Linux kernel based. Not sure what LG could be other than webOS? They bought webOS and it is Linux kernel based. Then again while webOS and ChromeOS and FirefoxOS (and Android) are Linux kernel based, I'm not sure it helpful for people group them together. They are web-based first and foremost, while Linux kernel based, not Unix-like in my opinion. comp.arch (talk) 13:23, 16 June 2014 (UTC)

Should we not have new categories now with new domains like Android wear, Android TV and Android Auto coming up ? Of the 3, Android wear is already a reality. — Preceding unsigned comment added by 72.163.217.102 (talk) 11:17, 31 July 2014 (UTC)

iWatch hasn't launched yet. And Android Auto, Apple Carplay, AGL (Tizen Linux), and Windows are currently gearing up. Does anyone have statistics on Apple TV vs Google/Android TV ? --Ne0 (talk) 10:36, 1 August 2014 (UTC)

I can't understand how the number(0.96) for "Linux other" section is achieved in the Web clients table. I had previously thought that "Linux other" included Symbian, but the Web clients OS bar Graph already shows Symbian as a separate bar. --Ne0 (talk) 08:20, 2 September 2014 (UTC)

Anyone know where 32 bit vs 64 bit stats could be sourced to be included? It seems like a significant development of the current times? — Preceding unsigned comment added by 130.185.67.135 (talk) 15:13, 30 October 2014 (UTC)

Since Microsoft has tried to make their X86/AMD64 tablets as close to desktop systems as possible, I am wondering if this might cause some confusion in the count. Unfortunately I don't have a decent source for this, so I'm just dropping it here hoping someone will find a source.

Also there are rumors that Microsoft has had a lot of returns. Again, I don't have a source I can pass you (and since I write tech articles I never edit this page). UrbanTerrorist (talk) 02:43, 3 February 2015 (UTC)

Yes and no, there is actually a dilemma (for me). I always assumed tablets should be counted with mobile. Then the other day, I saw desktop and tablets counted together, in one source (not used here). Looking now at StatCounter I see Syrian Arab Republic map (hovering over) I see: "48.49% Desktop 48.41% Mobile 3.1% Tablet 0% Console".[2][3] Desktop is only a tiny bit higher than Mobile and the country is blue ("Desktop majority"). I put in quotes as desktop is not majority (>50%) and only higher than Mobile, not higher than Mobile+Tablet.. *I* would have used green..

As most tablets are based on Android (or iOS), so-called "mobile operating systems".. I would have counted tablets with mobile. I think we can still do that (at least in some cases) if we just say what we are counting. I would want a source counting tablets with mobile so we can say some do that and others count with desktop.

Maybe we should just make all three (four) categories explicit with individual counts to make sure people realize what the counts mean. I'll try to not say "mobile majority" unless it (only) is clearly so. comp.arch (talk) 11:31, 28 April 2015 (UTC)

Comp.arch Per your edit summary question, there could be several sources of bias. (All of the following is either my own OR, or based off of StatCounter's FAQ)

• ipad users could browse more than android users
• Their stats are based on the websites that host their analytics code. There could be bias in the selection of those sites that favors android
• all ipads are human use devices, there are embedded or non general purpose android tablets that may not be usable for browsing
• There are a limited number of ipad models out to track. Easy to know you got them all for identifying tablet traffic. There are hundreds/thousands of android models. If StatCounter's list doesn't contain that model, then it doesn't count as tablet traffic (see http://gs.statcounter.com/faq#tablet-definition last update in 2013. Who knows if its just their FAQ or their real analytic s that are off)
• Per the phandroid link below, android tablets request the "desktop view" of webpages, so may be skewing the results.

