Occam's Razor
|
| Author | Content |
|---|---|
| Sander_Marechal May 06, 2009 10:25 AM EDT |
It's a good piece, but I don't agree with GreyGeek's assertion at the end:Quoting:It is quite obvious that NetApplications latest "report" is merely Microsoft's continuing attempt to control the news about Linux's success in replacing Windows on the desktop...." The way NetApplication measures leaves plenty of holes to explain the 1%, starting with the fact that NetApplications doesn't track websites frequented by most Linux users. No need to conjure up a Microsoft conspiracy here. Apply Occam's Razor. The study is flawed. Also, "Don't attribute to malice that which can adequately be explained by stupidity". I don't know who said that, but IMHO it also applies here. |
| softwarejanitor May 06, 2009 2:22 PM EDT |
Too bad that a few Linux orriented web sites can't somehow get onto NetApplications's list... LXer would be a great addition, but even sites like Slashdot which are these days largely infested by pro-MS trolls would still show a much larger number of Linux users than 1%. |
| jezuch May 06, 2009 3:10 PM EDT |
Quoting:"Don't attribute to malice that which can adequately be explained by stupidity" That's Hanlon's Razor. http://en.wikipedia.org/wiki/Hanlon's_razor |
| Sander_Marechal May 06, 2009 4:05 PM EDT |
So many razors. I must make sure not to cut myself! |
| gus3 May 06, 2009 4:12 PM EDT |
If someone contradicts Netcraft figures to such a great degree, they need to be prepared to explain why. |
| caitlyn May 06, 2009 5:39 PM EDT |
I can explain why. Netcraft's number is based on websites using their specific tracking software. That software only works on web servers running Microsoft IIS. Apache has a majority share in the web server market but sites running Apache can't be counted. How many Linux-oriented sites use IIS? |
| gus3 May 06, 2009 6:19 PM EDT |
http://uptime.netcraft.com/perf/reports/performance/Hosters?... Okay, I'm not able to reconcile "Windows-only tracking software" with this chart. It has 7 Windows platforms, two unknowns, two F5 Big-IP (whatever that is), and the 29 remaining are all *nix. |
| KernelShepard May 06, 2009 6:34 PM EDT |
While it is true that a huge percentage of web servers run Apache on Linux as opposed to Windows Server + IIS, adding Linux-focused websites to the list of servers in the study isn't necessarily going to get you a representative reading of the Linux desktop user population. However I do agree that the websites in the study do matter. For example, what would be really interesting would be to base the survey on popular commercial websites. I'm thinking along the lines of, say, Amazon, (maybe) Netflix, etc (I'm sure you get the idea). It's not that interesting what OS users who visit microsoft.com, apple.com or my-favorite-linux-distro.com are running, because odds are that they are running that OS. The only reason this is even the slightest bit interesting is if you are a commercial vendor wanting to make sure you are reaching the widest audience possible, but the way to discover this information isn't by reading meaningless Netcraft surveys, it's by actually logging your customers browsers (odds are this is already done anyway) and then writing some script to sift through the log file(s) and figure out what /your/ customers are using. If all your customers are using Windows & IE6, then don't bother wasting time trying to make a product (or website) work with Linux + Konqueror, because obviously it'd be fruitless. As I said elsewhere, it doesn't matter if Linux is on 1% of all desktops worldwide or 99% of the desktops worldwide, it only matters that you intelligently make your stuff work with the market that is most interested in your product(s). Even if Linux is only on 1% of all desktops worldwide, if that 1% constitutes 100% of your market, then there's no sense targeting Windows unless Windows users might be interested in your product too. Same for the other extreme. But to find that out, you need to do market research of your own - each product is going to have different amounts of interest for different groups. There's no one-size-fits-all survey you can refer to. |
| Sander_Marechal May 06, 2009 6:41 PM EDT |
Quoting:If all your customers are using Windows & IE6, then don't bother wasting time trying to make a product (or website) work with Linux + Konqueror, because obviously it'd be fruitless. Catch 22. If your website doesn't work on Linux, you'll never get any linux customers in your log files. 99.9% of your visitors are not going to complain. They simply go to a competitor who does support their OS. |
| hkwint May 06, 2009 6:45 PM EDT |
