More dispatches from the smartphone front

Four days ago I blogged about upgrading my old G-1 to CyanogenMOD and complained that the mod’s home site contained no high-level overview of the modding process to help pepople understand all the details in the modding instructions for individual devices.

It took some days of research, but I have now fixed that problem. I had that narrative description reviewed by experts on the IRC #cyanogenmod IRC channel while I was writing it, so it’s an outsider view with insider correctness.

In other news, see Android’s Share Of Mobile Web Use Soaring, iPhone Falling. If that share is tracking installed base accurately, Android installed base will pass iOS’s sometime between the beginning of December 2010 and mid-February 2011. This is just mildly more pessimistic than my forecast months ago of a crossover in 4Q2010, but it also needs bearing in mind that iOS webshare includes iPads as well as iPhones, and iPads don’t yet have effective competition. Thus the usage figures probably underestimate the rate of share decline for iPhones alone.

All indications continue to be that the iPhone is collapsing towards the same status of high-margin niche product with single-digit market share that has been the historic norm for Apple. The Apple fanboys who think that smooth UI and industrial design trump freedom from vendor lock-in and everything else (and champion the walled-garden content model!) have a lot more crow still to eat. I expect to enjoy every second of serving it to them.

The dirty little secret about Google Android suggests that Android’s triumph may not be a good thing after all. In its reading of events, the withdrawal of the Nexus One signals that Google is allowing cellphone carriers to seize back the power over feature lists that it seemed to be claiming with the announcement of the Android 2.2 feature list. And it’s true that some Android devices now have unhelpful carrier customizations that lock out features like USB tethering and lock in crapware apps that users might qet to delete.

Months ago, I thought that Google walking away from a deal with Verizon meant that Google was going to get tough with the carriers about not crippling Android 2.2. I may have been wrong about that, or right at the time but something has changed in their calculations since. They’re not resisting feature deletion and crapware with the intensity I expected then.

While this is cause for concern, it’s not yet time for alarm. I pointed out in Flattening the Smartphone Market, five months ago, that what’s really dangerous to the carriers is comparability of product. When customers start asking questions like “Sprint Android has tethering, why don’t I?” and “Why have you blocked me from deleting apps that you privileged?” vendor attempts to control features are living on borrowed time, whatever Google does or doesn’t do.

The existence of aftermarket upgrades like CyanogenMod for phones produced in huge volumes sharpens the edge of that blade by making unlocked, user-controlled phones a reality. Projects like OpenMoko’s FreeRunner are worthy and wonderful but can’t capture the economies of scale that come from expected production runs in the hundreds of millions; thus they’re doomed to remain expensive niche toys for hackers. CyanogenMOD is more effective because it co-opts the commercial success of Android and the financial mass of the handset manufacturers.

Which brings me back to CyanogenMOD documentation. Anything that lowers the barrier to installing that mod on mass-market phones increases the pressure on the carriers. The amount of grief they can inflict on their customers via Android customization is now bounded above by the hassle cost of installing CyanogenMOD. By making that installation easier, I have directly attacked their lock-in.

Are Google’s strategic planners counting on efforts like CyanogenMOD to help fight their corner? You betcha. They learned how to think about the open-source long term from me (I still have Larry and Sergei’s fanmail from 1998 to prove that), and they’re not stupid. If I can see where product comparability is leading, I’m bound to assume that so can they – and it recommends a very Taoist strategy. That is, rather than fight the carrier oligopoly for openness directly, give them enough rope to hang themselves with – and let their customers do the actual hanging.

41 thoughts on “More dispatches from the smartphone front

  1. I must personally express my gratitude. Very soon I’ll be getting a second-hand Droid and after having already seen the official restrictions from Verizon with regards to USB and wifi tethering on 2.2, I’m really glad to have a simple and understandable method to combat it :-)

  2. I echo Mike Swanson’s comment. I’m probably going to get a Samsung Captivate in the next few months (else I will switch providers entirely, and I don’t know who to flip to if I do: I want good call quality and coverage, obviously, but I also want 3G data and tethering.