However, broadly speaking the stats are probably accurate. Here are some other measurements from different sources showing similar results

• (Chitka ad network) http://www.patentlyapple.com/patently-apple/2014/02/apples-ipad-generates-4-times-web-traffic-than-android-tablets.html
• (raising some of the questions you do) http://phandroid.com/2012/12/04/ios-vs-android-web-traffic/
• http://www.businessinsider.com/ios-android-web-share-us-2014-7
• (2012, outdated) http://marketingland.com/study-ipad-controls-98-percent-of-tablet-web-traffic-26272

Overall mobile web traffic (not limited to tablet) android has taken the lead though, per almost all sources

• http://www.businessinsider.com/android-is-beating-ios-in-web-usage-2014-8

Gaijin42 (talk) 18:35, 21 November 2014 (UTC)

The tablet usage greatly underestimates Windows because it counts only Windows RT. Most Windows tablets sold use the full x86 Windows 8.1 and so are counted as desktop/laptops. Since some of them are used as convertibles Windows devices do not neatly fall into one or the other category. — Preceding unsigned comment added by 108.194.22.18 (talk) 15:37, 8 January 2015 (UTC)

Thanks (both), I missed when this was posted so I'm not sure what "edit summary".. I find: "android tablets request the "desktop view" of webpages" however interesting. Hopefully this article isn't too biased, at least on my account, towards say Android. It is very difficult to get good numbers. Besides web use numbers being biased, that is not all there is (see my latest edit). Are there any numbers on non-web use of smartphones/tablets/PCs? Would that be in minutes? Besides what minutes would you count.. minutes of your smartphone in your pocket or "actual" use.. comp.arch (talk) 09:56, 27 May 2015 (UTC)

I just looked through the StatCounter numbers from Dec 2014, which are very important because they drive the graph, and they do not jive up with what I am seeing when I click through to the Refs 4, 5, 8, and 9. The numbers I see are as follows from Ref 8 (click Download Data on page to see CSV):

• Win7: 33.52%; Win8/8.1: 10.91%; WinVista: 1.94%; WinXP: 8.14%; WP&RT: 0.79%; WinOther: 0.12% (Windows TOTAL: 55.42%)
• OS X: 5.34%; iOS: 11.91% (Apple TOTAL: 17.25%)
• Linux: 1.08%; Android: 20.62%; Other Linux (includes Chrome OS, bada, and Samsung): 0.77%: (Linux TOTAL: 22.47%)
• Other (100%-numbers above): 4.86%

If no one objects, I will update the table and charts on the page. --Nasa-verve (talk) 16:39, 22 January 2015 (UTC)

Bada is NOT based on Linux Kernel, it is based on Open Source forks of BSD, just like Mac OS X and iOS. So, should we be adding Bada under Linux Kernel ?

Also, as Symbian is not based on Linux anymore, should we make the "Linux - other" category into "Samsung devices" ? The Statistic 'Samsung' is probably Android, as this stat started right after the launch of the original Galaxy phone. So should we merge 'Samsung' with 'Android' ? --Ne0 (talk) 10:53, 4 June 2015 (UTC)

"Android has the largest installed base of all general-purpose operating systems" Where are the reference numbers to back such statement? From the rough data I could gather, by 2015, Windows should still lead the installed base with about 1.8 billion units in use (with more than 90% of market share in the desktop market), while Android should be at about 1.7 billion units . If that claim cannot be backed by data, it should be either removed or changed to indicate that the biggest installed base is not currently accurately known between Android and Windows. — Preceding unsigned comment added by Dockimbel (talk • contribs) 05:24, 10 July 2015 (UTC)

For the largest Install base citation, we can look at device shipment: [4][5], or add up install base: [6]. And as per IDC, the reference for your Computer estimate(Worldometers), 7-12 inch devices are Personal Computers, and NO-where does it say "Windows".[7] Also, Android is now the most popular OS on weekends,[8] and will soon become the most popular OS over all. I'm guessing that includes Android, as well as

Android Wear, and Android Auto. --Ne0 (talk

) 18:35, 7 August 2015 (UTC)

