Don’t panic over Honeycomb’s release delay

Google’s announced plans to delay the public release of the source code for Honeycomb, the tablet version of Android, are causing some indignation to be vented among the partisans of open source. But should it? This is a good time to reflect on what the freedoms guaranteed by the Open Source Definition actually mean and what they’re for.

Google says it wants to hold Honeycomb close to its chest for a while because it doesn’t want random hardware vendors trying to jam it onto cellphones before it’s ready. Release is being restricted to a handful of tablet vendors Google is partnered with.

Considering the disappointing crappiness and ridiculously high prices of the Android tablets that have shipped so far, it’s hard to blame Google for wanting to exert some quality control. There are difficult tradeoffs here. On the one hand, it serves nobody if handset vendors jump into shipping phones that are as flaky and nasty as the tablets we’ve seen so far; on the other hand, restricting source access also forfeits the code auditing and early field testing Google would otherwise get from early release. Also, Google pisses off the unfavored vendors and at least a substantial minority of the open-source community.

But the controversy isn’t about whether the move is a good idea, it’s about whether Google is violating the norms of the open-source community and (in the most hyperventilating accusations) has turned evil and deceptive.

I don’t think it has. This sort of release delay is an unusual thing to do, but it has happened before without a lot of screaming and finger-pointing. The Ghostscript and mySQL projects used to make a regular policy of time-delayed open-source releases lagging a leading-edge source tree they shipped only to paying customers; mySQL may still do this, I haven’t been keeping close enough track of them to know.

We didn’t see pitchforks and torches at the castle gates of the companies developing mySQL and Ghostscript, because the Open Source Definition doesn’t forbid behavior like this. Nor does the web of customs surrounding the GNU general Public License. Neither these nor any other community norms actually require any development group to release code it thinks is half-baked. They don’t even forbid selective close-to-the-kimono releases – in fact, even the hardest-core zealots at the Free Software Foundation have never fussed about that and they’re conspicuously not doing so now.

What the OSD and other community norms are designed to guarantee is that when there is a public release, you have a right to redistribute it, modify it, and reuse portions in your own code. Google has not attempted to infringe on this right and there is no sign that it intends to try.

One reason I’m relaxed about Google’s plans is that I think I understand Google’s grand strategy. Actually attempting to renege on the rights guaranteed by Android’s OSD-conformant licenses would be suicidally disruptive of that strategy – the change in other players’ cost and business-risk calculations would blow the Android coalition apart.

Thus, it makes sense for the Honeycomb delay to be exactly the interim measure that Google says it is. I have frankly been a bit puzzled about why they’re catching as much flak as they are, which is why I’m a bit late responding to the controversy. Everybody calm down, please! This is no big deal. There’s precedent, and Google’s self-interest will require it to release Honeycomb source publicly when it’s fully baked.

165 comments

  1. @Breeze, he’s most likely referring to mySQL.

    ESR: That’s correct. Which’ll teach me to post in the dead of night…

  2. Could you elabourate on the definition of “public release”? Does releasing devices running it count as a “public release”?

    1. >Could you elabourate on the definition of “public release”? Does releasing devices running it count as a “public release”?

      Probably. Of course, then we get to have an argument over what constitutes the release of a device. At what point in the progress from prototype through trade-show demo to field testing to early street distribution does that trigger?

      I do not think we are well served by attempting to erect hairsplitting legalistic distinctions and then ginning up controversies about them. Remember Santayana’s definition of a fanatic: “He who redoubles his efforts after he has forgotten his aim. ” And don’t be that guy.

  3. This is off-topic here, but the old post I was reading no longer had a reply box. If you haven’t read Thomas Sowell’s Intellectuals and Society yet, you should. He makes many of the same points you did in A Brief History of Firearms Policy Fraud, but applied to nearly all aspects of liberal intellectuals’ beliefs and claims, and documents them. The book is pretty close to being an enlarged and updated version of A Conflict of Visions and, especially, The Vision of the Anointed.

    ESR says: I have re-enabled comments on that post. Not sure how they got disabled.

  4. Why the flak?

    Because conflict sells and harmony doesn’t. In addition, there are those who win if Android loses.

    So you see “angrry” articles and “the end is near” articles. Articles that yell “nothing happened”, like the above will be ignored.

  5. I have frankly been a bit puzzled about why they’re catching as much flak as they are, which is why I’m a bit late responding to the controversy. Everybody calm down, please! This is no big deal. There’s precedent, and Google’s self-interest will require it to release Honeycomb source publicly when it’s fully baked.

    Those screaming the loudest will be the ones that view access to the source code as an entitlement, rather than as a privilege granted out of self-interest.

  6. @Morgan Greywolf
    “Those screaming the loudest will be the ones that view access to the source code as an entitlement, rather than as a privilege granted out of self-interest.”

    I suspect that those that scream the loudest are those that want Android and Google in general to fail. As Eric wrote, the FSF is conspicuous for its absence in the criticism. It was a Businessweek article that started this brouhaha. And Businessweek is not exactly a hard-core Free Software/GPL publication.
    http://www.businessweek.com/technology/content/mar2011/tc20110324_269784.htm

    (we discussed this earlier at NOD and WINCE)

  7. Disclaimer: Most of this comment is copy pasta from my comment in the older thread.

    It is possible that part of the reason you don’t hear much crowing about mySQL and why you do about Google could have to do with that to the best of my knowledge, mySQL doesn’t have executives mocking their competitors with such comments as:

    “the definition of open: “mkdir android ; cd android ; repo init -u git://android.git.kernel.org/platform/manifest.git ; repo sync ; make””

    I’m sure at least a significant chunk of the “controversy” is people finding Google’s actions amusing in light of such above comments. Especially since Xoom tablets with honeycomb are shipping, which I think by anyone’s definition would count as a “public release” without any hair splitting.

    But none of this is really surprising. In fact, I would be more surprised if Google didn’t slowly move to a more “closed” process like this. It appears the most truly successful uses of open source are a sort of Cathedral with a Bazaar on the front lawn model, we can see this in Red Hat, Mozilla, MySQL, Apple and many others. Compare to a completely open Bazaar model like say wikipedia, where anyone really can contribute [and then note that Wikipedia has been slowly closing their doors over time]. When your product becomes more than just a product, when it becomes a brand, there are certain benefits to a cathedral. Closing your doors, even just a little, gives you better control over how your brand is perceived. As Android and Google become more powerful brands, Google is going to want to exercise more control of what “Android” actually is.

    Despite the differences, it is apt to look at Android and the lessons to be learned from windows. Microsoft made it so that in the consumer’s minds, Windows was the PC. Although there were always manufacturer brands (dell, HP, gateway), if you asked someone what type of computer they had, you had a better chance of getting the windows version than the name of the manufacturer. Now look at what sort of headaches that managed to bring Microsoft. How many problems do you think were attributed to Windows, when an honest evaluation would have determined it was the hardware?

    So it can be with Android, and that appears to be what Google is afraid of here. They don’t want people putting Honeycomb on phones because they don’t know that it will work well, and they know if it doesn’t work, that Android will take the blame for it, not Samsung or HTC.

    I think you’re going to see a lot more of this. In fact, it wouldn’t surprise me if in a year or two, you see Android, and Android Community Edition or some other distinction between Android (Cathedral) and Android (Bazaar).

  8. Google has a TM on the name “Android”, right? So, why can’t they just forbid vendors to ship devices labeled as “Android” until Google considers the code baked?

    That would make it possible to open up the sources early for anyone to see, nullify all claims that Android isn’t open source, and ascertain the quality of the shipped devices.

    No?

  9. @tmoney
    “It appears the most truly successful uses of open source are a sort of Cathedral with a Bazaar on the front lawn model, we can see this in Red Hat, Mozilla, MySQL, Apple and many others. ”

    It depends on how you define “successful”.

    In my list, the most successful Open Source projects are: GCC, Linux, Apache, Mozilla, KDE, Gnome, in no paticular order. They behave a lot different from MySQL, KDE, or Apple. I bet that most readrs here can point out even more successful Open Source projects I missed that also do not behave like that.

    The difference between these groups of projects is that at one hand we have projects where it is attempted to extract money directly from the code, and in the others no such attempts are made. It seems that the most profitable way to get money out of Open source is brand management combined with selling Services.

  10. Get ready for the Hissyfits asserting this post proves, PROVES, I tell ya. Eric is evil, eeevull I say. And he can’t hack, and he’s funny looking, and he rewrites history, and he’s stoooopid, blah blah blah. IN 5…IN 4…IN 3…

  11. Android is very much a Cathedral-model development and always has been; there isn’t much code written by non-Google employees in Android, aside (of course) from the projects (Linux kernel, for example) that are pulled in whole.

    I have no doubt that Google will continue to release the source code for Android – if for no other reason than there’s a lot of GPL code in there that is copyright to other people and they’d be breaching the GPL terms by not releasing source.

    One of the original comparisons in “Cathedral and the Bazaar” was the Linux kernel vs HURD. HURD was very definitely Cathedral, while Linux was the archetype for the Bazaar. Part of the argument for the Bazaar being more successful than the Cathedral is that the Bazaar allows for larger teams. While that’s true, large software companies can put together an in-house team of comparable size to the Bazaar-mode teams like Linux and Firefox, and Android’s structure means that the design arguments that Firefox, in particular, spends a lot of time resolving are simply irrelevant; Google says how it will work and that is that.

    You can see something similar with Chrome/Chromium, though that’s with a public source-tree.

  12. I suspect that Honeycomb right now isn’t really usable, or at least doesn’t have anything over Gingerbread for single-core processors.

    Instead of releasing it and getting a hundred forks – and yes, they will be forks, incompatible fixes around problems, that will all break when they do fix it in general.