  3. >I’ll be getting a second-hand Droid

    There’s a CyanogenMOD port for the CDMA version of the Droid that should cover you.

    >Samsung Captivate

    Ah, that’s the Galaxy S. There’s a CyanogenMOD 6.0 port for that, too.

  4. “it’s an outsider view with insider correctness.”

    This is very important. You wrote it while you were still newb enough to not know the process so well that you took any of it for granted. One of the toughest things for people with specialized knowledge to do is to explain it to those who don’t yet know the specialty. (I do think you’re one of the rare people who can have high level knowledge and still explain it to those who need to learn it.)

    “You are in a twisty maze of man pages, all alike.”

  5. All indications continue to be that the iPhone is collapsing

    What indications would those be? The fact that each iPhone launch has set new sales records?

    The Apple fanboys who think that smooth UI and industrial design trump freedom from vendor lock-in and everything else

    Oh, for heaven’s sake. The number of people willing to put up with an inferior product for political reasons like you do is insignificant. Android’s high-water mark will be the day that the iPhone goes multi-user in the United States.

    Just look at all the work you had to go through to update the OS on your Android device. There are hundreds, maybe a couple thousand people in the whole world who would do likewise. Meanwhile, the tens of millions of iPhone users have a device that “just works”.

    I expect to enjoy every second of serving it to them.

    Your expectations have a habit of diverging considerably from reality, Eric.

  6. >Your expectations have a habit of diverging considerably from reality, Eric.

    Laugh while you can. I predicted the present shape of the smartphone market sooner and more accurately than anyone else, beginning in late 2008. It wasn’t even difficult, because I grokked the underlying economics. And if you haven’t scurried into a hole in four months, I will take great joy in thrusting this quote right back at you.

    Here’s a prediction: if and when the iPhone goes multi-carrier in the U.S., it will be not only too late to break Android’s already-established dominance, it won’t even cause a perceptible blip in relative market share. You think Android’s mass userbase is composed of people who settled for not-an-iPhone because they had to, but that’s a crack fantasy just like the crack fantasy that the iPhone 4 (remember the iPhone 4?) was going to reverse the decline.

  7. Re: the carriers pushing locking patches to their Android installs: Never having used Android, my understanding of the process is limited. However, from my knowledge, the process of rooting an Android phone is analogous to jailbreaking an iPhone, except that the initial restrictions are fewer and the process is less tedious and hackish. Presumably, then, an Android phone once rooted is similar to any other Android phone, regardless the carrier customisations, because the user can always undo those mods (at least in theory). Is this incorrect? If not, then shouldn’t a rooted phone with app-locking or similar restrictions be able to simply bypass them?

  8. Meanwhile, the tens of millions of iPhone users have a device that “just works”.

    Except when they grip it the wrong way.

    As I said back then… Apple lost the ability to say that when Steve Job’s initial response to the dropped calls issue was “you’re doing it wrong”.

  9. Laugh while you can.

    Thanks, I will. I’ll be laughing at your histrionics for years to come, I’m sure.

    You think Android’s mass userbase is composed of people who settled for not-an-iPhone because they had to,

    Those who settled, those who didn’t care, those who were swayed by a skilled salesman, and those who simply didn’t know any better.

    it will be not only too late to break Android’s already-established dominance,

    You really seem to have a hefty amount of your ego invested in this. Why is that? Is it because you’re still pissed off that Linux on the Desktop was a total washout? Are you just jealous that Apple’s enormously successful, while VA Linux turned out to be a flash in the pan?

    I will take great joy in thrusting this quote right back at you.

    I wonder, do you develop the same kind of irrational emotional attachment to, say, sports teams? The Android platform is inferior to Apple’s offerings. If it were better, I’d have bought one.

    I’ve had the thing in my hands, I’ve evaluated its development system, and the sad fact is that it’s a turd. It has all the UI sensibility of the various X window GUI abortions I’ve seen over the years, the development environment is Java, which introduces all the same suckage that Java phones had before the iPhone, and on top of that, it’s a fragmented platform which requires developers to target the least common denominator. The political aspects that you care so much about matter not one whit.