Desktop/laptop shares

It's nice to get up to date figures, but there's something wrong with these August 2015 stats. They total up to 105.38% I don't have access to the subscription service at Net Applications so I don't know what's happening, but it seems unlikely that the Windows share has suddenly increased from 90 to 95%. I doubt that no users have switched from Windows 8.1 to 10. Chris55 (talk) 21:49, 25 August 2015 (UTC)

It is referenced to statcounter bar graph. Windows 10 automated bandwidth usage is ultra high: data collection[9] + update distribution[10]. Maybe some part of this Internet usage is mis-calculated as user activity ? --Ne0 (talk) 10:01, 29 August 2015 (UTC)

I was a little puzzled by seeing only a weeks worth of statistics used. Now there is a full month of statistics available after Windows 10 was released. I think We should as a rule use month statistics (as other statistics compared to are for full months, and may not be available for weeks (without paying); also the stats could fluctuate, I guess somewhat within a month).

In a way having full months would be ok, when relative popularity is kind of stable. It isn't however.. Windows 10 is new, but the first burst popularity burst has died down.[11] Still, a month including weeks just after release, would give lower number than the current one. That can be fixed to have a footnote with *only* Windows 10's number for a week or "instant". I guess the numbers do not matter too, much unless they would give different relative ranking. comp.arch (talk) 15:40, 1 September 2015 (UTC)

This is not entirely surprising.. I just hadn't looked at such a short period to see this clearly before..

In the US: iOS (all versions) just barely became more popular than Windows 7 (yes, only one version of Windows) on August 2 (and close on others). iOS was then highest ranked, but that would be unfair as you should add all Windows versions together.[12]. Aslo even more pronounced for Oceania/Australia[13]

[When I say huge drop in Windows 7, this is still apparent even if mobile ([smart]phones) are excluded AND consoles, leaving only desktop and tablets. It doesn't matter if its desktop vs. mobile (phones) are compared or desktop vs. tablets, indicating the drop is in desktops, and is absolute. Increase in mobile or tablets may not be absolute, only relative to desktops. To see an OS top Win 7, however including mobile (phones) is needed.]

Same can be seen worldwide, then Andoid topping Windows 7 and then more pronounced, say on 15 August.[14]

Say something about this?

Seems mobile is what people want.. just are tied to a desk[top] at work/school[days].. I wander if it is preferred or just what people have.. comp.arch (talk) 10:03, 4 September 2015 (UTC)

Apple devices: 1> Apple market-share is partly hype/loyalty for customer service; 2> OS X and iOS are a (proprietary)fork of BSD, and Apple open sources part of their code.[15]

Android devices: 1> Android is a fork of Linux, and open source; 2> dosen't require payment to Google for app development; 3> Most android devices provide for a way to unlock bootloader.

Most new Intel motherboards(except Purism[16] & Chromebooks) are boot-locked with UEFI, and therefore restricted to Microsoft signed operating systems. No one likes to have their computer software system locked away, & now Windows 10 updates are enforced by Microsoft. On the other hand, Android provides a great ecosystem for developers, & manufacturers to jump right in. And being open source, Android isn't just limited to mobile phones/tablets. This includes Remix mini [17], Console OS [18], Andromium OS [19], 3D mobile game console[20], etc. --Ne0 (talk) 19:20, 4 September 2015 (UTC)

Ne0Freedom, this seems good info and probably all true, but not related to the section I started (or even the page). I wander if you meant to put this elsewhere? comp.arch (talk) 22:34, 4 September 2015 (UTC)

Sorry, I thought you were wondering how Android is topping Windows 7. I wanted to point out that Android is not only used as a 'standard mobile' device. When

Steam Machines launch, we may see Desktops rise again, this time on Linux. --Ne0 (talk

) 09:06, 5 September 2015 (UTC)