    The Xoom likely has many hacks tuned just for it that won’t work well elsewhere. And will be true of the other vendor tablets.

    But when released, the full version should be backward compatible and have good code across all platforms.

    Also, Google can’t force vendors to update. So if there are many crappy ports and forks, and it becomes hard to update to the fixed version, you will get a bunch of bad tablets.

    It is probably (untuned) an early beta. Releasing it would cause more problems than it would solve. I’m annoyed and wish Google would be a little more detailed but can understand.

    They haven’t said they won’t ever release it, merely delay it for quality. Unlike the linux kernel which Linus will delay as needed, there isn’t a history of RC versions everyone knows are buggy for Android.

  13. So how long, exactly, can you wait to provide source before you’re in violation of the GPL? A month? Until the code is “finished”? A year? Ten years?

  14. @Richard:

    You post assumes that Google isn’t using Bazaar-mode development internally. Having actually interviewed with Google, I would say that I think that Google probably does run some projects this way.

  15. @Richard:

    I have no doubt that Google will continue to release the source code for Android – if for no other reason than there’s a lot of GPL code in there that is copyright to other people and they’d be breaching the GPL terms by not releasing source.

    @Michael Hipp:

    So how long, exactly, can you wait to provide source before you’re in violation of the GPL? A month? Until the code is “finished”? A year? Ten years?

    You guys appear to be assuming that google isn’t releasing the GPLed code that has shipped in Xoom. Care to back that assumption up with any cites?

  16. One thing that no one seems to have mentioned is that near as I can tell, the gpl only ‘requires’ that the source be offered to someone you give the binary to. So, if anyone who actually got a tablet with honeycomb also has a method of obtaining the source, google has it covered. They are not *obligated* to make the source available to anyone else.

    So, if you really want the source, get a tablet with honeycomb on it and you’re in business. As Patrick noted, I would bet a fairly large nickle that you can get it without hassle.

  17. @Patrick Maupin:
    You guys appear to be assuming that google isn’t releasing the GPLed code that has shipped in Xoom. Care to back that assumption up with any cites?

    Uh, how about Eric’s blog post.

    If we’re not talking about releasing GPL code (I said nothing about Xoom, neither did Eric) then what *are* we talking about?

  18. > Uh, how about Eric’s blog post.

    There are three separate conflated issues here:

    1) Google only has to ship source to those they ship binaries to. Probably they don’t even ship binaries, so they aren’t even required to ship source…

    2) For the released devices (e.g. Xoom), it is up to the vendor (e.g. Motorola) to release GPLed sources.

    3) Google has done a fine job of separating licenses. Motorola can release all the GPLed kernel source, and doesn’t have to release anything above the kernel.

  19. > I said nothing about Xoom, neither did Eric

    (Forgot to hit this in the last post)

    OK, but where else would you see Honeycomb binaries in action? Or in other words, what would be the event that prompts release of the source?

    Look, a zillion companies all over the world use and modify open source internally, and some of them work directly with public repositories, but a lot of work is done internally, and then made public when it’s ready to ship. (A company that doesn’t actually ship isn’t required to make anything public, but a lot of them do to accrue more of the benefits of open source.)

    In other words, Eric’s post was about how google is most likely going to keep going down the open source path for business reasons. Your post strongly implied that google might be in violation of the GPL. These are two completely separate issues, and for you to imply that, simply because Eric didn’t address the Xoom, your question doesn’t require you to address the Xoom is completely specious reasoning.

  20. Speculating a bit here, there might be yet another explanation. Appearently there are rumors on the big interweb, that Google is working on releasing it’s own tablet with Honeycomb, as early as this summer. Now, if that is the case, it’s not entirely inconcievable that the release of the source for HC would coincide with the release of this device. It would simply follow the model of Nexus-One/Froyo and Nexus-S/Gingerbread.

    On a related note, I’ve got to say that I’m also slightly baffled by how seemingly “Google bashing” has become quite the fashion in the last couple of years, even kosher in FOSS circles. It might be the automatic knee-jerk reaction that “they are big therefor they must be evil”, but still… I mean how can FOSS folks ignore that the company Google is a massively successful Open Source story? And how can they ignore the staggering mass of FOSS software they release on an almost daily basis? Is it simply because they forget easily, or is it envy, or… what? I’ve gotta say, as software behemoths go, I would put my money and trust with el Goog any day, rather than any of the competition.

  21. @Morgan

    > You post assumes that Google isn’t using Bazaar-mode development internally. Having actually interviewed with Google, I would say that I think that Google probably does run some projects this way.

    One of things I’ve always understood about the cathedral/bazaar metaphor is that the bazaar is open-air, available for all to view and participate in. Put a stone roof and windows over it and it ceases to be a bazaar….it becomes a shopping mall, with a carefully controlled shopping experience and overbearing management, with only the outward appearance of a open-air market. It gets even worse when they paint blue skies and clouds (Android-based-on-Linux) on the ceiling.

  22. @Jim Hurlburt
    “One thing that no one seems to have mentioned is that near as I can tell, the gpl only ‘requires’ that the source be offered to someone you give the binary to. ”

    That has been a common interpretation before the Internet. However, the current interpretation is that the GPL requires anyone who distributes a binary to offer source to everyone who has a legal copy of the binary. In practice, this means anybody who has a binary. So the common sense solution is to just give everybody who bothers to ask a copy of the source.

    The GPL kicks in the moment you hand anybody outside of your “organization” a binary without source. It is up to you to convince a judge that the person you gave a binary is legally part of your organization.

    If Google gives Motorola source plus binary, they are absolved from any requirements that follow Motorola’s distribution of binaries. And Motorola only has to offer source for GPLed binaries. Which is expressively not anything related to Dalvik, the part people outside of Google are interested in.

  23. > If Google gives Motorola source plus binary,

    Google and Motorola are probably smart enough that they mainly exchange source code, at least for the Linux kernel. Motorola needs to be able to build the entire phone stack to control their own destiny, and if Google never ships binaries, there can never be any GPL issues.

  24. @Matt Minton:

    One of things I’ve always understood about the cathedral/bazaar metaphor is that the bazaar is open-air, available for all to view and participate in.

    The bazaar has many different vendors’ stalls in it. Over time, the most popular vendors are going to be the ones with the best quality products. Unfortunately, when network effects kick in and products become so popular that communication becomes a burden, it turns out that filters are required to keep the SNR to a level where development can still actually occur. This means that, while most bazaar stalls keep their wares and most of their internal communication public at all times, the myth that anybody can participate in any particular bazaar project is just that — a myth. Attention of the project leaders and main participants is a valuable finite resource, and must be managed carefully.

    Android has been open-sourced for around 2 1/2 years, and development is highly concentrated within a single company and, to a lesser extent, a few select partners. Unlike Linux, which has been in public development for two decades, Android doesn’t have the carefully cultivated but organically grown infrastructure to easily insulate the lead developers from the masses. One of the tools they do have to manage the influx of communication they are not yet ready to deal with is to manage the outflows of communication that will produce that influx. It’s a crude tool, but to some extent, it’s all they have so far.

    Another issue facing google is PR. We’ve all seen the whining about how this person or that person was mistreated on LKML, or about how desktop linux will never happen because linux devs hate average users.

    Well, guess what? Google adores average users and deeply cares about the experience of the average user on Android, and all the attendant PR issues. Paradoxically, this means they have to be more constrained in their communication, because they can’t afford the meme that Android isn’t for Joe Sixpack.

    This doesn’t mean that they have to wall off internal development like Microsoft, merely that it would be extremely surprising, and probably counterproductive, if their coping mechanisms for reducing the demands on the lead developers were the same as the mechanisms used for Linux kernel development. I wouldn’t be surprised if they get more open over time, but they’re still struggling with needing to move extremely quickly against a very capable opponent, and while open source is important to them, it won’t do them any good to have always fully opened the source for something that nobody is using, or that has a needlessly damaged reputation.

  25. Eric,

    I agree with your point, but I want to mention that the two examples you gave (MySQL and Ghostscript) both require copyright assignment to their respective corporate entities for outside contributions. Thus the entire code base was copyrighted to a single source, and they were thus able to offer *dual* licenses, a proprietary closed source license and the GPL. What Ghostscript did (completely legal) is simply only offer the proprietary license for the latest and greatest versions, and dual license older versions.

    Is this a similar situation with Android? It is my understanding that they incorporate GPLed software that has third party copyrights… most notably a modified version of the Linux kernel. This may be the reason people are upset about Android but not Ghostscript and MySQL.

  26. .The Google Android development team have the release of the source code for version 2.2 of their open source Android mobile operating system code named Froyo . Originally announced at this years developer conference Froyo is the seventh platform release since 1.0 and the next major update for Googles mobile OS.

  27. >If Google gives Motorola source plus binary, they are absolved from any requirements that follow
    >Motorola’s distribution of binaries.

    Boy, there’s a business opportunity…

    1) Start a small company, incorporated in some country with weak / nonexistent copyright law.
    2) Offer to act as a “distributor” for any closed source shop wanting to distribute modified GPL software without having to release their source.
    3) Obtain binary code and source. Lock source in vault, distribute binaries.
    4) Ignore C&D letters from FSF
    5) Profit

  28. Addendum:

    Kidding aside, it does open an interesting question about at what point your responsibilities are absolved. If you change GPL software, market those changes and product, and partner with someone else to distribute those changes in their product using your trademarks, are you still responsible for providing source code, or is the fact that you use a distributor, enough to absolve you?

  29. @tmoney:

    How are you going to profit if you don’t get anybody to pay you?

    Who is going to pay you? Why would they? How long would it be before potential customers saw your name associated with “scam” on the internet, and decided not to send you any money?

    If you’re distributing pure GPLed code, people can get it from elsewhere. If you’re attempting to distribute value-added code, why would any publisher trust you with their own code?