  10. Except when they grip it the wrong way.

    I’ve got an iPhone 4, and its radio performance exceeds all of my previous cell phones. If you’re in a location where the signal is very weak, (where many other phones wouldn’t have connected at all), you can cause the call to drop by bridging the gap between the cell and wifi antennas. You can do the same thing with any other phone at the limit of their ranges, too.

  11. esr> because I grokked the underlying economics

    The telcos carriers have a profit motive to not share revenue with Apple, and perhaps they *think* they can more easily co-opt an open OS and without threat to their network access oligarchy. And more choice always dilutes a more refined competitor, because evolution (economics) demands random diversity. I explained recently in layman’s terms, that although more possibilities seems more complex, a reduction in possibilities is more like O(n*n), because for example one is in the middle of an urgent project and suddenly realizes a dire need for a variation of a tool that is unavailable immediately. However, the iPhone will not be reduced to single-digit percent niche marketshare if it can go multi-carrier and can continue to sport significant exclusive popular usability advantages (and no vendor crapware bloat), but I doubt Jobs can find enough new uncopied popular features to stay ahead of the wolves. Perhaps Jobs’s next
    strategic move will be to push greater ease of integration, but without open protocols that is another form of lockin.

    Although Mr. Raymond’s overview appears to be well written and appreciated by those of us who like to hack, until I see evidence that more than 1% are rooting their smartphone, this is not an effective strategy to break the telco oligarchy. A decade of customer angst
    (crapware, etc) has not and will not break that oligarchy (different flavors of poison) without a paradigm to effectively route that angst around the oligarghy’s monopoly over network access. Math helps. I have been trying to explain (in prior 2 blogs) that the winning strategy is to target the way we write programs, such that programs become unsignable, because they will be *dynamic*, network amalgamations of 10s, 100s, or 1000s of referentially transparent code snippets (signed platforms won’t cope or won’t compete with the (n! – k!) / k! rate of innovation). The Taoist
    understands that the insidious nibbling of the termite is more destructive to structure than than overt sting of the wasp. To allow such
    amalgamation, security theater has to end. It is our benefit to end security threater by making it harder for us to root our own devices without physical access to the device (i.e. push a physical button), because (given we improve the way we write programs) the Apache indians decentralized model of social structure is the most impervious to control. For those who didn’t follow, profileration of code permutations can’t happen with effective principle of least authority in the OS. Calling for us to be more intelligient and not fight against ourselves is not cranky, granted it may be too visionary (far-fetched) for some. I realize most still won’t get the point (big picture). Ambot nimo, baThala ka dong. I appreciate discussing ideas, yet I take no solice whilst my vaporware makes me a moUtherfluster.

    Some Guy> political aspects [...] matter not one whit

    For now agreed, but even if we ignore the political doomsdayers, I don’t think anyone can deny that consumers would go bananas over a device that is devoid of security issues and rocks with unlimited *dynamic*
    exponential rates of software innovation. My thought is kill two birds with one stone. My big problem with an iPhone is I don’t want to write software that dies with the platform one day. Open protocols have a longer half-life. Jobs is brilliantly mercurial– blasts of innovation without holding power because his strategies hinge on control. I admire his passion for refinement and art of the tools of the mind. But on a
    different scale, anarchy is more beautiful and effective. The ying and yang of evolution (blasts of order and decay along the overall trend to maximum uncontrolled possibilities).

    Some Guy> Linux on the Desktop was a total washout

    It didn’t offering any compelling advantage, and offered numerous disadvantages (e.g. not running all my software, hard to install, etc) on the desktop. On servers *nix offers many advantages. Had E2E principle (clients also servers) not been destroyed, perhaps *nix would have mattered more on the desktop.

    But other than fixing the security theater, the future is an OS will not be able to distinquish itself significantly. The real battle is in the application space (assuming all move to a super-secure micro-kernel with full disk encryption, memory protection, and most OS services become applications running in user space).

    Some Guy> hefty amount of your ego

    Someone told me that more constructive discussion occurs in those forums when everyone is anonymous.

