Linux On the Desktop: 0.24 Percent?

Canyon Rat writes: "According to this story, less than a quarter of a percent of desktop users have adopted Linux. The survey was based on web surfers so it may be accurate." Anne Onymus adds a link to an interesting reaction over at lowendmac.com.

Comments

The problem is..

[Rosonowski]

The problem with a web survey is that websites are targeted, much like television, to a specific audeince. That audience is more or less likely to be a windows/linux user, and as such, the results are likely flawed. Kind of like if you tried to do an OS survey on slashdot. Linux would have a much higher rating, would it not?

Re:The problem is..

[ergo98]

But I think that's really their point: If only small engineering circles use Linux, then it's a fundamental fact that the deployed base is small. The dream of Linux, and all other alternative OS', is that the oft stated scenario of "grandma using SuSe" will come true, and naturally grandma isn't going to start her browsing at Slashdot just because she installed Linux: She'll have the same general browsing as most other grandams.

In other words, if you're saying that websites always cater to a certain crowd then I 100% agree (though note that that stat came from information gathered from some 125,000 sites so it'd be less biased than, say, howtouseacomputer.com), however you're conceding defeat if you say that those who use Linux are of a different breed.

Re:The problem is..

[noser]

From time to time I take a look at the pie chart on Google's Zeitgeist page, where they display the relative proportions of operating systems used to access Google. I figure it is a pretty good rough benchmark, as I know they get a lot of traffic from Linux users, so I would expect the representation of Linux on that chart to be high, but we are reading one percent!

It is sobering to see how much the Microsoft browsers have really taken over on the internet. One thing that does make me rest a litte easier about it though is the Mozilla project, and how AOL basicly forces people to use their gecko-based browser instead of IE, so the web is not in too much immediate danger of falling into a MSIE-only club.

I understand that it isn't really reasonable to expect that there would be a large proportion of Linux users though. I agree with some of the other posters that measurements like this are probably more likely to move our way once more people begin to access the internet through Linux embedded devices like cellphones and PDA's, set-top boxes, etc. "Linux on the desktop" probably won't seem like such a big deal as the desktop paradigm begins to fade. I imagine a future where the only people who even use a PC like we do now would be developers or scientists. Regular types will probably surf the web with all manner of specialized devices, and maybe not even think of it as 'surfing', but 'checking the weather', or 'looking something up'.

Re:The problem is..

[ralmeida]

Actually, I kinda wonder about that. I *own* a Linux box, but I don't use it for surfing, I use it as a server (web/ftp/mail/etc).

That's why it's called Linux On The Desktop.

Re:The problem is..

[zmooc]

Proxy logs? That must be even more inaccurate than a webpoll:P Who use proxies? The users that have installed the software they got from their ISP. And what did it do? It set up the proxy-settings. Those that install another OS usually don't bother setting up proxy-settings. So proxies are probably mostly used by newbies. And they usually run Linux. So what the survey actually should say "Linux has 0.24% desktop coverage among users of ISPs that use HitBox". This survey is like saying "69% of my friends runs Linux so the world-wide Linux-on-the-desktop-coverage must be about 69%".

Re:The problem is..

[cworley]

The problem is, browsers in Linux must masquerade as IE on MS/Apple in order to be allowed to render content.

I exclusively surf from Linux desktops (I don't do Windows at all), but I have all my ID strings changed to indicate that I'm running IE on NT because many sites won't allow access to non-IE browsers and/or non MS/Apple OSes.

Re:The problem is..

[telecaster]

The problem is that Linux relies on X, and X sucks as a desktop for the masses. We all know that, and don't try and convince me that an almost 20 year-old architecture is going to bode well these days with my 62 year-old Dad who asks me "Hey, should I use this Linux thing since you like it so much". Let's face it, its klunky.
I told Dad to stick with his Mac...

Here are the problems as I see it (my opinions, I've got about 20 years into this stuff, so I atleast I get to spew my opinion, heh. )

First and foremost:

- No killer desktop application: Yep, thats right. There is no compelling reason for people to use Linux on the desktop. Why switch or even start there? If there isn't an application they can only run on Linux (or run *better* on Linux). BeOS talked about this, never got one... Mac had Photoshop/Quark etc. Windows had Word/Office/IE... Linux needs a similiar app.
Now maybe this application isn't really a desktop application but something like a content creation product, media/video/audio, who knows, but Linux doesn't have one now.

- X Windows: Relying on a HUGE layer for your graphical underpinnings is a big mistake. Remove X. Its too complicated to install, too big and too slow. I could give a hoot about all you so called "Linux Hackers" who say that Linux is for the elite. I look at it as I see it -- it shouldn't be this damn hard, and this damn big! Windows 98 installs in 10 mins -- nice goal to shoot for. Xfree86 my ass, move off that clunker and have a nice thin layer at the bottom... sheesh.