    Kidding aside, it does open an interesting question about at what point your responsibilities are absolved. If you change GPL software, market those changes and product, and partner with someone else to distribute those changes in their product using your trademarks, are you still responsible for providing source code, or is the fact that you use a distributor, enough to absolve you?

    Courts will look at agreements to see if they are third-party, arm’s length. If not, they can treat two entities as one. If they are valid agreements between different entities, then why wouldn’t the GPL work as written? Are you worried Motorola doesn’t have deep enough pockets to pay for damages if they neglect to distribute the Linux Kernel for the Xoom?

  30. @JPL:

    Is this a similar situation with Android? It is my understanding that they incorporate GPLed software that has third party copyrights… most notably a modified version of the Linux kernel. This may be the reason people are upset about Android but not Ghostscript and MySQL.

    If you have a Xoom, and Motorola won’t give you the sources to the GPLed kernel, then by all means complain. If you can even find such a person, let us know. Otherwise, it’s all FUD.

  31. There’s nothing illegal or unethical about saying to someone “Here’s a binary version of GPLed source code, some from me, some from others. I’ll give it to you if you explicitly agree not to ask me for source code.” If you accept this deal, you have bargained away your right to source code, presumably for the sake of something else you value more.

    This model isn’t suitable for distribution to mass-market customers, who don’t want to sign contracts before they buy things (and you don’t want the effort of signing them and enforcing them either). For dealing with a few partners? Sure. No problem.

  32. I am wondering to what extent this release delay is specifically aimed at the Chinese market. Earlier Android releases have been quickly forked with industrial volumes, which I’m sure is not making Google happy. Delaying source release would give their official in-country partners some support. (Similar things happen in other countries, but China is the elephant in that room.)

    > Thus, it makes sense for the Honeycomb delay to be exactly the interim measure that Google says it is.

    I have no trouble believing this. When Apple first shipped the iPad, it was with an interim version of iOS (3.2) that was just for the iPad, got leap-frogged by 4.0 and 4.1 on the iPhone *only*, until 4.2 finally unified the two forks. This kind of thing happens when you’re coding as fast as you can, and you want to ship product this year instead of next.

    > I have frankly been a bit puzzled about why they’re catching as much flak as they are, which is why I’m a bit late responding to the controversy. Everybody calm down, please! This is no big deal. There’s precedent, and Google’s self-interest will require it to release Honeycomb source publicly when it’s fully baked.

    There’s some push-back against Google’s “we’re open and they’re closed” marketing campaign; it sets them up for the obvious “oh, yeah?” chorus whenever they do something that looks even vaguely “non-open”. Think of it as a cost of the PR strategy they’ve chosen. I don’t think the (GPL) legalities matter much here.

    Cheers
    — perry

  33. There’s nothing illegal or unethical about saying to someone “Here’s a binary version of GPLed source code, some from me, some from others. I’ll give it to you if you explicitly agree not to ask me for source code.”

    Yes, there is. There’s a third party you’re not accounting for, and they are the ones with the actual rights: The original authors. If you do that, you’ve lost your right to distribute the code by violating the terms of the GPL, so now you are violating the copyright of the authors if the only rights you have on the code are those granted to you by complying with the GPL.

    If that move worked, the GPL would be useless, because you could just steal anything and make that part of your new license (transitively).

  34. This is a mess. Google have shot themselves in the foot a bit, but I don’t think they’re in violation of the GPL.

    Anyone that has a Honeycomb device can ask their manufacturer for the GPL code, and the manufacturer is required to supply it.

    On the other hand, I don’t believe Android is open.

    >the definition of open: “mkdir android ; cd android ; repo init -u git://android.git.kernel.org/platform/manifest.git ; repo sync ; make”

    Which is true iff the code the internal Android devs are working on is in that repo, ie. that repo is bleeding edge.

  35. @LyleD:

    > On the other hand, I don’t believe Android is open.

    HoneyComb is not open (yet). But you can’t seriously argue that previous versions of Android aren’t. To address this point, I have copied, with a few minor edits, a comment I made on a previous post.

    This isn’t the first time that google has delayed the release of Android source. They have done this periodically from the start, when they don’t want people coding to internal interfaces that will be changing, for example. There is no question that, from the perspective of the outside world, it is more a cathedral model than a bazaar model. Of course, one of the benefits of a cathedral model is that you can polish your stuff a bit more before you show it off, but one of the downsides is that you’re expected to.

    Google certainly inhabits this in between world, driven by, among other things, the hardware schedule demands of its closest partners, and its own image/branding/reliability concerns. But some of the uninformed, unthinking vitriol at slashdot, for example, would be amusing if it didn’t play straight into Apple’s and Microsoft’s hands. I have seen, in the same slashdot comment, allegations that google should be doing development in the open (bazaar-style) and that they shouldn’t have released the (allegedly broken) HoneyComb software yet!

    I don’t claim to know all google’s reasons for the delay. It may simply be, as google suggests, that they want to clean things up before the “Ice Cream” release. Or it may well be, as some at slashdot allege, that one goal is to let some of the bigger manufacturers (who monetarily helped development) get a headstart on shipment of HoneyComb hardware.

    Obviously google has their own internal business reasons to do whatever it is they do, but this certainly isn’t any sort of ethical issue. In any case, I will be extremely surprised if google’s business issues drive any kind of significant change in their Android business model, and I’m actually kind of amazed at the number of people who (obviously on purpose, in some cases) misunderstand how google interacts with open source, and treat this as proof of some underlying honesty issues and/or the beginning of the end.

    Google has handed the FOSS world the dominant cellphone software stack on a silver platter. This is huge, and nothing google does in the future can detract from this. FSF-fetishists[1] don’t like it because it’s not properly viral, and point to that fact as evidence of google’s evil behavior, conveniently ignoring that Aladdin, TrollTech, MySQL, and countless others have extremely similar business models, where they develop in a cathedral and push software out to the world without taking contributions. The difference, of course, is that those companies charge rent to those of their customers who want to use the software in proprietary ways, while google derives their profit in a more circuitous fashion. But to an FSF-fetishist, although proprietary software licenses are evil, it is permissible, even encouraged, to consort with those enemies for the express purpose of rent-seeking to support more GPLed programs, but completely ignoble to give your efforts to the world with no strings attached.

    What the FSF-fetishists fail to note, however, is that once google open-sourced Android, they put the same sort of competitive pressure on themselves as any of those for-profit GPL/proprietary companies. If they stop innovating with Android, others will pick up the slack. Of course, if at some point google wanted to take a breather, they could hand off a lot of the maintenance work to others — a tactic that would be sure death for those GPL/proprietary companies.

    But I can actually envision a world where a cellphone or tablet manufacturer contracts with the cyanogenmod developers to develop and maintain Android drivers for their device, and everybody[2] is happy — google because Android is shipping on another device, the manufacturer because their software development and support costs are even lower, and cyanogen mod developers, because they’re getting paid to do what they want to. cyanogenmod could be the next RedHat.

    [1] Please note that my term “FSF-fetishists” does not include RMS or, to my knowledge, anyone actually at the FSF. Instead, it includes unthinking knee-jerk reactionaries, and the closed-source puppetmasters that like to stir things up when it suits them, both of which can be readily found all over the net, weighing in on the lack of HoneyComb source code availability.

    [2] In this context, “everybody” obviously doesn’t include Apple, RIM, or those walking dead men Microsoft and Nokia.

  36. @ patrick

    >Who is going to pay you? Why would they? How long would it be before potential customers saw your name associated
    >with “scam” on the internet, and decided not to send you any money?
    >If you’re distributing pure GPLed code, people can get it from elsewhere. If you’re attempting to distribute
    >value-added code, why would any publisher trust you with their own code?

    The customers would be the companies wanting to use / ship modified GPL code without having to give away their source, though admittedly, the model only works for guys who don’t mind outsourcing their hardware. Basically the point was that it’s not just so simple as saying “we gave the source to our distributor, it’s their problem now” with this sort of thing. If it were that simple, companies would be doing end runs around the GPL just by dealing through a distributor as long as the cost of dealing with the distributor was less than the cost of licensing the GPL software under different terms (assuming such alternative licensing was even possible).

    Instead of Linksys routers, you might see “Linksys Routers by XXX Distributor”. Same hardware, same software, but because XXX Distributor is the actual distributor shipping the device and code, Linksys hasn’t violated the GPL (at least according to the interpretation that says if you give the source to your distributor, your distributor is responsible for further GPL compliance).

    >Are you worried Motorola doesn’t have deep enough pockets to pay for damages if they neglect to distribute the Linux Kernel for the Xoom?

    I’m not worried about anything. To be honest, if all it takes to do an end run around the GPL is laundering your code the same way the mob launders money, more power to the corporations. I find the GPL almost as onerous as some proprietary licenses and anything that discourages its use (or breaks it) is fine by me.

  37. @Patrick Maupin

    I guess I should have worded what I said more precisely, and for that I apologise. Android is not completely open, in that we can’t see what the devs are doing until the whole bundle is pushed into the publicly viewable git repo. I don’t have a problem with this development model and I understand there are legitimate reasons for doing this, but I also understand that some hackers aren’t happy unless they have the bleeding edge running on their devices.

    From my point of view, Google is completely in the clear. As far as the FSF-fetishists go, I must quote Commander Sinclair from Babylon 5. “When you become a zealot, you become the enemy” or something to that effect.

  38. @tmoney:

    Instead of Linksys routers, you might see “Linksys Routers by XXX Distributor”. Same hardware, same software, but because XXX Distributor is the actual distributor shipping the device and code, Linksys hasn’t violated the GPL (at least according to the interpretation that says if you give the source to your distributor, your distributor is responsible for further GPL compliance).