  12. >You really seem to have a hefty amount of your ego invested in this. Why is that? Is it because you’re still pissed off that Linux on the Desktop was a total washout?

    Er, total washout? How many times do I have to point out that most market-research terms who track this stuff think Linux still beats Apple OS X in total market share, it’s just that Linux is less visible because the installs are off anyone’s sales radar.

    So if you’re going to call desktop Linux a washout, you have to dismiss OS X on the same grounds. Or have a double standard…oh, wait, you already do.

  13. >However, from my knowledge, the process of rooting an Android phone is analogous to jailbreaking an iPhone, except that the initial restrictions are fewer and the process is less tedious and hackish.

    My knowledge of jailbreaking iPhones is secondhand, but I believe this is correct.

    >Presumably, then, an Android phone once rooted is similar to any other Android phone, regardless the carrier customisations, because the user can always undo those mods (at least in theory).

    Correct.

    >If not, then shouldn’t a rooted phone with app-locking or similar restrictions be able to simply bypass them?

    Yes. To do it you’d need to do exactly what the CyanogenMOD people have done, which is gen up your own firmware image with the carrier restrictions patched out.

  14. You can do the same thing with any other phone at the limit of their ranges, too.

    Sure except that when you do an objective test of phones comparing signal attenuation in dB(not the crappy non-standard signal bars version) this is easier to see as the apple spin that it is. The cliff notes version is that from the ninja turtle grip to holding naturally the iphone 4 loses roughly 19dB compared to about 10dB on the N1. Turn that into the kind of death grips they were showing in their “the other kids do it too” example and the iphone 4 loses 24dB compared to the N1s 17dB.

    Now to put that into the same scale of readouts(the “incorrect” pre iOS 4.01 one). If we’re on the cusp between max(5) bars and 4 bars we’re registering about -91dB. At this point, a 24dB loss takes you to -115dB which is 0 bars. At the same time the N1 would be at -108dB which is just hitting the 1 bar territory. According to Anandtech UMTS starts going south at -107dB and has pretty much found a cave for hibernating at -113dB so the iphone4 is dropped and the N1 is hearing a little bit of static. (Caveat: i haven’t been able to verify those dB values from a separate source).

    This isn’t the greatest rigor, just one anandtech analysis. They also have an article giving further signal attenuation values. So i’m open to seeing objective analyses of signal attenuation that you can find.

    Interestingly the N1 was the second worst phone of the lot, with the iphone 4 in _an open palm_ being worse than all but the N1 when held naturally. In contrast, the Iphone 3GS gets basically no attenuation until you start squeezing. That is actually very impressive.

    But all this is besides the point. Apple’s catch cry(that you’re repeating) is “It just works”.
    Not “It just works”(unless you hold it naturally).
    Not “It just works”(except it has failure cases that are similar to other phones).
    Not “It just works”(once you put a case on it).

    If they had put a layer of insulation on the sides (e.g. made it the standard rounded edge where the roundness is all insulation) we wouldn’t be having this conversation. Ultimately the ideal behind “it just works” is that good design is unobtrusive and that is what Apple has failed at here.

  15. I think ESR’s view is bit to US centric. In the US the the competition isn’t so much between phone models, but between phone/carrier bundles. How much of Android’s success is due to the fact that it is offered on other networks than AT&T? How much of the antenna problems are really just network problems?
    When the iPhone 4 was launched in Switzerland a few colleges of mine got one immediately. We then wanted to see if we could reproduce the antenna problems that were reported from the US. We couldn’t. No matter what we tried, how we hald it, we could not get the phone to drop a connection (and we were doing this in a basement). So maybe the whole “antenna gate” thing has more to do with the quality (or lack thereof) of the AT&T network than with an oversight on the part of Apple’s engineers. So in the US maybe the biggest flaw in the iPhone is AT&T. Over here every iPhone 4 customer I know is very happy. I’ll probably get one for myself too, to replace my iPhone 3G
    In markets where bundling of phone and subscription is uncommon (or even downright outlawed) every phone is available for every carrier. A true “flat” market, and there the iPhone still outsells android smartphones by a nice margin.
    So I have an iPhone. In fact, all the Linux/Unix geeks I know have one. I have looked at the Samsung Galaxy S which looks like an very good phone, but will not buy it. The apps I use don’t exist for this phone, and off the shelf android phones are as much a walled garden as iPhones are. So moving to Android would mean losing a lot of functionality without gaining much.