- No Office platform worth caring about: OpenOffice is pretty good, but its NOT MS Office, lets be honest. If OpenOffice started inching more towards the MS side of the world, it wouldn't be such a bad thing -- hey, why not include Evolution in OpenOffice, come on Ximiam, that might help things?? An installable *complete* solution with a good e-mail product. Hmmm... I think OpenOffice is getting there, but its about half the way there... It needs more time and a lot more MS compatibility.

- No good browser: Konq. Nice start. Opera, getting better (although, I could without all the added poop). Mozilla -- please. Netscape -- please. Those two suck on Windows too.
Linux needs a completely IE compatible browser. 100% compatible (there I said it, hate me for it). From a Web designer perspective, to a developer -- the browser choices on Linux are horrible -- we need IE, or an IE clone. I think our best bet is to have Konq. lead the way...

- Fonts suck: Every default font on KDE/GNOME that I've seen pales in comparison to a Mac or Windows desktop/environment. We need good fonts. Freetype2, ok, now we're getting someplace. But thats only recently...

- KDE and GNOME desktop's look like crap: I find every GNOME and KDE environment I try, just looks like junk compared to a Mac or Windows experience. Things just seem patched together, and not completely thought out. Now, I'm not saying Windows or the Mac is the best GUI's around, but boy, didn't anyone learn ANYTHING from those GUI's?!?! I don't see anything in the KDE or GNOME desktops that I would say "hey, thats better than windows", or "wow, I like this better than the Mac".
Personally, I think Linux GUI developers should steal the hell out of both and create a GUI thats even better!

- No good printing: Yeah, yeah, theres CUPS, but you have to GET that and its a pain and driver support is fairly weak. I like the Mac: You plug a printer in to the USB port, click print... boom! printout. No futzing around.

- There are two competing graphical platforms. GNOME and KDE. Thats not helping the overall cause.

- No Desktop "Champion": We need someone (thought we had it in Eazel) that can champion the desktop and create that "killer desktop" that everyone has to have. Linux needs a Linus of the Desktop.

Here are the solutions as I see it:

1. Remove X, standardize on one low-level graphical kernel -- DirectFB anyone?

2. Standardize on one API layer for the GUI, much like Win32, we should have a set of API's that are "God" when it comes to writing GUI under Linux. None of this, Bonobo vs. Qt vs. blah, blah, blah. One standard API thats small, easy and well maintained -- I vote for an API that uses XML/XSLT as its abstraction layer.

3. IBM. IBM should step in and see this as a chance to re-kindle the OS/2 vs. Windows war that was waged from 90 - 92 (which they got their asses kicked).
By dumping some money into the Desktop side of the world (not just the server side), they could create a platform that can beat Windows and Mac.
Hell, I have an idea: Why not port the Workplace Shell right over to Linux, Open Source it and then simply support it as a business platform replacement for OS/2 (if you work at a big company, you know how HUGE OS/2 got during the mid-90's)? Well, in my mind, this could happen again, but only with a platform like Linux under it.
Sure, this is a task, but I bet it would start making believers out of people. Plus, since the Workplace Shell already has a decent amount of applications for Business (3270/5250 emulators), it wouldn't be hard to start eating away at that Fortune 500 companies spending money on XP Professional. IBM could say "hey, why buy that when you can simple get Linux installed on an IBM Desktop for free, oh and we run those same applications... Believe it or not IBM still "kills" at the business level over Microsoft.

3. Gain critical mass with that killer application. Linux need a Photoshop or MS Office.
Anyone got one out there?

4. Stop battling between desktops, choose one or create a new one and settle on that -- but don't use X.

Re:The problem is..

[Sj0]

Opera pretends to be IE5, doesn't it?

Besides, the numbers will lie. I use Linux and BeOS at work, and I use Windows at home because of technical reasons (winmodem in a laptop....). I also use Linux at home, but as a part of my stereo, not as an internet PC(an old p200 with a decent CD-ROM drive makes for a terrific MP3 and CD player, and I can access it remotely through either telnet or an X-server -- p200s are quiet!) so every one of these OS's will hit the logs. Not to mention Atheos and QNX, which I tried recently, and I even log in using Arachne in DOS sometimes.

The point of all this? I forgot. I think it had something to do with misleading web logs. :)

Re:The problem is..

[looie]

If its got anything to do with Hitbox, it's not sites that are polled but ISPs... [HitBox take proxy logs from ISPs and process those - generating the stats and providing clickthrough information to sites].

Hitbox does nothing of the kind. It uses JavaScript included on the site pages. Every time the page is opened in a browser, information about the visitor is returned to Hitbox.