    I almost addressed the hardware play in my previous post, but then I didn’t bother because I thought the answer was too obvious. Once Linksys or their agent ships a binary, in theory anybody with that binary who sells it to somebody else is committing copyright infringement. See, for example, Omega v CostCo.

    So, whether it’s Amazon, WalMart, or Best Buy, the hardware play gives plenty of juicy targets to the FSF. The foreign company pure software play might make it easier to work around the laws, but the questions I asked in my original post are still relevant — where does the money come from? If there is no value-add, why would anybody give them any money? If there is enough value-add to attract revenue, then there are real programmers involved who theoretically shouldn’t need to be violating the GPL. Also, to the extent that the customers were aware of the culpability of the vendor, the FSF probably wouldn’t have any qualms about going after them for copyright violation, using the Costco ruling. I’m really having a hard time seeing a viable business model here.

    To be honest, if all it takes to do an end run around the GPL is laundering your code the same way the mob launders money, more power to the corporations. I find the GPL almost as onerous as some proprietary licenses and anything that discourages its use (or breaks it) is fine by me.

    To the extent that some programmers require the reciprocity agreement in the GPL to produce useful stuff, the GPL is a tool that increases available open source stuff, both for use and for study, and that’s good. I have several issues with the hard-core pushers of the GPL, some of whom, spread FUD that the license is actually more viral than it really is, and some of whom demonize all proprietary programs as unethical or even “evil”, etc.

    But I don’t have any real problems with people who use the license to release useful stuff, although I do find it incredibly annoying on those occasions when I come across a 25 line script that has the GPL slapped on it. That’s really the tail trying to wag the dog.

  39. @Perry The Cynic:

    There’s some push-back against Google’s “we’re open and they’re closed” marketing campaign; it sets them up for the obvious “oh, yeah?” chorus whenever they do something that looks even vaguely “non-open”. Think of it as a cost of the PR strategy they’ve chosen. I don’t think the (GPL) legalities matter much here.

    I think the actual cost involved for google isn’t actually all that big. Most of the developers who matter code on both Android and iPhone. The rest probably know the score. I’d be interested to see if anybody can point to an actual developer of any significance who claims this is an issue. And as far as the users go — Joe Blow doesn’t care about open in the sense of source code, but he might care about whether he can get his porn in the app store. (And no, he doesn’t spell it “pr0n”.)

    Even non-porn apps like VLC are going to be a huge differentiator to some users. It’s actually nice to have the GPL on some useful things…

    The only really disappointed people are the modders of the world who have to wait a few months to be able to mod their tablets, and the Microsofts and Apples of the world who don’t realize why their arguments about this supposed hypocrisy on the part of Google aren’t getting that much traction.

    BTW, the charges that Android isn’t truly open source are over two years old.

  40. @tmoney
    “we gave the source to our distributor, it’s their problem now”

    Indeed, the one distributing the binaries is the one who has the obligation to deliver the sources. That is reasonable. If the distributer has the source, how can I be sure the binary is the one he got from upstream? And, if we go upstream, why would, say, Linus Torvalds be responsible for delivering the build system for some no-name router hardware?

    It happens a lot, though. Some OEM buys a cheap operating system in the far east to put on their hardware, say a TV set. Then it is found out that the set runs linux/busybox. The resellers and OEM do not have the source so cannot comply even if they wanted. As a result, all the equipment is confiscated.
    (it really happened http://arstechnica.com/open-source/news/2010/08/court-rules-gpl-part-of-a-well-pleaded-case.ars )

  41. @Patrick Maupin
    “BTW, the charges that Android isn’t truly open source are over two years old.”

    The article has some amusing quotes with 20/20 hindsight:

    ‘Unsurprisingly, in Williams’ view an open platform requires a community behind it.

    “If you were to ask me [the Symbian Foundation] roadmap for the next two years I’m going to tell you I don’t know – I can tell you… what the plans are but other than that it’s up to the community to take it where it wants to go,” he explained. ‘

  42. I think the Register is a prime example of a news-outlet that lives by conflict. And they prove worthy of that qualification. The case is made that you can ignore the opinion of the coders and the author of the license. But you can indeed trust lawyers who have no clue about coding.

    Google’s ‘clean’ Linux headers: Are they really that dirty?
    When lawyers and Linus collide
    http://www.theregister.co.uk/2011/03/29/google_android_and_the_linux_headers/

    When we spoke to Raymond Nimmer – where this whole thing began – he reiterated that he believes “there is a risk” with Google’s use of the header files. “If I was developing an Android app of any significance, I would tell the CEO, ‘we have a judgment to make’,” he told us. But then he added: “Of course, I’m not a coder. Someone else would have to look at the code and make that judgment.”

  43. >Once Linksys or their agent ships a binary, in theory anybody with that binary who sells it to somebody
    >else is committing copyright infringement. See, for example, Omega v CostCo.
    >So, whether it’s Amazon, WalMart, or Best Buy, the hardware play gives plenty of juicy targets to the FSF.

    Thats another interesting can of worms. As a retailer, how liable are you for ensuring the manufacturers comply with licensing rights to the software in the products they ship? Is wal-mart liable for license violations in Windows? Does best buy need to have a code auditor on the payroll to ensure that every DVD player Samsung makes doesn’t contain GPL code? My guess would be that at best, the FSF could get an injunction to stop the sale of these devices while a case is on going, but I doubt they could get much in the way of damages out of the companies. I could be wrong of course.

    It goes the other way too. If we say that anyone who sells a product containing GPL code (whether they modified the code themselves or not) is responsible for supplying the source, does that mean if I sell my Xoom second hand that I’ve become responsible for providing the GPL source code as well? And does that mean that Motorola has no obligation to provide the source to the second hand buyer?

    Is it any wonder businesses try to avoid the GPL like a plague?

  44. >It happens a lot, though. Some OEM buys a cheap operating system in the far east to put on their hardware,
    >say a TV set. Then it is found out that the set runs linux/busybox. The resellers and OEM do not have the source
    >so cannot comply even if they wanted. As a result, all the equipment is confiscated.

    To be fair, the judgment there was summary judgment over failure to comply with discovery and the court proceedings and doesn’t really tell us if resellers have an obligation under the GPL. It just tells us that a GPL copyright claim is sound enough not to be thrown out on the face of it. This is perhaps the most frustrating things about the GPL (and I think someone mentioned it here before, maybe esr?), almost all the cases that have tested and tried the GPL have been relatively weak rulings, often concluding without any real test of the GPL itself

  45. @tmoney
    “As a retailer, how liable are you for ensuring the manufacturers comply with licensing rights to the software in the products they ship? ”

    How could they?

    The GPL is a license to perform actions under the umbrella of copyright law. A reseller or retailer does not copy or change anything. So how can they be bound by copyright law?

    The same holds for resellers in general. If they do not make new copies, there is nothing copyright law can do them. The one and only exception is when the goods themselves are illegal copies.

    @tmoney:
    “Is it any wonder businesses try to avoid the GPL like a plague?”

    My response above is also the reason businesses are very fond of GPLed software. All the cheap (wireless) routers and TV sets run Linux. I even get a GPL license page on my (not so cheap) TV set.

    The problem only exists where ignorant businesses buy no-name binaries without “asking” for the relevant copyright papers. Because if you buy software, you must ask for a written permission to copy and distribute. And not asking for these papers is “odd”.

  46. @tmoney
    “This is perhaps the most frustrating things about the GPL (and I think someone mentioned it here before, maybe esr?), almost all the cases that have tested and tried the GPL have been relatively weak rulings, often concluding without any real test of the GPL itself”

    In Germany and France the courts were very determined to uphold every letter of the GPL.

    GPL passes acid test in German court
    http://www.linux.com/archive/feed/57353

    Thanks to the developer’s research and the lawyer’s understanding, on September 6th, 2006, the court in its judgment, backed the claims made by gpl-violations.org and acknowledging the violation of the GPL by D-Link ordered it to reimburse gpl-violations.org for legal expenses plus the costs of purchasing and re-engineering the product.

    Big Win for GNU GPL in France
    http://opendotdotdot.blogspot.com/2009/09/big-win-for-gnu-gpl-in-france.html

    In a landmark ruling that will set legal precedent, the Paris Court of Appeals decided last week that the company Edu4 violated the terms of the GNU General Public License (GPL) when it distributed binary copies of the remote desktop access software VNC but denied users access to its corresponding source code. The suit was filed by Association pour la formation professionnelle des adultes (AFPA), a French education organization.

    Enforcement of the GNU GPLin Germany and Europe
    http://www.scribd.com/doc/47584756/enforcement-of-the-gpl-in-Germany-and-Europe

    Abstract:
    GPL enforcement is successful in Europe. In several court decisions and out of courtsettlements the license conditions of the GPL have been successfully enforced. In particular, embeddedsystems are the main focus of such compliance activities. The article describes the practice ofenforcement activities and the legal prerequisites under the application of German law

  47. @Winter:

    The GPL is a license to perform actions under the umbrella of copyright law. A reseller or retailer does not copy or change anything. So how can they be bound by copyright law?

    That would have been my assumption a year ago, but given that you’re assuming it now, I take it you didn’t actually read the Costco decision? :-)

    @tmoney:

    Thats another interesting can of worms. As a retailer, how liable are you for ensuring the manufacturers comply with licensing rights to the software in the products they ship? Is wal-mart liable for license violations in Windows?

    After the Costco decision, I would say yes.

    Is it any wonder businesses try to avoid the GPL like a plague?

    There are two separate issues here. The first one, which Costco covers, has to do with a copy that was not legally created. If a hardware vendor is violating the GPL (or Windows or other license), then his copies are not legal. As you point out, the original manufacturer might be out of the country, so it is only fair that the retailer can be sued, but also only fair that he can join his upstream suppliers into the suit.