  16. >A true “flat” market, and there the iPhone still outsells android smartphones by a nice margin.

    I’ve heard this claimed before, but nobody has offered figures with a provenance. Can you?

  17. I googled for information on wortldwide smartphone market share.

    New information: according to ZDNet, a report by Piper Jaffray contains the following interesting tidbit about worldwide market share:

    We estimate Google will control 14.9% of the smartphone market through Android in 2010, growing to 23.2% in 2012. For Apple, we expect the iPhone 15.9% of the smartphone market in 2010, growing to 17.6% in 2012

    Looks like they think Apple and Android market shares are about 1% apart now, which is within statistical noise. This is certainly not K’s “nice margin”. And look at how the 2012 projections diverge; they expect iOS to lose market share relative to Android,

  18. I googled for information on wortldwide smartphone market share.

    I’ve tried to find information on smartphone numbers in Australia but the best I could find is a nebulous and unsupported “the groundswell of interest is there” type comments(mostly fueled by US stats) that I couldn’t in good conscience treat as more than advertising.

    Having said that though, I saw my first Android phone advert(galaxy s) only a couple of months ago so any numbers Android has here are almost surely not representative. (Although in the group of people I know there are many Android phones, including at least one other Milestone, which isn’t bad since none of them would have been on a plan)

  19. > Er, total washout? How many times do I have to point out that most market-research terms who track this stuff think Linux still beats Apple OS X in total market share, it’s just that Linux is less visible because the installs are off anyone’s sales radar.

    What are your “figures with a provenance” for this? All the web user statistic of any kind of representative sampling put Linux desktops at 1 – 2 % in the US and Europe and Mac OS X somewhere upwards of 5 %. Of course, web user agent counts are a problematic measure in many ways, but I don’t see anything much better for estimating the installed base of desktop Linux. You can find statistics with high percentages of Linux user agents, but these all seem to have samplings that are strongly biased toward web or other developers. Then there’s that slide in Steve Ballmer’s presentation some time ago that had something like 7 % for Linux or whatever it was. I’d like to believe it, but I never saw any basis for the claim, or anyone else saying such stuff.

    Whatever the size of the installed base, “washout” is nonsense. Linux is used by millions of people on the desktop every day and the system is being actively developed and improved all the time.

  20. >What are your “figures with a provenance” for this?

    Actually, I think the strongest evidence is the Ballmer 7% report that you cite. It’s what lawyers call an “admission against interest”; Ballmer was talking to investors, not the DOJ, so he had every incentive to minimize Linux market share rather than maximize it.

  21. >However, from my knowledge, the process of rooting an Android phone is analogous to jailbreaking an iPhone, except that the initial restrictions are fewer and the process is less tedious and hackish.

    I have never tried this as I am not an iPhone user, but http://www.jailbreakme.com claims to make jailbreaking an iPhone as simple as navigating the native browser to their website. The site is notorious for enabling pranksters to jailbreak the iPhones on display at Apple stores.

  22. Does that mean Microsoft may believe the market share is larger than 7%? Wow that’s pretty interesting.

    Hell even if 7% were a totally non-smudged number, that’s a fairly significant portion of users. How many computers are out there? Facebook has 500 million users which would put approximately 35 million of them on Linux. 7% is not small =p

    Around this time I really wish Google would generate and publish some usage statistics — they’re probably the closest thing that really does have nearly every web user on the planet going to.

  23. Actually, I think the strongest evidence is the Ballmer 7% report that you cite. It’s what lawyers call an “admission against interest”; Ballmer was talking to investors, not the DOJ, so he had every incentive to minimize Linux market share rather than maximize it.

    I just haven’t seen anyone else make that claim. You said

    I have to point out that most market-research terms who track this stuff think Linux still beats Apple OS X in total market share

    Can you point us to such research? Ballmer undoubtedly has at his disposal resources that the rest of us don’t, but if the market share really was 7 %, I think someone else should have noticed, too. And Ballmer has been known to be full of shit at times.