Because it uses JavaScript, it can get exactly the same useragent information that would show in the server logs -- and more.

This method of retrieving information is an accurate reflection of what web browsers are doing at a particular moment, with caveats. They must have JavaScript enabled in the browser and usually they must accept cookies. (About 4-5% of browsers disable JavaScript and 10-15% disable cookies.) The kinds of sites that are using Hitbox and its competitors are likely to be commercial and windows-centric. Also, I think it is worth noting that an estimated 40% of web browsing is done from work, where people have access to high-speed connections not available at home. This naturally skews the results, since offices are presently far more likely to be Windows-based.

This type of traffic analysis is rapidly becoming popular among the major enterprise-level sites, such as Major League Baseball, Ticketmaster and so forth. If you have a Windows machine, you can run bugnosis for a while and see what I mean.

mp

Re:The problem is..

[MSG]

The problem is that users don't really know what they want, they only know what they're told they should want. The parent post may be a troll, and it may not, but it's full of frequently posted bullshit that people need to stop believing.

don't try and convince me that an almost 20 year-old architecture is going to bode well these days

OK, why are you even considering Linux then? It's a 10 year old OS replicating a 30 year old architecture. It can't *possibly* be any good, right?

Modular, extensible software isn't new. X11 was designed that way years ago. The only problem has been the proliferation of slow, monolithic implimentations. XFree86's implimentation is much much better than many in the past. X11 itself is a fine drawing layer, even if libx11 is a bitch to interface with.

There is no compelling reason for people to use Linux on the desktop

Maybe not. I don't know. My mom's been using it for 3 or 4 years, since before Windows had ICS. That was the killer feature. Even after that, Windows didn't have a good personal firewall. Even still, it's vulnerable to about a million virii that will never affect her computer. Everyone has something that they desire from their computer...

Xfree86 my ass, move off that clunker and have a nice thin layer at the bottom

There's that frequently quoth bullshit. X11 IS THIN. Thin == little memory: X11 works on Compaq iPAQ's in something like 2MB of RAM, and provides better services than the Linux frame buffer. Thin == low level, which is what most *real* X11 programmers bitch about. X11 is so thin that it provides a mechanism without any policy! That's its design goal. It's just the mechanism, so policy can be decided by anyone who needs a graphical environment without rewriting their drawing layer from scratch. As GUI's evolve, and their internal designs change, X11 will always be there for them to be built upon, without rewriting the low level hardware interface bits.

Why does everyone bitch about X11, but no one ever thinks that Linux should be replaced with something that isn't 100MB of source, and 20 MB of binary? What? No one thinks that an OS would be much faster if it were "thin"?

Remember, THIN == few features. X11 provides all of the features that you need to draw in an extensible architecture, without anything that belongs somewhere else.

Linux needs a completely IE compatible browser.

Compatible in what way? If browsers on Linux aren't compatible with IE, then the fault lies not in the Linux browser developers; it's with MS. There *IS* a standard for this crap, you know? It's all written out, and anyone should be able to understand it. Mozilla and Opera are far better at being standards compliant than IE, so why don't you bitch at MS. Why should we have to degrade from written standards to implimentation standards that are likely to change as IE does?

Fonts suck

*GOOD* fonts are really hard to create, and therefore very expensive. Perhaps you would like to develop some? Or maybe fund their development? Not that you're wrong here... The fonts we've got would be a lot better if they were scalable and hinted, but that's where we loose out.

Re:The problem is..

[GiMP]

In debian, X is really hard to install..
apt-get install x-window-system

There are even programs for X that will let you do that via clicking little buttons.

Very hard.

I suppose mandrake is slightly different:
rpm -i XFree86-*.rpm

Re:The problem is..

[ahde]

I wonder what the ratio of linux users is after 5pm, as opposed to during working hours?

I know my daytime hits are usually Mozilla/NT (or Lynx/Solaris if the boss is looking -- or the animated gifs get too annoying.) Of course, at night I'm usually working on something, so I don't have time for Slashdot.

Re:The problem is..

[jackbox]

Does anyone recall how NT 3.1 was supposed to be the desktop follow-up to Windows 3.x back in... like, 1994? When did NT finally achieve notable penetration on the desktop? Maybe around 1998? Maybe only last year with Windows 2000?

I don't have exact stats. My point is: it's taken at least 4-5 years for Microsoft to push their own "industrial strength" OS onto the desktop. (Win 9x was a stopgap measure because people were sticking with Win 3.x and not moving to NT.)

Whoever is doing doomsaying on Linux by claiming "it's been years and it's not on the desktop yet - therefore it's a loser" has been brainwashed by the MS PR spinners.