Just to clear up something here. Android is topping Windows 7 by being more popular, but not in the "desktop"-category. It's just that Android – only on "mobile platforms" already has a larger installed based than Windows on "PCs", desktop/laptops/tablets. We'll see, when/if they are bigger on Microsoft's (older) turf. Your info on, say, RemixOS (that I didn't know of) was interesting, thanks. I think that Android-variant, may have have a chance to dent Windows. Unchanged Android, works also with a mouse and keyboard, few use it that way or know of (I did..), but people may prefer it (or not). Anyway, what will happen falls under

WP:FORUM. comp.arch (talk

) 15:54, 5 September 2015 (UTC)

added the info about Android. As for downfall of Windows 7, I think it should go into Microsoft Spyware or Criticism of Microsoft Windows. --Ne0 (talk) 08:51, 6 September 2015 (UTC)

No, unless you have a really good source. As I pointed out, in the next section, numbers of Windows can, and probably are, be going up (absolute), while relatively going down. And people using mobile more (or even "desktop"/Windows less), doesn't mean it is a criticism of Windows, except in the

mainframes redundant (mostly, they are still around but a tiny minority uses them directly, while many indirectly..). comp.arch (talk

) 11:38, 7 September 2015 (UTC)

Windows 10 growth is slowing down

The general consensus in tech news is that windows 10 growth is slowing down.[21][22][23][24][25][26] --Ne0 (talk) 04:52, 7 September 2015 (UTC)

I said in an edit summary, "exponential growth", that is true, but realize now that even that could mean flatlined relative growth or a decline.. [Expenential of Win10, faster than the others, would eventually get you close to 100%.] Note, stats only give you relative numbers, I would love to see more absolute numbers. I understand Windows 10 got to 45 million fast, still, I understand Android activates 100 million every day(?). What is most appropriate is the most "current" statistics, but that could be a day (and happen on a weekend.. I notice Win10 (and Win8.1) gets higher on weekends, like Android and iOS, but unlike Win7) or a week, but we've used to have numbers for whole months only. I think, until we have relatively good handsight, showing Win10 stable, for say two months, explaining is in order. comp.arch (talk) 11:26, 7 September 2015 (UTC)

Ne0Freedom, about this edit. The reason for the 7.06% number, that you deleted (a number for a week, somebody put week numbers, instead of the usual month number, and I reverted that and put in a "footnote", for fairness), is that it is higher than the 5.38% month market share stated, in fact it is a 31% higher number (1.68 percentage points). That is a fair increase from the months average. In fact, last whole week was 7.49% (39% higher), so Windows 10 is growing. The growth may be slowing down.. The first week number 3.78%, is also 30% lower than the month average or 46% lower than the last week number.

I didn't check, but I think you'll see no such increase – in a month – for the other operating systems. I do not plan too add week numbers every month, at the end of this one you can put in only the month average, and unless someone puts in a higher number I'm ok, with just that. I do not have a strong feeling about having week numbers, except not as the default (too much hassle to update every week, and some statistics only have month numbers), but I can see that having a lower number for an operating system on the rise being "unfair". comp.arch (talk) 08:50, 15 September 2015 (UTC)

Shows Android with 51.12% from Net Market share numbers.

At least StatCounter shows 62%[27]

[and while Desktop/laptop did use Net Market share numbers, at least now currently shows StatCounter.]

I believe Net Marketshare may be biased towards US (only?) use (seems has been fixed, then I do not know why the big difference, except for only 40,000?; from 2008: "tracks usage across its more than 40,000 client websites. Although these sites are located all over the world, they're skewed towards Europe and North America. [..] they will start weighting their statistics by country in January"[28]), should we switch also for mobile? StatCounter also allows separate number for "mobile" (e.g. w/o tablets, only smartphones) and separate numbers for tablets if we choose:

Then showing iOS still at iOS 66.25%[29], maybe both numbers combined and/or separately? comp.arch (talk) 16:39, 17 September 2015 (UTC)

http://thecloudmarket.com/stats#/by_platform_definition — Preceding unsigned comment added by 186.115.91.4 (talk) 14:32, 23 September 2015 (UTC)