    The other issue is how much control an author has over a legally created copy. If you follow some of the reasoning in Apple v. Psystar, it could get really ugly. Of course, that case wasn’t really adjudicated properly, so some of the reasoning might not stand, but according to Apple, I might be in violation of the license on their precious IP if I bought a Mac and then gave it to my daughter.

    But Apple theoretically (at least according to 9th district and others which uphold shrinkwrap) licenses software rather than sells it. This is like Costco (where the copy was not legal) in that the consumer doesn’t actually own the copy, which eviscerates first-sale. With GPL v2 I was pretty sure that I would own a legally created copy. With v3 I’m not so sure, but it still looks like that might be the case. If I own a copy, then I don’t have to worry about the license — I can give the copy away. If I don’t own the copy, then you’re right, there could be issues.

    Of course, many of the GPL proponents (e.g. over at groklaw) hope and pray that first-sale doesn’t apply to GPL, because if it does, then somebody might, in theory. be able to get around the license with something similar to some of the wacky schemes you’ve been proposing. Of course, the can of worms you point out, e.g. “can I sell my used router 10 years later without going to jail?”, argues heavily in favor of first-sale applying to a legally made copy of GPLed software, but to my knowledge, this issue hasn’t made it to court.

    In any case, your question about the GPL is properly aimed at any non-permissive license. Microsoft, Adobe, and Apple all reserve the right to take your firstborn…

  48. @Patrick Maupin
    “Of course, many of the GPL proponents (e.g. over at groklaw) hope and pray that first-sale doesn’t apply to GPL, because if it does, then somebody might, in theory. be able to get around the license with something similar to some of the wacky schemes you’ve been proposing. ”

    I have never ever seen a GPL proponent that wants to abolish the first sale doctrine. And I have been reading Groklaw from the early beginnings. But that does not mean you have not met them. They might very well exist.

    If you do not alter or copy a program, the GPL is rather simple: It does not object to doing whatever you want with a program, including giving it away. The one thing the original GPLv2 said was that you must hand over the “offer for source” with the copy you sell or give away. As the “offer for source” has been dead since the advent of the Internet, that is irrelevant.

    The source must be offered by the creator of the binary copy. If they refuse to make the source available, then it is he who has to appear in court. The GPL is very clear about the fact that the user does never lose her right to use the code.

    If you can point out any action taken in the name of the GPL to sue someone who simply resold a legal copy, please inform us.

  49. @Winter:

    I have never ever seen a GPL proponent that wants to abolish the first sale doctrine. And I have been reading Groklaw from the early beginnings.

    Well, if PJ believed in comment searching I think I could easily show you a few dozen interesting posts related to GPL and first sale. In any case, I can easily show you that PJ herself thinks that the first sale doctrine is “a topic dear to the heart of anti-GPL trolls.”.

    If you can point out any action taken in the name of the GPL to sue someone who simply resold a legal copy, please inform us.

    I know of no such action. That’s precisely why I wrote “but to my knowledge, this issue hasn’t made it to court.”

    When we spoke to Raymond Nimmer – where this whole thing began – he reiterated that he believes “there is a risk” with Google’s use of the header files.

    Raymond Nimmer is a putz who realized 4 decades ago that the serendipity of being a lawyer with the same last name as the preeminent worldwide expert on copyright law was a meal ticket that was too powerful to ignore. Without his last name, he would be nothing; with it, he is dean of a law school.

  50. Hey Eric, now that the honeycomb tablets are on their way can we get an update on your position on tablet computers? The last thing you wrote was:
    “Now I’m not sure there’s enough reality there to sustain the product category at all.”

    1. >Hey Eric, now that the honeycomb tablets are on their way can we get an update on your position on tablet computers?

      The iPad has been selling well enough to make it clear that there’s some reality there. But it’s still a crazy market with a lot of hype and unrealistic expectations floating around in it. There are two weirdnesses that particularly stick out to me:

      1. What the hell is up with the pricing on Android tablets? It’s ridiculous. They have to come in $100 below iPad to have a prayer of selling in volume. Why are half the electronics companies in the world are building trade-show demo gadgets with a MSRP that gives them no hope of ever making it to a mass market?

      2. What the hell-squared is up with tying these things to cell-phone plans? I already have a device for making phone calls; it’s called (duh) a phone. The only way I’ll buy a tablet is if I can pay for it once to have a cheap Web terminal I can leave on my kitchen table.

  51. @Patrick

    …Another issue facing google is PR….
    …Google adores average users and deeply cares about the experience of the average user on Android, and all the attendant PR issues…
    …Google has handed the FOSS world the dominant cellphone software stack on a silver platter. This is huge, and nothing google does in the future can detract from this…

    It seems we have two visions of Google. I won’t deny that Google has done the open source community a HUGE favor with Android, but I also have to look at Google’s actions outside of Android, and they’re not good. Perhaps this is my libertarian lack of trust in large, singular organizations, but Google as an organization seems enamored with control. There is plenty Google can do to stain itself that has nothing to do as a software development house and everything to do as a dominant player in two of the greatest communication tools ever created, the Internet and mobile phones. There is a certain ‘creepy’ factor I can’t get past, which is why I thought my shopping mall analog was appropriate. It’s all well-constructed to be pleasing to eye and quickly part you and your money (or privacy), but what is behind those unmarked doors? I simply don’t know. We fail to heed Orwell’s warnings at our peril.

  52. @Matt:

    Google makes its money by reading peoples’ minds. The better they can read minds, the more money they make. There is no question that mind-reading abilities are dangerous, and there is no question that google is deliberately cultivating their mind-reading skills.

    Nonetheless, in my book, they haven’t (yet) gotten anywhere near Apple or Microsoft or Sony on the evil scale, and are certainly working hard in some quarters to put some goodwill in the bank for that inevitable day when one of their employees screws up royally.

  53. The Android tablet-makers mostly seem to be trying to come in above Apple on hardware features; given Apple’s current volume advantage that is difficult and means they’re at best equal in price.

    Most interesting exception now is the Nook e-Book reader; it’s $249, and will be receiving an update in April that will basically make it a proper Android tablet. (officially, although plenty of people have installed stock Android already). I may have to have a look at that… it wouldn’t be surprising to see something similar from Amazon, although their current e-paper based reader isn’t really suitable for normal computing interfaces.

  54. @esr:

    > What the hell is up with the pricing on Android tablets?

    > What the hell-squared is up with tying these things to cell-phone plans?

    I think the cellphone manufacturers wrongly assumed they could somewhat escape the cellphone pricing bloodbath by moving up to tablets. In their fantasy world, there is a huge pent-up demand for the ability to access a cell network with a non-Apple device. Once they have been disabused of this notion, expect prices to plummet.

    In related news, the $249 Nook Color is gaining its own app store with titles like Angry Birds. According to some reasonably reputable sources, B&N has already taken shipment of 3 million Nook colors, or approximately 50% of the North American non-Apple tablet market…

    B&N has apparently realized that they are in a position of strength here. Instead of clinging stubbornly to the idea that it’s an eReader and suing hackers, they seem to be treating it more like the general purpose computing platform it is.

  55. Expensive Android tablets?

    I think it is simple, they cannot deliver. Honeycomb is not ready, the hardware components are not available in bulk, production lines unavailable, etc.

    So if you have a slick product but supply is deadly constrained, you hike up the price enough to keep demand on a level that you can do custom builds.

  56. @Winter: “Expensive Android tablets?

    I think it is simple, they cannot deliver. Honeycomb is not ready, the hardware components are not available in bulk, production lines unavailable, etc.

    So if you have a slick product but supply is deadly constrained, you hike up the price enough to keep demand on a level that you can do custom builds.”

    If we’re just talking about Tegra 2 tablets (particularly with 10″ screens) I might agree. But it completely ignores the case of the Nook Color, which seems to be becoming something of a major sleeper success story.

  57. Of course, 4 months ago, the stories were about Galaxy instead of Xoom:

    http://www.informationweek.com/news/smb/mobile/showArticle.jhtml?articleID=228300095

    The point is, if they’re ramping production down, that doesn’t indicate a supply constraint.

    No matter how you slice it, in a new product category like tablets, where Apple is kicking butt at the high end and Barnes & Noble (a bookstore, for deity’s sake!) is kicking butt at the low end, if you’re ramping down average production a month or two after product roll-out, you’re probably doing something wrong. Mind you, these companies are probably royally screwed no matter what they do. Apple gets enough grief when they drop the price by $100 6 months after product rollout. If these companies man up and do the right thing and cut their prices in half, the anguished cries of the less savvy early adopters will be ringing in their ears for the next quarter.

  58. I think the cellphone manufacturers wrongly assumed they could somewhat escape the cellphone pricing bloodbath by moving up to tablets. In their fantasy world, there is a huge pent-up demand for the ability to access a cell network with a non-Apple device. Once they have been disabused of this notion, expect prices to plummet.

    But even that doesn’t fully explain why they have to be tied to cell-phone plans. There’s no reason they couldn’t ship the devices with a de-activated mobile broadband radio.

  59. > But even that doesn’t fully explain why they have to be tied to cell-phone plans.

    It does when you add even more greed to the equation. If Samsung and Motorola thought it was enough of a seller’s market that they could demand kickbacks from the operators, that pretty much explains it all.

    Remember, there was plenty of evidence last summer that Apple was supply-constrained. Even if you’re merely opportunistic, you operate differently in a supply-constrained market, but when you’re a crudely drawn caricature of a businessman, and you have supply and your competitor apparently doesn’t, so you assume that his would-be customers are now yours to own, it’s easy to get greedy and assume that you can turn the poor punters upside down and shake them until their pockets are completely empty.