These changes take time. And Linux has made incredible progress considering the many hurdles it has to overcome in the marketplace. Now is no time to stop.

er?

[yatest5]

The survey was based on web surfers so it may be accurate

Er, or it may not. Does the web surfing population necessarily represent all computer owners? I would suggest that web surfers are slightly more likely to be tech-savvy and therefore web-surfers will have a higher percentage of Linux desktop use than non-web users. So the figure may be even lower.

Slashdot Stats

[WildBeast]

Maybe cmdrTaco should post the Slashdot stats of it's users OS

How to craft a response.

[glowingspleen]

[Insert Pro-Linux Outcry]

[Insert Rambling Out-Of-My-Ass Reasons why Survey Can't Be Correct]

[Insert Attack on Microsoft]

[Insert Short Insult To Silly Un-learned Users Who Don't Know Better]

[Insert Reminder That Survey Can't Be True]

[Close with Name, Followed By Witty Anti-M$ Slogan, Being Sure To Substitute A Dollar Sign For The "S" Because Doing That Is Inventive And Hilarious]

Re:How to craft a response.

[release7]

I've had my Linux box running for over 9 1/2 years. Try that with you Windows 3.1 machine. I'll tell you it can't be done.

Anyway, I've seen surveys by this company before and they've actually had to retract some of the statistics they've published in the past because of some oversights. Notice how they don't provide a breakdown of the 125,000 sites in the survey---just a little suspicious, don't you think? I've talked to almost everyone in my lab and at least 20% are using Linux.

In the long run, about 15 years, I predict Microsoft will be toppled. Let's face it, their software just plain sucks! Once the gov't wakes up and realizes they are nothing but pure monopoly, their market share will get washed away like a brillo pad on soap scum.

However, I am sick of lusers who are too lazy to want to learn anything new. It's like, hello! God help us if they actually put in just 2 hours a day for a couple of months into learning a new user interface so they could duplicate what they can do with Windows, they'd have a kickass OS that never crashed and can simulate many of the same things you do in Microsoft Office. Stupid lusers. Keep buying Microsoft products but don't come whining to me when Bill Gates owns the mortgage on your house.

Anyway, I just wish /. would be a bit more responsible about the articles it posts. The stats are obviously skewed.

preZZure--->Friends don't let friends buy M$.

Re:How to craft a response.

[onion2k]

Of course MS is bloatware. Now, what did I do with my 7 cds of Debian..

Here we go...

[forgoil]

I bet that there will be at least 100 posts saying that you can't trust this kind of data, that it's complete bollocks and yada yada yada Linux is so good it will for Bill to eat Linus used shorts.

Please don't care about that article, it's not interesting really. It's not really news. We all know what we use ourselves (XP and linux in my case) and I suggest that our time should be spent on something better than surveys and such things.

Writing serious and useful documentation for linux for instance, and putting it into XML and making it readable and searchable in different applications (such as the exellent Konqi, the only other browser besides IE I would ever dream of using). Go do that instead of reading all the pointlessness that this news consists of.

Where are the Appliances

[dunstan]

I think we can all start from the premise that these statistics are:
a) flawed
b) backward looking

What would be more interesting is some insight into where browsing is headed. For example, there will be some sites which will attract mobile traffic much more readily than others - traffic updates, or train running info, or today's tube (as in London Underground) breakdown. Then we are going to see amounts of traffic from appliances such as set-top boxes.

But then I suppose "We produce rubbish statistics" won't be as headline grabbing as "You Linux folks are all losers".

Dunstan

Lynx will never show up on these stats

[andyr]

Since the stats are gathered in one place, a hitcounter, my lynx-browsing will never be tallied, as I do not download those little GIFs. Even under Galeon I flag it to not download pictures from other sites - so I will not show up there either.

Cheers, Andy!

Will Get Faster then More Popular

[KarmaBlackballed]

I tried using Linux KDE as a desktop last year and was disappointed with the speed of the graphical interface. I could watch the dialogs painting and this was on a 900MHz machine.

This is not an issue with Servers.

I, like most users, expect performance to be at least as snappy as on other systems using comparable hardware.

As hardware gets faster, the GUI sluggishness will be less apparent. That along with the advent of more mainstream compatible apps will make it more prevalent as a desktop OS.

.24 percent?

[Minupla]

Damn, that's much better then I would have guessed. Think about it, that means, one out of every 400 users is using linux as a desktop system. I'm impressed, honestly, I didn't know there were that many clued lusers out there.

Wow!