Quoting:If someone contradicts Netcraft figures to such a great degree, they need to be prepared to explain why. Given the fact that Netcraft itself doesn't want to disclose where both their data and money comes from (Microsoft is one of their clients), I'd say that's bollocks. Netcraft is the one who should explain their data, not the ones doubting the data. BTW Xitimonitor (now AT internet institute?) shows 1,24% for Linux: http://www.atinternet-institute.com/fr-fr/equipement-interna... and such a 25% difference in market share seems rather big. Recently, a chart showed by Ballmer gave 5% market share to Linux. So what's the truth? |
| KernelShepard May 06, 2009 6:48 PM EDT |
Sander: yes and no. If you notice a competitor targeting the same audience you are... PLUS some people using a different OS and the people on OtherOS are flocking to them in droves, then you obviously need to reassess what platforms your products/website are built for. But that can all be gathered via real market research. I didn't mean to suggest you use website stats alone ;-) Also keep in mind that Linux+Konqueror (as an example) users aren't going to know that your website doesn't work with that setup until they try it ;-) |
| Sander_Marechal May 06, 2009 6:52 PM EDT |
Quoting:what would be really interesting would be to base the survey on popular commercial websites. Even that wouldn't work well I think. Just take a look at the average Linux user. What are they? A tad geeky, technically inclined and well versed at finding their way around the internet. I'd say we are less inclined to use the big commercial sites and more likely to use niche sites that suit us better. I would even go as far as saying that the only large commercial website that would give a somewhat accurate sample would be Google. |
| Sander_Marechal May 06, 2009 6:57 PM EDT |
Quoting:If you notice a competitor targeting the same audience you are... PLUS some people using a different OS and the people on OtherOS are flocking to them in droves You'll never know. Unless the website in question is tech oriented or requires special technology then you can't tell if a website caters to Linux users or not. A site about growing your own vegitables or your local gym aren't going to scream Linux compatibility from their websites. They simply work or they don't. There's no way of knowing that they specifically cater to Linux people. And you'll never know how much Linux visitors your competitor gets without getting access to their web statistics. |
| herzeleid May 06, 2009 7:02 PM EDT |
I know lots of linux users who e.g. shop @ amazon.com or search on google - a look at those server logs would probably be a lot more meaningful than looking at the stats from some obviously slanted site e.g. microsoft.com |
| KernelShepard May 06, 2009 7:10 PM EDT |
Sander: yes, Google would absolutely be a great example of a site that could actually shed some half-way decent light on what the actual linux desktop install base is. However, I think that if Linux desktops really are only used by "tech geeks" as you say (and I have no evidence to disprove your assertion), then it will never have much of a real desktop market share simply because the "tech geek" population is tiny. Maybe you are right, I don't know, but if you are right, then it's even less reason to get all upset (not that you are specifically) about about the 1% figure ;-) Think about it... I heard somewhere that worldwide there are only about 9 million programmers. What is world population? Something like 7 billion? That would mean that even if all programmers in the world ran Linux on the Desktop, Linux would only have 0.13% desktop market share. What's the ratio of "tech geek" to programmer? I doubt it's more than 10:1 (which would give us the >1% mark assuming every techie in the world ran Linux on the desktop). Of course this is only 1% of world population, not computer-user population, and I have no idea what that ratio is. In an event, it's entirely reasonable to conclude that in order to take much more than 1% of the computer-user market share, Linux will have to appeal to more than just tech geeks because a vast majority of computer users are not and do not care to ever be tech geeks. |
| Sander_Marechal May 06, 2009 7:43 PM EDT |
Shepard, a couple of things: First off, tech geek != programmer. There are many, many more tech geeks than programmers. Also, I refuse to believe that there are only 9 million programmers out there. That figure seems to be off by at least an order of magnitude. As you say, 9 million makes about 0.13% and I can't believe that such a tiny market share could support the large amount of programmer oriented websites out there. I do believe that most Linux users out there are tech geeks. Not all, not by far. And we're actively out marketing and growing in the non-tech non-geek market pretty fast and that's a good thing. But the browsing habits of the average Linux user (or even average geek) is very, very different from the average internet user. |
| caitlyn May 06, 2009 8:03 PM EDT |