    That sometimes works in the market for necessities. Unfortunately for Motorola and Samsung, although consumers are certainly warming to the idea of a tablet, they still view it as a luxury item. Apple’s execution in this market has been stellar to date. They actually needed to price it low enough to create enough demand so that shortages would create enough buzz to work in their favor. Their marketing guys have probably read all the psychological research on the slots at Vegas.

    Viewed in that light, the temporary embargo on Honeycomb source code may turn out to be a really healthy thing for Android tablets, by damping the pricing oscillation that inevitably happens after a lot of capacity comes online based on inflated demand projections.

  60. esr wrote:
    The iPad has been selling well enough to make it clear that there’s some reality there.

    Tablets basically replace low end laptops, netbooks, and e-readers. That’s their niche.

    I got my iPad through work, as I had no intention of buying one, but now that I have it, I have found it indispensable. It made me realize that I never wanted a laptop in the first place. That is, it made me realize that my computing habits are desktop + tablet.

    If I need to create content, I use my desktop. If I’m in the mood to mainly consume content, I use my tablet. And outside of work, I find that I’m a consumer far more than I’m a creator.

    Even on vacation, I really only consume content, so I only bring my iPad and a keyboard these days, and the keyboard is mainly for email. If I actually have to work, like during a business trip, I’ll bring the laptop, but I do not miss lugging around the laptop during vacations.

    I have to say, I’m amazed that only Apple got it right when it comes to cellular plans for tablets. The iPad has a simple on demand usage plan. You basically buy cellular data service for 30 days at a time, no contract, no automatic renewals, no early termination fee, no activation fee. It is the way all tablets should handle cellular data.

  61. Also, Google pisses off the unfavored vendors and at least a substantial minority of the open-source community.

    What, all twelve of them?

    (The “substantial minority”, not the “open-source community”.)

    I kid, but only a little – mostly in that Google can piss off that “substantial minority” more or less endlessly to no effect on their bottom line.

    Because, well, where are those few Open Source Zealots (who are mad that they don’t have the source right now, that is) going to go? iOS?

    Mike: Problem with the Nook theory is that it’s not a very good Android tablet; looking at the specs, there’s a reason the people rooting them are saying it’s for the Android enthusiast market.

    For Joe Schmoe Who Wants A Tablet After Playing With His Buddy’s iPad, well… an Android Nook is exactly what’ll tell him to save the extra $200 and change and buy an iPad.

    Low-spec tablets suck, at least given the current definition of low-spec – presumably that’ll change in the future, the way “low-spec hardware sucks” always changes.

  62. “What the hell-squared is up with tying these things to cell-phone plans? I already have a device for making phone calls; it’s called (duh) a phone.”

    It’s got something going for it. Outside of this blog, there are still some people that don’t want to know from wifi and computers. With one of these, you could take a picture with your cellphone camera, dial up your friend’s phone-tablet, and say, “Hey! Look at this!”. No need to get out a magnifying glass, or boot up a computer.

  63. Am I being too hypercritical if I think that Google is being disingenuous with its open claim?

    For everything said in the entire thread, I still think it’s odd that Android as a project doesn’t actually give access to the code as it is made; only after a “release” is it later opened.

    Is it an open source project if you wait until some indeterminate time after release to actually free the source? And the majority of the world never sees the work in progress? Is merely putting last release’s code under an OSL enough to qualify as an open project?

    (I’m not denying of course that Android makes releases open source. But to call the project itself open seems…)

    That seems radically different from the way other open source projects (Mozilla, KDE, GNOME, kernel) operate.

    But do I need to cut Google some slack, or I am seriously too enamored with Apple to see it in perspective? I’m genuinely curious here. Or am I being too pedantic in my use of “open” and need to allow more wiggle room for a nebulous, non-semantically binding term?

  64. Or, my personal grasp of the idea of open source (warning: ignorance imminent) was that if you will share your code, then it should, for the most part, be available during the entire development cycle. I don’t mean that you can’t have private branches, but there should be a public tree that is developed in full view, and that new releases are based off of the public code – not a sudden dump from a private branch.

    That is, the public code would be open source from development to release. Anyone can get read access and see the current state of the project.

    Which is to say, the idea that “for every possessor of the binary who requests, may receive a copy of the source” would be necessary but not sufficient to meet my idea of an open source project.

    Am I being too severe, or do I have a point? (Which I suppose would be, putting releases under an open source license is not enough to actually meet the requirements of my supposed open source principles, which may be too severe or unrealistic.)

    1. >Or, my personal grasp of the idea of open source (warning: ignorance imminent) was that if you will share your code, then it should, for the most part, be available during the entire development cycle.

      That’s a good idea, but it’s not the law.

      That is, it’s the way to collect maximum possible benefits from the many-eyeballs effect, but there’s no rule about it. How could there be? Who would enforce it? Many projects (including my own) start out with a week or three of private development before you declare 0.1 and throw it in a public repo. This is normal.

  65. Many projects (including my own) start out with a week or three of private development before you declare 0.1 and throw it in a public repo. This is normal.

    I certainly think that’s acceptable. I just think there’s a big difference between “private before first release/private work on my branch” versus “for every single version N+1, the entire source tree is private and unavailable until some indeterminate time after release.” There’s not even a concession to the idea of any development in the open.

    I admit I’m dancing in a giant gray zone not actually enumerated under law or license, but in terms of conduct I think there’s a big distinction between the white-gray and black-gray. I just think it matters. Legally all of this is irrelevant, but Google and the entire FOSS community talk of the ideals as, well…ideals, not laws.

    It just seems inconsistent on Google’s part to claim the ideal, but only meet the bare legal requirement.

  66. Or, my personal grasp of the idea of open source (warning: ignorance imminent) was that if you will share your code, then it should, for the most part, be available during the entire development cycle.

    There exists plenty of precedent where this is or was not the case: Ghostscript in the Aladdin days (that changed after ESP/Apple took over), OpenSolaris since Oracle took over, VirtualBox (even before Oracle took over), etc.

    I’m not so sure about MySQL. The SVN repositories are public, but it may be like VirtualBox where patches are applied to a private repository first before being released publically

  67. There exists plenty of precedent where this is or was not the case: Ghostscript in the Aladdin days (that changed after ESP/Apple took over), OpenSolaris since Oracle took over, VirtualBox (even before Oracle took over), etc.

    Right, but in keeping with my argument, I don’t think it’s unfair to say that Oracle (neé Sun’s) projects were surrounded by a hint of “open in name but not in practice.” Hence OpenOffice’s governance disputes and LibreOffice, or am I not correct?

    There’s definitely precedence for gray-area projects that dance the line of actually using open development, if that’s what you mean.

    And if they’re clearly dancing the line, what’s the point of ideals? Who would call Oracle’s projects a good example of open source?

  68. Or, if Google has to dance the line to create a mass-market successful product with open source ideals riding on the coattails, that’s actually pretty cool – I’m all for that, and I think making multiple source releases of a complete mobile operating system stack is pretty darned good anyway.

    But does the dance required to make it a success still grant them the right to use the claimed ideals as their flag standard? Or if they’re really playing it gray as I think they are, why don’t we just call it that and stop letting them be corporate PR for “openness”?

  69. @twilightomni:

    > I still think it’s odd that Android as a project doesn’t actually give access to the code as it is made; only after a “release” is it later opened.

    That’s a fairly normal commercial way of doing releases. Even for open source. Google is a commercial entity.

    > Is it an open source project if you wait until some indeterminate time after release to actually free the source?

    While a lot of benefits can accrue to doing development in the open, there are good reasons not to in some cases. From the google FAQ on this (emphasis added):

    Why are parts of Android developed in private?

    It typically takes over a year to bring a device to market, but of course device manufacturers want to ship the latest software they can. Developers, meanwhile, don’t want to have to constantly track new versions of the platform when writing apps. Both groups experience a tension between shipping products, and not wanting to fall behind.

    To address this, some parts of the next version of Android including the core platform APIs are developed in a private branch. These APIs constitute the next version of Android. Our aim is to focus attention on the current stable version of the Android source code, while we create the next version of the platform as driven by flagship Android devices. This allows developers and OEMs to focus on a single version without having to track unfinished future work just to keep up. Other parts of the Android system that aren’t related to application compatibility are developed in the open, however. It’s our intention to move more of these parts to open development over time.

    > That seems radically different from the way other open source projects (Mozilla, KDE, GNOME, kernel) operate.

    All those projects are operated by not-for-profit entities. Except, of course, that KDE is based heavily on for-profit QT.

    QT itself didn’t have an open development process until less than two years ago, so by your own definition, a huge portion of KDE wasn’t open from 1996 to 2008.

    > But do I need to cut Google some slack, or I am seriously too enamored with Apple to see it in perspective?

    Google doesn’t need your slack. Your unwarranted criticism will not slow down the juggernaut, hurt their feelings, or sway any important opinions.

    > Or, my personal grasp of the idea of open source (warning: ignorance imminent) was that if you will share your code, then it should, for the most part, be available during the entire development cycle.

    OK, so you don’t even need/like/use google, and you’re criticizing their gift to the world because they like to tidy it up and put a bow on it before they hand it out?

    > there should be a public tree that is developed in full view,

    Why? What’s it to you? You don’t mind, in fact prefer, Apple iOS, where you don’t get to see the source code at all. Seriously, why?

    > Am I being too hypercritical if I think that Google is being disingenuous with its open claim?

    No, I think when you read what you want into what somebody else says and then hold them to your own arbitrary standard, the word isn’t spelled quite like that.

    > Am I being too severe, or do I have a point? (Which I suppose would be, putting releases under an open source license is not enough to actually meet the requirements of my supposed open source principles, which may be too severe or unrealistic.)

    If projects develop according to your stated open source principles (e.g. with an open development model), certain benefits definitely accrue. Google uses that model for some of Android, but for other parts, they have, for business reasons, eschewed this model in favor of one that lets a small, dedicated team move more quickly without worrying about breaking code for lots of too-eager users.