Representative data

[Cpyder]

well, as the mac article points out, the Hitbox users aren't really representative... But what about Google??? About everybody uses google... So let's see what they have to tell:
Windows (all versions): 93 %
Macintosh: 4%
Linux: 1%
Other: 4%

Detailed figures on browsers and operating systems on their site. I think Google can be considered quite representative, not?
(posted with Konqueror / Linux)

Re:Representative data

[nEoN nOoDlE]

That only proves that Linux users don't search for stuff online cause they're so knowledgable while Windows users have to resort to search engines to get where they're going! Your statistics are LIES! Or maybe Windows users just search for more porn.

Survey says...

[adubey]

I've had some training in statistics, and I see a number of problems. First, the slashdot editors are making the perennial journalists' mistake of misinterpretting statistics. Statmarket only claims to be measuring web client usage, and doesn't make any claims about the desktop market in general (at least from what I saw).

In terms of the study itself, statmarket admits that the sample is "self-selected" rather than randomly selected. This results in a biased sample. In particular, since they are offering a service to business users, the sample is likely biased in favour of business sites. The bias is then against more "arty" or technologically-oriented sites, resulting in lower-than-expected numbers from Macintosh and Linux users. It might also be biased against home users.

That said, while the survey may be off by an order of magnitude, I wouldn't expect it to be off by more than an order of magnitude. Most other surveys don't put Linux usage at more than 2 or 3%

Our Stats ...

[SirSlud]

.. as taken from our counter for the site www.pseudodictionary.com, a site that has NOTHING to do with computers.

Win 98 80178 (45%)
Win 2000 33183 (18%)
Unknown 17948 (10%)
Win NT 15051 (8%)
Mac 13085 (7%)
Win 95 11717 (6%)
Linux 2459 (1%)
Win 3.x 1055 (0%)
Unix 761 (0%)
WebTV 226 (0%)
OS/2 24 (0%)
Amiga 4 (0%)

The scariest thing is that win98 is still 45%. If not being part of that 45% is wrong, I don't wanna be right, baby!

Re:Our Stats ...

[JiveDonut]

Why is it scary that 45% run win98?

Windows 98 is a perfectly good OS for home users. I have two linux boxes in my house that I use for programming, file serving, wireless network, etc.

Guess what the third one runs? Win98.

Browser Identification Strings

[Rushuru]

The other problem that may drive *nix browsing market share is that there is a gazillion browsers who all have different identification strings. Very often, poorly designed stats system will not even notice that a given browser is actually a linux one, and will classify it as unknown.

Also, many poorly designed sites ony lets people with Ms IE 4 or Netscape 4 visit the site. Opera, mozilla, konqueror users have to fake the identification strings to be able to see the site. And, as a matter of fact, I know several people who have set their browsers' id string to be IE like, to avoid troubles.

There's no arguing that Linux's desktop market share is far lesser than that or windows and mac, but I do think and hope it's above 0.24%

Where they get their stats.

[Carnage4Life]

According to their Research methodology page

StatMarket publishes statistics based on the combined data from tens of millions of daily Internet users visiting the tens of thousands of sites that use WebSideStory's HitBox Enterprise and other HitBox Web audience analysis services. HitBox is an outsourced Web site measurement and analysis service that provides real-time statistics about online visitor behavior.

...

While the 125,000+ Web sites worldwide that HitBox monitors are self-selected, StatMarket's figures are culled from more than 50 million unique visitors who visit those sites every day

Re:Where they get their stats.

[zmooc]

Still, that doesn't have to be a guarantee at all. It is very well possible that the sites that use hitbox are for some reason visited more by windows users (newbies?) than other sites. Sites with content that's more interesting to us geeks usually don't use hitbox (slashdot, google, blah). Porno sites for example work much better in windows (movies!). I'm not saying that Porno sites use HitBox more, but it's just one of the many examples. The only way to do such a survey right is by picking a few people randomly and then contact them by telephone. And then it's still possible that users of OS A are more willing to cooperate than users of OS B:)

Circular circle? :)

[ishark]

I like the answer on lowendmac. Not the article, but the statistics. Beside that, could it be that we're witnessing the same "netscape effect" of the web? The article says that lots of web developers use those statistics to build sites. Translation: they only target IE. I can believe this, since I use galeon and I often have quirks in commercial sites. Now, if your site works well only with IE I'm not surprised that 98% of the visitors use IE.... Just like netscape-enhanced sites used to justify their attitude by saying that "90%+ of the visitors use netscape"....

(Note: I use Windows == IE. I don't know the statistics of Ns/Mozilla/Opera vs IE on windows, am I guessing right that they are a tiny %?).