    @Mike Swanson

    35 million is actually not far from what the various Linux distributors estimate, but that’s not 7 % of desktop systems or web users, even if it might be 7 % of Facebook users. Google used to publish web user agent statistics and Linux was somewhere slightly above 1 % when they stopped publishing. I assume they produce the statistics internally still, or at least are able to easily do it if they want to. I never quite understood why they stopped publishing. Do they sell that stuff now to these analysts?

  24. >Can you point us to such research?

    Hm. Maybe not, any more. Now that I look, I see that other than the Ballmer cite all my sources are several years old and seem to have dropped off the web. I hadn’t noticed this before because I considered the 7% figure from Microsoft strong enough evidence to make most of the previous figures uninteresting.

    I’ll have to do some deeper digging. I did a little Googling last night, but all I found were estimates based on browser IDs and I don’t trust those much.

  25. >Does that mean Microsoft may believe the market share is larger than 7%? Wow that’s pretty interesting.

    Yes, it is. But tempting though it might be, let’s not jack this thread to discuss that; as someone pointed out previously, “washout” would be falsified at share well below 7%.

  26. > Er, total washout? How many times do I have to point out that most market-research terms who track this stuff think Linux still beats Apple OS X in total market share, it’s just that Linux is less visible because the installs are off anyone’s sales radar.

    If you talk about desktop units, then what you could do is gather round the linux distributors, convince them of adding a simple ‘Linux’ tag to (let’s say) the browser and do it consistently, or perhaps ask permission to count the install as a new desktop linux, (like ubuntu does when installing, if you want), so that these figures can be better calculated. Still, I very much doubt your claim.

    If you are thinking ‘servers and small phones are linux too’, sorry, but that helps Desktop Linux exactly 0 for now, and will continue not helping until those actually resemble a desktop in use. Counting a graphic-less server as ‘one linux install that competes with OSX’ is terribly dishonest. So is counting a smartphone that has Linux underneath but a completely different, perhaps proprietary, cover. It doesn’t even help for device driver recognition, because those environments are either already solved or completely locked down and purpose-built.

  27. >Counting a graphic-less server as ‘one linux install that competes with OSX’ is terribly dishonest.

    Of course it is. I’m not advocating that, nor is anyone else.

  28. Ok, there aren’t that many important desktop Linux distros: Suse, Fedora, Ubuntu, Debian, Mandriva are the biggest I know for that segment. Coordinating them in this effort is that difficult? It is, after all, in their own interest.

  29. gather round the linux distributors, convince them of adding a simple ‘Linux’ tag to (let’s say) the browser and do it consistently,

    This is what they have been doing already, anyway, at least with the common browsers on the common distros. Firefox user agent string on CentOS looks like this:

    “Mozilla/5.0 (X11; U; Linux x86_64; en-US; rv:1.9.2.7) Gecko/20100726 CentOS/3.6-3.el5.centos Firefox/3.6.7″

    It says ‘Linux’ and the name of the distro in it. All of the large distros do this, and even if ‘Linux’ is missing, it’s trivial to add the distros together. People sometimes say that the statistics are skewed by users spoofing Windows browsers because some sites refuse Linux clients, but I highly doubt that this is common enough to have a significant effect. I don’t know of any Linux user who’d routinely do this or of a major distro doing it by default.

  30. What I’m trying to point out: the people who stand to gain the most from an accurate, big figure of linux desktops are both distros and users. That they haven’t said anything, tells me more than Eric’s beliefs on the matter. If desktop linux had really got a 7% provable penetration on the desktop, flowers would bloom, birds would chirp merrily on my window, and i’d be able to install more games without having to dual boot.

  31. >What I’m trying to point out: the people who stand to gain the most from an accurate, big figure of linux desktops are both distros and users

    Sadly, desktop share doesn’t matter to any of the big distros except Ubuntu. The other Linux players have all decided that high-margin sales to corporate customers are more fun than fighting the Fraunhofer Institute over MP3 codecs and the MPEG consortium over video. Screwy as it seems, Microsoft is probably the only player with both the motive and the money to do a careful study; they need it for threat assessment.