    I can understand criticism of a non-open-development model, as compared to an open development model. I cannot understand criticism of a non-open-development model as compared to iOS, nor can I understand conflating the development model with the openness of the resultant code. The released Android open source code is truly open and free — do what you want with it, and if you don’t like google’s development model and think they’re dragging their feet so much they are hurting the project, feel free to start your own fork.

  70. Alright. My ideal of open isn’t Google’s model of open and I shouldn’t hold them accountable to it.

    Caveat: Your quote explicitly says the point of the Android release model is to allow developers and OEMs to focus on the open ‘stable release’, while Google focuses on the next version for a flagship device. In reality I think we know that most Open Handset Alliance (I think that’s the name) OEMs all get in on the “next flagship version”, while it’s only really the AOSP community that has to play nice with the “safe release” until the next one is “ready”.

    But this last one:

    Nor can I understand conflating the development model with the openness of the resultant code. The released Android open source code is truly open and free — do what you want with it.

    A central point of my contentions can be debunked or confirme here.

    Is open source a set of development principles or not? I assumed it is. I assume the definition of open source project involves the development model itself.

    I assume the very point of open source is not just to publish code, but to show stages of its progress in public where anyone can see what am I working on, to some degree, as I work on it. (Not *always* after I’m done with it.)

    If release (and the license) is the only thing that matters, then yes, you’re right and I have no reason to complain and no deception I should hold against Google.

    1. >Is open source a set of development principles or not? I assumed it is. I assume the definition of open source project involves the development model itself.

      It definitely does involve the development model itself. I wrote the foundational papers about the development model, not licensing technicalities

      That said, it is definitely not the case that one size fits all here. The community has settled on licensing as a minimum constraint because trying to legislate about development procedure (including what would satisfy some legalistic definition of “openness”) is a snakepit nobody wants to wander into.

  71. Right, but in keeping with my argument, I don’t think it’s unfair to say that Oracle (neé Sun’s) projects were surrounded by a hint of “open in name but not in practice.” Hence OpenOffice’s governance disputes and LibreOffice, or am I not correct?

    Sun, MySQL, QT, etc. make all their money off selling non-GPL versions of the GPL code. That’s why the projects aren’t open, even though the code is. Theyonly accept patches with a code license assignment so that they can continue to ship proprietary versions, while disallowing everybody else the same privilege. Google accepts patches. Without requiring a code license assignment. And since Android (other than the kernel) isn’t GPL, everybody else has exactly the same privileges as google.

    There’s definitely precedence for gray-area projects that dance the line of actually using open development, if that’s what you mean.

    Yes, open source can be a great advertising medium for selling your closed source version. But for that to work, you have to maintain ownership of the source.

    And if they’re clearly dancing the line, what’s the point of ideals? Who would call Oracle’s projects a good example of open source?

    But google isn’t dancing this line. Sure, google has commercial reasons to want Android to be successful, but those reasons are, for the most part, aligned with the average consumer. Google feels that a successful open platform will mean more ad sales for them, but a closed platform could, e.g. redirect hits to bing or others.

    So, to the extent google doesn’t develop in the open, I feel it really is because they are moving so fast they don’t want the extra communication cost associated with the process. The fine line they are walking is about when the code is baked well enough to where opening it benefits them more than it costs them, not about keeping others from doing whatever they want with the code. Google lets you do anything you want with the code, including (like the Chinese) modifying it to use your own app store instead of google’s.

    1. >So, to the extent google doesn’t develop in the open, I feel it really is because they are moving so fast they don’t want the extra communication cost associated with the process. The fine line they are walking is about when the code is baked well enough to where opening it benefits them more than it costs them, not about keeping others from doing whatever they want with the code.

      I agree with this analysis. I think they may be overestimating the friction costs of a project run along more open lines, but that’s a different argument from how they may ethically respond to an evaluation that those costs are too high.

  72. Minor odds and ends.

    Why? What’s it to you? You don’t mind, in fact prefer, Apple iOS, where you don’t get to see the source code at all. Seriously, why?

    Because I hold open source projects to a different standard than closed source projects, because by definition they *aspire* to a different set of behaviors and philosophies.

    By your own definition, a huge portion of KDE wasn’t open from 1996 to 2008.

    You are correct. By this definition, KDE (1998-200x pre GPL-Qt) would technically fail my definition of an open source project. So, for that matter, would Netscape’s first release of the Mozilla code, as by my definition a post-facto source release is not enough to claim as an open source project.

    …I’m ok with that from a philosophical standpoint, but I must grant that it’s incredibly strict, and hence the point of analyzing my own fallacies. After all, isn’t philosophical strictness the reason for the creation of GNOME in the first place?

  73. Is open source a set of development principles or not? I assumed it is. I assume the definition of open source project involves the development model itself.

    As I said, I think benefits accrue with an open development model, but so do costs. The openness of the development model is actually completely orthogonal to the openness of the code.

    You can have (and I have seen) code which is developed in the open which has a license on it that says “for academic use only.” That is useless to me. But source code I am allowed to use where the latest revision does exactly what I need? That’s golden. It’s nice if the developers are responsive and take bug reports and patches and actually show me what is going into the next revision if I care to look, but that’s the cherry on top of the whipped cream. If the code does exactly what I need and the license is OK, I don’t care if the aliens left it behind at Roswell and are never going to maintain it.

    I assume the very point of open source is not just to publish code, but to show stages of its progress in public where anyone can see what am I working on, to some degree, as I work on it. (Not *always* after I’m done with it.)

    That’s very reassuring when the project is going slowly. At least you can see that something is happening. But if you know something is happening, and is happening much faster than you and 100 of your closest friends could match working full time, why not just sit back and wait for the source to show up?

    If release (and the license) is the only thing that matters, then yes, you’re right and I have no reason to complain and no deception I should hold against Google.

    Obviously, the development model matters. Tremendously in some circumstances. But, from google’s perspective, the development model probably seems radically open. They probably accept contributions from lots of people in their partner companies. Different people working all over the code stack. Everybody seeing what’s going on. They probably have an internal “open” community that dwarfs that of the interested communities for most open source projects. And while they would like to open it further, that costs time and energy they don’t want to spend right now.

    From my perspective, what google is doing is not at all like what MySQL, TrollTech, etc. did (open source as a teaser/sales tool for proprietary licenses), and not at all like what the Symbian Foundation and others did (open source as a last ditch effort for a dying product), but is more in the true spirit of open source — here is something we are doing, we give it away free of charge, you can help, you can look at it (most of it, most of the time).

    If you think about it, their model excels where many closed source companies fall down badly — maintenance. Despite the catcalls about fragmentation, the “many eyeballs make bugs shallow” actually happens with Android, and patches happen.

  74. I think there are many good practical reasons for Google’s governance model over Android, and I think you’ve outlined them all extremely well.

    My only problem is that they obviously strike a (literal, not political) compromise between open release and “closed” (to the outside world) development model, yet in spite of this, they sell a central appeal of Android (according to their own statements) on philosophical ambiguity.

    Or perhaps, they’re acting like any other business (granting that the business wishes to use open source as a tool for product development), but their PR pretends they’re even better than their compromises suggest.

    Or good grief, I don’t think that’s unique…

    But, from google’s perspective, the development model probably seems radically open….[lots of rationalization how a closed development model with enough participants has the same benefits of scale as an open model]

    Yes, and yet…I suppose the philosophical argument is just pointless.

  75. @esr:

    I think they may be overestimating the friction costs of a project run along more open lines,

    I think they have deliberately injected some additional friction into their own processes, probably because it’s damned annoying to try to debug changing hardware and changing software at the same time:

    http://source.android.com/source/submit-patches.html

    If you read that link and grok everything that’s going on, you’ll realize that when somebody at google triages incoming patches, and the guys at cyanogenmod submit something simple that doesn’t show up in the main tree for three weeks because google’s busy with partner bugs, and the guy at cyanogenmod has to keep adding his patch back in to quickly changing downloads from google… Well, that’s a scenario with, not only friction, but a lot of public bad-mouthing that google just doesn’t need. The abuse they are seeing now pales in comparison to the abuse they would get about not accepting perfectly good patches for obviously broken stuff in the development branch.

  76. Or, if you consider development as a means to end [release], then Patrick would seem to suggest that the end justifies whatever wish-washy stuff happens before the release part. And from a results-standpoint (open stack, free to fork, still tons of participation), that seems exactly right.

    That is, as long as Google makes the final releases open-source, all of the benefits you’ve outlined above still hold.

    Maybe I just need to grow up a little.

  77. @esr:

    > It definitely does involve the development model itself. I wrote the foundational papers about the development model, not licensing technicalities

    @twilightomni:

    [lots of rationalization how a closed development model with enough participants has the same benefits of scale as an open model]

    > Or, if you consider development as a means to end [release],

    There is no question that the bazaar development model can result in higher quality code more quickly. However, I really do believe that if you take dozens of developers weaned on the bazaar development model, and ask them to develop stuff together in a private repository, (a) a lot of the benefits of the bazaar model will still accrue; and (b) the developers themselves will constantly push for more openness.

    > Maybe I just need to grow up a little.

    I think it’s reasonable to be suspicious of people who suddenly get the open religion. I really don’t think google falls in that category. They hire a lot of people based on their publicly visible resumes and then let them spend 20% on personal projects. It’s an excellent way to get great developers, but trying to close things more than necessary is an excellent way to bleed them right back out again, and I think they’re smart enough to know that.

    And yes, put bluntly, from a purely pragmatic, non-philosophical, selfish user viewpoint, the development model is a means, not an end.

    1. >I think it’s reasonable to be suspicious of people who suddenly get the open religion. I really don’t think google falls in that category.