I'm not registering any Linux User Agents

[Vic Metcalfe]

I host web sites. Here's a webalizer chunk on User Agents from a piece of November I called up just to see if it was close:

Top 15 of 5486 Total User Agents
# Hits User Agent
1 200870 9.80% Mozilla/4.0 (compatible; MSIE 5.0; Windows 98; DigExt)
2 169779 8.29% Mozilla/4.0 (compatible; MSIE 5.5; Windows 98)
3 161822 7.90% Mozilla/4.0 (compatible; MSIE 5.5; Windows 98; Win 9x
4.90)
4 73991 3.61% Mozilla/4.0 (compatible; MSIE 6.0; Windows 98)
5 72181 3.52% Mozilla/4.0 (compatible; MSIE 5.01; Windows NT 5.0)
6 70011 3.42% Mozilla/4.0 (compatible; MSIE 5.5; Windows NT 5.0)
7 63082 3.08% Mozilla/4.0 (compatible; MSIE 5.01; Windows 98)
8 54560 2.66% Mozilla/4.0 (compatible; MSIE 6.0; Windows 98; Win 9x
4.90)
9 46702 2.28% Mozilla/4.0 (compatible; MSIE 6.0; Windows NT 5.0)
10 43299 2.11% Mozilla/4.0 (compatible; MSIE 6.0; Windows NT 5.1)
11 41167 2.01% Mozilla/4.0 (compatible; MSIE 5.5; Windows NT 4.0)
12 37536 1.83% Mozilla/4.0 (compatible; MSIE 5.0; Mac_PowerPC)
13 33620 1.64% Mozilla/4.0 (compatible; MSIE 5.5; Windows 95)
14 29224 1.43% Mozilla/4.0 (compatible; MSIE 5.0; Windows 98)
15 28778 1.40% Mozilla/3.01 (compatible;)

Ok, #12 says it is Mac, and #15 doesn't say at all. I host the primary site for the UNIX Socket FAQ, which you would expect to bring in a significant chunk of Linux users, but it isn't even in the top 15. Maybe users are masking their user agent? Maybe some, but not many.

Take from this what you will, I just thought it was interesting...

Re:I'm not registering any Linux User Agents

[ishark]

15 28778 1.40% Mozilla/3.01 (compatible;)

Ok, #12 says it is Mac, and #15 doesn't say at all.

It's probably junkbuster, which screws up the user-agent field with some obscure old stuff.

My own stats

[BillyGoatThree]

Having recently set up my own domain, I have some stats. The site is non-technical (book reviews) so should represent a relatively random sampling of users.

Total sample: 10000 hits

Windows 98 is way in the lead with 46.5%.

ME comes in at 15.9%

95, 2000, MacOS and NT are all roughly equal at 9.1, 8.8, 7.4 and 6.1 respectively.

XP has 3.6%.

Linux has 1%.

(there are a few others, including "Unknown" so those won't add up to 100)

Considering the differences between some of those Windows OS's, that's fairly diverse. What's more disturbing to me is the following:

IE has 81.3% of the browser stats, Netscape has 16.8%. "Unknown" and Opera together have less than 2%. WebTV brings up the rear with a measly 8 hits (0%) and that's it. No other browsers.

Considering that desktop OS is largely irrelevant to the Internet whereas browser is VERY relevant, this points out a disturbing trend: Microsoft Owns The Client-Side of the Internet.

Yep, it's pretty small.

[infiniti99]

In fact, the only people I know that use Linux on the desktop are all developers (myself included) except for one guy.

Not that this is a problem. For us developers, Linux/KDE is a wonderful system to use. It all comes down to needs. Does the average user need multiple tabbed sessions in Konsole? No. Does he need to be able to play Dark Reign? Yes.

Unfortunately, the "games" problem is not one that can easily be solved. Most software you buy at the store is only for Windows, and I've heard more than one person say that Linux can't succeed with normal users without it being able to run Windows programs. IMO, making it a requirement of Linux to run Windows software (a la Lindows) is too much to ask. Not only is reverse-engineering difficult, but companies these days are making it harder to pull off. And sometimes, it can even be illegal (see DMCA).

So is all hope lost? What can anyone do? Linux is basically done.. Linus said so himself. Now the focus is on the user. Well, what is left for KDE? It is already more configurable than Windows. Ok, so that's done. Now what? If we're done, but we have no users, there is obviously a problem somewhere.

It's the apps. Linux is not scary anymore. The "one guy" I mention above knows nothing about coding, but uses Redhat just happily. But why can't he play his games? And where is Adobe?

We've done all we can do. I think it's just a waiting game now. I'd like to see some improvements with more general (non-distribution specific) software installation. And for video drivers to be kernel controlled, and have X just ride on the framebuffer. But issues like these won't stop average users from using Linux. Just ask a normal Windows user why he would not want to switch to Linux. It will come back to the apps.

Linux has only become more popular, not less. More companies join in the game as time goes. Sure, some have left, but at the end of the day the number is bigger. The general computer user will get his games and his apps.