  32. What I’m trying to point out: the people who stand to gain the most from an accurate, big figure of linux desktops are both distros and users. That they haven’t said anything

    Of course. They have said something. This is Fedora’s take on it:

    http://fedoraproject.org/wiki/Statistics

    That’s missing a lot of downloads from mirrors and probably a lot of sites like my place of work, where we have two dozen Fedora desktops updating from a local mirror. They never get counted on these server statistics. The total number of unique IPs is accumulated over a long period. Unfortunately they don’t publish e.g. the number of unique IPs in the last year regardless of distro version. Maybe the number would be unflattering. I’m pretty sure that it’s a lot less than the 24 million total they have on the page now. Guessing from the individual numbers of the more or less current distros, I’d say something like four million.

    Fedora also has an opt-in registration of the hardware at installation time (Smolt), but those numbers are a lot lower. Apparently not many people are interested in being counted voluntarily. Of course, adding a default, automatic ‘ping home regularly’ feature would be trivial, but the users would likely go berserk over something like that.

    Ubuntum is claimed to be the most common distro. If the web user agent data is to be believed, and if at least the proportions within a given sample are significant, Ubuntu is supposed to be about half of the Linux desktop, which would leave Fedora’s however many millions it is within the remaining half.

  33. >This is Fedora’s take on it:

    Yes, of which the most important sentence is: “Currently, there is no reliable way to determine the total number of Linux users, or even count the total number of users of any Linux distribution which does not have a mandatory per user registration process.”

  34. Mike H the 2nd got me thinking.
    Like his, my User Agent string:
    Mozilla/5.0 (X11; U; Linux i686; en-US; rv:1.9.2.3) Gecko/20100423 Ubuntu/10.04 (lucid) Firefox/3.6.3
    omits the gnu/ in front of Linux.
    What if 7% report their OS as Linux, but only 1% report their OS as gnu/Linux?

  35. WordPress ate my ‘ducking and covering’, probably because they were enclosed in angle brackets. I did not intend to troll, but to only to provide some light humor.

  36. What if 7% report their OS as Linux, but only 1% report their OS as gnu/Linux?

    Then obviously RMS needs to grow his beard longer.

  37. Speak of the devil… This article on the O’Reilly site puts Linux market share at 8 %, approximately on par with Mac OS X, based apparently mostly on netbook sales. The writer also refers to the presentation by Ballmer, but no basis is given for his numbers.

    http://broadcast.oreilly.com/2010/09/debunking-the-1-myth.html

    I still wonder why the web user data is the way it is. If there are such problems with sampling, shouldn’t the share of Mac OS X vary as wildly as that of Linux between the different surveys? Maybe it does, but from what I’ve seen, it’s pretty consistently a lot more than the share of Linux. Of course, the user agent data measures installed base and not market share, but if Linux systems have always been sold and installed so much more than we thought, shouldn’t the installed base be on par with Mac, too?

  38. Thanks for the wiki page man! :) I’m surprised to see that the only thing I was ignorant about was the “Radio” section.

  39. Eric, here’s someone independently arriving at some of your own conclusions (with a dose of self-interest, since Jeff Atwood is comparing his software with android in some way). I don’t know if you’re familiar with StackOverflow, but I find it a wonderful site (and no, I’m not associated with them in any way beyond ‘user’).

  40. >Eric, here’s someone independently arriving at some of your own conclusions

    He uses one of my favorite historical references – John Boyd on agility vs. speed in fighter combat. Make me wonder if it’s completely independent or if I influenced his thinking. In any case, I agree with the analysis.

Leave a Reply

Your email address will not be published. Required fields are marked *

You may use these HTML tags and attributes: <a href="" title=""> <abbr title=""> <acronym title=""> <b> <blockquote cite=""> <cite> <code> <del datetime=""> <em> <i> <q cite=""> <strike> <strong> <pre lang="" line="" escaped="" highlight="">