"I do believe that most Linux users out there are tech geeks." That certainly was the case a few years back. It's been changing for quite a while now and I no longer believe it's true. A few million Linux netbooks have been sold and I think a lot of the folks using those aren't technical at all. I also know more and more businesses are using Linux on the desktop. While the percentage is small it trickles down to home users. Folks who are forced to use something at work may discover they like it and start using it at home. I remember how many IBMers used to run OS/2 at home not because they had to but rather because they had gotten used to it and found it did a lot more for them than Windows with less aggravation that Windows. I think the tech geek:Joe/Jane Sixpack user ratio is certainly higher among Linux users. I just don't think it's an overwhelming percentage at this point. |
| Sander_Marechal May 06, 2009 9:03 PM EDT |
I think it is, but I define "tech geek" pretty broadly here. In this case, it doesn't even necessarily mean linux geek or even computer geek. Millions of netbooks are sold, but what group do they mostly sell to? People who love tech toys and such things I think. People who already have one or more computers and who'd love to have a netbook on the side for some couch surfing or commute surfing. People who want to be online. A lot. And those are IMHO the people who don't rely on big, commercial websites to get their daily fix. They are very experienced surfers. They get their fix from a number of smaller, better targetted sites that deliver exactly what they want. Sites that don't get tracked by NetApplications. |
| KernelShepard May 06, 2009 10:18 PM EDT |
Sander: well I did give an estimate of geek:programmer being 10:1 ;-) The 9 million estimate came from Joel Spolsky, I'm not sure where he got it but was talking about it in one of his recent podcasts. I'd have to go back and listen to it again but I'm not that interested to be honest. |
| tracyanne May 07, 2009 12:04 AM EDT |
Wet finger in the air time again. We are getting a number complaints/commemts (when it's possible to view Silverlight on Linux, our company would like to sign up with you) directed at our personal training on the net client http://www.ptonthenet.com, they still only acount for a fraction of a percent of our users, but the fact that fitness companies/finess trainers/professionals are actually using Linux, is, I think, significant, and points to a siginificantly higher user base than NetApplications 1% would suggest. The requests/complaints are significant enough that we will be testing our websites on Linux based browsers. |
| Sander_Marechal May 07, 2009 5:41 AM EDT |
@Tracyanne: Will they also be dropping Silverlight in favour of something else? |
| gus3 May 07, 2009 11:50 AM EDT |
Or just dropping Silverlight on general principles? |
| tracyanne May 07, 2009 10:10 PM EDT |
@sANDER, no, too much invested in it for starters, second it's almost too easy to develop Silverlight applications. The answer to that is not politics. It's a better mousetrap. |
| tuxchick May 07, 2009 10:34 PM EDT |
Quoting: That's Hanlon's Razor. I prefer Stihl's Chainsaw. Crude but effective. |
| Steven_Rosenber May 08, 2009 12:08 AM EDT |
Gillette or Schick? |
| gus3 May 08, 2009 12:45 AM EDT |
Hanlon's Razor is just a psychological twist on Occam's Razor, which reduces to "Make as few assumptions as possible." |
| chalbersma May 08, 2009 8:04 PM EDT |
I don't know. 1% seems like a reasonable estimate. Maybe I'll get flamed but I could definitly see 1% of personal computers running linux. |
| azerthoth May 08, 2009 8:36 PM EDT |
So you see a 3.2% to 1% overall decline in desktop over a 2 year period reasonable? Curious. |
| tracyanne May 09, 2009 12:44 AM EDT |
Quoting:So you see a 3.2% to 1% overall decline in desktop over a 2 year period reasonable? Especially when Linux has even more mind share than 2 years ago (at least the Ubuntu version), and more people than ever are very dissatisfied with Windows, and Net books are available and selling with Linux, in some countries at least. |
| Sander_Marechal May 09, 2009 2:57 AM EDT |
I stumbled across the source of the "9 million programmers" quote. Joel Spolsky said that there were 9 million professional programmers in the English speaking world. He made that statement with regard to the number of visitors on stackoverflow.com. So that's 9 million people in the US, UK, Canada and Australia that get paid to write code full-time. That excludes a lot of countries, excludes non-professional programmers (and FOSS has a ton of those) and also excludes perl and bash hackers who are sysadmins rather than full-time developers. |
Posting in this forum is limited to members of the group: [ForumMods, SITEADMINS, MEMBERS.]
Becoming a member of LXer is easy and free. Join Us!