In the meantime, everyone just continue doing their thing.

Their stats are wrong ... here's why:

[Skapare]

websidestory.com and statmarket.com are basing their statistics on their web tracking technology through the use of advertising. The problem is, they use web bugs (see here, here, and here) to accomplish this. Windows users typically do not take actions to inhibit these web bugs, but Linux, BSD, and even many other Unix users do. There's software out there to help, too. Those who do block these web bugs, or all the hitbox.com sites, as I do, won't ever be counted.

Statistics based on web bugs should never be counted to determine platform penetration. Instead, actual HTML loads from a wide variety of real sites should be used, and the distribution variations show, too. I'm sure Slashdot gets more Linux and BSD just because of what it is.

Find out what other sites that /.ers visit, then get platform stats from those sites, and only for their main page HTML hits (not for images or ads or anything else). Then check the variation of that.

I had to go remove them from 6 different blocks in my network to just to view the linked page.

My version of Linux reports itself as MSIE 5.5

[TheFlu]

I use Junkbuster and alter the browser info headers to fool sites into thinking I am using IE 5.5, instead of Opera on Linux. I do this because lots of sites have the annoying practice of thinking that a non-Microsoft browser won't render their pages correctly, but it usually works fine.

0.24% isn't bad!

[stonewolf]

Considering everything that stops people from using Linux on the desktop and the huge lead that MS has in established customer base and marketing, 0.24% isn't at all bad. It is a lot higher than I thought it was.

Truth is that there are huge barriers to using Linux that can only be blamed on Linux. A simple example. I recently installed NVidia graphics cards on both my Win98 and Red Hat 7.2 desktop machines. The hardware installation was the same of both machines. The software was another story...

Even though NVidia included binary RPMs for linux drivers on the disk those drivers were useless because they were for a different version of the kernel. So, I had to down load the drivers from the NVidia site and install those. Of course, even though they claimed to be compiled for the same RH kernal as the one I was using they didn't work either. So, I downloaded the source tar.gz files and compiled them and the installation went just fine. Then I had to edit the XF86Config-4 file and then figure out that for some reason AGP just wasn't going to work... and most of an afternoon and an evening later I had a working high performance OpenGL monster of a Linux box.

The install of the Windows drivers took about 5 minutes, but since I was at NVidia's site anyway I down loaded the newest drivers, installed them, and started playing games. Total time, counting the down load, about half an hour.

Did I mention that I spent 5 years as an X server developer in another life? So, I have an above average knowledge of the server. Did I also mention that I have several computers all networked so that after I lost my desktop and web browser (no graphical interface == no browser for most people) I was still able to access the NVidia web site and down load drivers and help files? And when they lose their desktop they are completely helpless.

All in all, just the hassle involved in loading an accelerated graphics card made by the most pro-linux graphics card manufacturer in the world (MHO) is enough to keep anyone who is not a hard core geek from even considering using Linux.

Lets face it folks, right now Linux is still actively hostile to the average human being. The fact that drivers have to be recompiled to match the kernel makes Linux actively hostile to all device manufacturers. And, that makes Linux hostile to all software developers that depend on specialty devices.

I'm about as pro-linux as anyone I know, but that doesn't change the fact that a company like NVidia needs to provide a 68 page installation manual with the Linux drivers for a card and doesn't need to provide any instructions for the Windows drivers for the same card.

Like I said, 0.24% isn't bad. On my web site I see closer to 1.3% Linux, 1.9% Mac, and 91% Windows.

Stonewolf

B.F.D: There's still a monopoly out there fellas

[Locutus]

Come on now, how is anybody supposed to get Linux out on the desktop if nobody worth a hoot can pre-install it? Not even in a dual boot configuration. I've got two friends who went out and bought $1500 PC's to do email and web surfing. Only some of the fringe players like Ellisons company,etc. do Linux such that consumers could use and how do they compete in a Windows-only press world?

Hell, OS/2 had/has a much higher usability rating, IMHO, yet only in one country in the world could IBM get pre-installs, Germany. I'd heard that OS/2 had 25% of the desktops in one year. BeOS was available for free to anybody who wanted to pre-install. They couldn't. Can you say monopoly?
BAD Monopoly?

Linux will remain out of the desktop space as long as Microsoft can hang anybody who lets Linux get close to a pre-installed Windows box. PERIOD. No operating system in existance today or tomorrow will break this strangle hold cause users take what is pre-installed.

IMHO

hitbox ignores client without javascript

[gregor_b_dramkin]

from the FAQ at http://www.hitboxcentral.com:

"HitBox will record nothing if:

• The referring site does not link to the page containing the HitBox Main Code or you have not installed the HitBox Main Page code properly (Free HitBox V5 only).
• The HitBox code does not fully execute.
• The reverse DNS lookup is unable to complete.
• The visitor's JavaScript is disabled. "

I'd say that a Linux user is much more apt/able to turn off Javascript in their browser than an IE user.

50,000,000 "unique" users is nothing

[Erris]

When you consider that 125,000,000 people in the US alone have internet access, 50,000,000 so called unique users is not so big. I have to wonder if the tool they used to determine uniqueness is a MSIE only thing.

Web surveys are not a good measure anyway. Linux users may have something better to do than surf comercial websites all day. Consider the number of Sun users reported. Linux is used by the physics community for workstations. I doubt any of those "desktops" got counted. They might not even have a browser (gasp!), or a GUI for that matter.

Great News!

[nathanh]

A number of posters seem to be moaning because the figures range from 0.25% (HitBox) to 1% (Google). I see wild theories attempting to discredit the figures and additional arguments trying to justify why the figures should be higher.

Wake up to yourselves. Almost 1% is great! The current estimate for the number of Internet users is 513 million people (according to NUA http://www.nua.ie/surveys/how_many_online/). So even taking the lowest figure from HitBox that's 1.3 million people using Linux as a desktop. It could be as high as 5.2 million people if Google provides a better sample.

But that's only desktop users! I will claim (and I think many people would agree) that the percentage of Linux *servers* is much higher than the percentage of Linux desktops. I can't guess how many machines this equates to (I don't know the relative number of desktops to servers, or the percentage of servers that are Linux) but it's going to be more than zero.

It's brilliant news that Linux usage is this high. Every single person that uses Linux is a success story for Linux. There's no need to have huge marketshare, or be the dominant player. You just need a critical mass of users and several million users is definitely a critical mass. The early years of Linux had just a few 100 users and it was enough to propel the snowball forward. Millions of users equates to an avalanche!

Keep reminding yourself, just by using Linux you are helping to make Linux better. You are another person who can help a newbie. You are another person who might buy a book or CD and thus indirectly fund a developer. You are another person who might find a bug, suggest a feature, write some documentation, or perhaps even write some code.

You are part of the Linux community, and even the most pessimistic figures suggest that this is a community with MILLIONS of members.

Not conspiricy theories

[einhverfr]

HA, and the conspiracy threories come out already! I really don't think any one has motive to pretend that Linux has a lower end-user figure than it has, when it is such a minuscule figure - or are you suggesting that the figure is out by 20-50%???

The IDC states that 2% of corporate desktop users are using Linux. This is rouchly 8 times what this survey reported and I would think that there would be slightly more home users using it now than corporate users.

My estimate is 2-5% of users are using Linux. Still small but not as small as 0.24%...

NVidia dosn't get Linux

[Royster]

All in all, just the hassle involved in loading an accelerated graphics card made by the most pro-linux graphics card manufacturer in the world (MHO) is enough to keep anyone who is not a hard core geek from even considering using Linux.

If they were really so pro-Linux, they would have Open Source drivers so that you wouldn't have to jump through the hoops that you did. Place the blame where it belongs -- with NVidia.

X11

[einhverfr]

X11 is good technology. It is extensible and flexible. It is thin, with a low memory footprint, and has layered features, much like TCP/IP (actually, the analogy could be followed here quite well).

The problem is not actually with X at all. It is the office suites which are useful in small settings but are not enough of a development platform for the enterprise (this is where MS rules, security aside).

I remember when the Linux desktop was clunky (that was a year and a half ago) and now it is much more smooth, without getting rid of X. KDE and GNOME have both come a long way in that time. I am waiting for the office suites to do likewise.

The basic thing is that I think that the study has underreported Linux by 8 to 10 times (still a small percent though), but I don't think that the problems are as severe as they used to be.

Ways the measure is inaccurate:

[DunbarTheInept]

1. It only measures sites with their tool being used.
2. Users can lie with their user-agent strings and sometimes *have* to to get into a site at all.
3. Even when the user-agent string is honest, the user might have javascript disabled by default. Those who don't use Javascript don't get counted into this counter. Now, who is more likely to have JS turned off - a Windows user or a Linux user?
4. Web hits per day cannot measure computer *ownership* percentages, only user *traffic* percentages. Users with innefficient web browsing habits will tend to score much higher in the measure than those with sensible web browsing habits. The next time you see some guy clicking back and forth between pages instead of opening two browser windows, think "HitBox thinks there's 10 times as many of him as there are of me".

What hitbox does isn't necessarily wrong. It is a useful thing to know how much of the web traffic is coming from what users. It's just when this data gets misinterpeted by hack reporters that there's a problem.