      Certainly Google does not. The company’s founders sent me fan mail back when they were in the garage stage; for this and other reasons I believe that open source has been a central strand of Google’s DNA from its beginnings. One of the reasons I’ve been having so much fun analyzing Android’s rise is that I think my writings on the production economics of open source make me one of the grandparents of Google’s Android strategy, with Clayton Christensen as another obvious ancestor — what they’ve been doing reads very much as though they juxtaposed his insights about technology disruption to my work on the sustainability of open-source-centered coalitions and thought “Hey! Let’s tackle the smartphone business as though we take this seriously!”

  78. I think it’s reasonable to be suspicious of people who suddenly get the open religion.

    I think Google’s not-well-defined PRspeak is probably (or in fact already has) created a new generation of these people.

    Yes, Google does good things. I suppose I can trust in their good intentions enough not to worry about their deviation from straightforward ideological honesty.

    (I conclude this is the viewpoint espoused here, mainly one of earned trust. In fact the very situation of depending upon promises of openness reinforces this, and perhaps is just as useful and maybe slightly less deceitful than adherence to a supposed set of ideals that cannot be legally enforced.)

    But you could say these benefits accrue for many closed systems. At this point in the discussion, I feel a little unsatisfied.

  79. Please pardon my lapse into the dramatic in the previous post, as I clearly jumped certain foundational points of the conversation (i.e. Google *is* obligated to release certain parts of source by license), so it’s not like it’s entirely a blind trust relationship.

    I do think though that depending on Google being true to their word for eventual complete Android source releases is probably a good bet, and much less dramatic than arguing about whether they meet some set of principles for open development. However it just requires me to look past Google’s philosophical-tinted public statements about Android and rather analyze what actually *happens*, and frankly I’m either too unwilling to disregard their public statements, and I just feel a little disappointed that a great advocate of open source (which I know Google to be) isn’t afraid to bend the rules for its own personal mobile juggernaut.

  80. @ Patrick

    >Google accepts patches. Without requiring a code license assignment. And since Android (other than the kernel) isn’t GPL,
    >everybody else has exactly the same privileges as google.

    Small point of order. Unless I’m misunderstanding you, Google certainly does require a code license agreement:

    http://source.android.com/source/cla-individual.html

  81. > Small point of order. Unless I’m misunderstanding you, Google certainly does require a code license agreement…

    So they do. I confess I missed that.

    But, on reading it, you’re not really granting them anything more than they grant you. For example, sections 2 and 3 are taken almost verbatim from the Apache 2.0 license text (slightly modified to reflect that you’re the grantor, not the grantee), which is the license they use for the bulk of the project.

    Other than that, they’re just asking you to to represent that you actually have the authority to grant them the license.

    Contrast this with, e.g. OpenOffice, where they can do whatever the heck they want with your code, and you can do whatever the heck the GPL grants with their code. Oracle is in a superior position — nobody else can ship proprietary versions of OpenOffice, but they can and do (StarOffice), but Google puts you in exactly the same position as them, with the minor exception that they are asking you to “agree to notify the Project Leads of any facts or circumstances of which you become aware that would make these representations inaccurate in any respect.”

  82. @twilightomni
    “Is open source a set of development principles or not? I assumed it is. I assume the definition of open source project involves the development model itself.”

    @esr
    “The community has settled on licensing as a minimum constraint because trying to legislate about development procedure (including what would satisfy some legalistic definition of “openness”) is a snakepit nobody wants to wander into.”

    I disagree. Open must be about use of to the code, and most certainly not about the methods.

    No set of development guidelines can ever cover all projects. Statistics tells us that half the projects will be one-man projects, a quarter will be two person projects.. It is completely impossible to write something that would cover every project’s development team’s requirements. This is like marriage. You can legislate a procedure that recognizes people married. It is very, very unwanted to legislate how they should live together, or even whether they should live together.

    @esr
    “The company’s founders sent me fan mail back when they were in the garage stage; ”

    That explains their success ;-)

    Seriously, they would have failed had they not based their search engine business on open source software. Imagine they would have used Windows or Solaris. How could they then compete against MS or Sun/Oracle?

    1. >Seriously, they would have failed had they not based their search engine business on open source software.

      Much as I would like to agree with this, I’m not actually sure it’s true. The PageRank algorithm was some pretty powerful secret sauce which they still keep secret; it’s possible to at least imagine an alternate history in which they built the search-engine business as an entire closed-source stack. The Android play would have gone nowhere without open source, but that’s a different matter.

  83. I conclude this is the viewpoint espoused here, mainly one of earned trust

    I don’t even think you need to go that far.
    The viewpoint is less about earned trust(though i acknowledge that there is an element there) and more about “enlightened self-interest”. That is to say that Google has seemingly aligned itself (knowingly or not) to a path whereby defecting from open source harms them more than it helps. This neatly (and not co-incidentally) mirrors the state of GPL/OSS community. (E.G. how quickly the fork of open office took after Oracle went nuclear)

  84. There has been some discussion about tablets vs laptops. But what about netbooks? I find them more convenient than tablets mostly because they have full keyboards. I imagine a netbook in which the touchscreen panel that slides over a tiny keyboard panel, so you can see only the touchscreen when you want to and get the fully capable keyboard when you need to.

  85. > I agree with this analysis. I think they may be overestimating the friction costs of a project run along more open lines, but that’s a different argument from how they may ethically respond to an evaluation that those costs are too high.

    I suspect that the Android division of Google has naturally secretive tendencies and is not particularly good at open source. They were very quiet when they were Android Inc and seem to have generally embraced a cathedral development style. Then we had the wakelocks issue, where they decided to fork linux, rather than get the functionality they needed into the kernel. Now this. I think the issues are at the level of the Android subdivision, rather than at Google itself.

  86. What the hell is up with the pricing on Android tablets? It’s ridiculous. They have to come in $100 below iPad to have a prayer of selling in volume.

    Quite simply, they can’t do it. Apple’s got an economy of scale advantage that the also-rans can’t begin to touch. Apple is far and away the biggest buyer of flash ram, for a start.

    1. >Apple’s got an economy of scale advantage that the also-rans can’t begin to touch.

      I know this is one prevailing theory, but I don’t think it’s sufficient to explain what we’re seeing. Apple buying displays and flash RAM in ton lots is a signal that says to the producers of same “build more capacity!”. They’ve had a year to do it now and Android tablet prices are still crazy. Something else must be going on.

  87. @esr
    “One of the reasons I’ve been having so much fun analyzing Android’s rise is that I think my writings on the production economics of open source make me one of the grandparents of Google’s Android strategy, with Clayton Christensen as another obvious ancestor ”

    Google embody a lot of what Bruce Perens described in “The Emerging Economic Paradigm of Open Source”
    http://perens.com/works/articles/Economic.html

    Given the time-line, it is more likely Bruce got some of his ideas from Google instead of the other way around. In my memory, that paper too is from around 2000. However, the earliest copies I can find are from early 2005. So it must be my memory that is playing games on me.

    1. >Given the time-line, it is more likely Bruce got some of his ideas from Google instead of the other way around. In my memory, that paper too is from around 2000.

      It was last edited in 2005, but by 2005 it is impossible that Bruce wouldn’t have read The Magic Cauldron, which I released in mid-1999 and which covers the same territory in more depth (including game-theoretical analysis of incentives, a price theory of software and seven derived business models). After my paper this would have looked like an imitation; the only way it makes any sense that it exists was that Bruce actually wrote it between early 1998 (founding of OSI) and July 1999 (release of The Magic Cauldron) and it represents his attempt to think through the economics in parallel with mine.

      It is just barely possible that Bruce influenced the Google founders during that crucial period. Highly unlikely that they influenced him; they were just another small startup then, and I know they didn’t infuence me; not visible or exceptional enough. The fanmail I got shows they were paying attention to my work. In 2000-2001 my economic analysis hit the NYT best-seller business-books list and hung out there for 7 weeks, but I’m pretty sure Dave and Sergei were hip to it almost eighteen months sooner.

  88. Quite simply, they can’t do it. Apple’s got an economy of scale advantage that the also-rans can’t begin to touch. Apple is far and away the biggest buyer of flash ram, for a start.

    The whole “biggest buyer of flash ram” meme seems somewhat played to me, I can’t find any support for it over and above an isuppli statement to that effect. I wonder what they define to be flash ram and how SSD manufacturers come into it. However thats not really the point i want to make.

    If flash ram supply is the problem then I’d imagine the price takes a nose dive in the next 3-6 months as supply ramps up(i’ve seen one link that states Toshiba is implementing a manufacturing process that will double production in Q2 2011). I think its vastly more likely however that the price is fueled more by limited competition in Android terms (there’s what, 3 official android tablets and a couple of also-rans?) or alternatively that general demand is just not as high as the rosy glasses suggest.

    I am however interested in the ASUS Eee Transformer.

  89. Ah, the prediction below actually should go in the WP7/Nokia articles, but it is not completely off-topic here.

    Note that IDC predicts a compound annual growth of Microsoft’s platform over the next four years is expected to be 67.1 percent.

    That is not correct, I think. The growth is expected to come from Nokia, who will not sell a WP7 phone before 2012. So the growth should be calculated from 2012 on, that is over 4 years, not five. That would then be a CAGR of 90% over 4 years. How they could ever achieve that when entering a market with some 750 million active phones (IDC’s numbers over 2010+2011) is anybody’s guess.

    I think this is pure marketing.

    IDC: Android, Windows Phone to rule mobile
    http://news.cnet.com/8301-13506_3-20048236-17.html

    According to the market researcher, more than 450 million smartphones will ship in 2011, up from the 303.4 million units last year. IDC predicts that Android will account for 39.5 percent of all smartphone shipments this year, easily besting Symbian’s 20.9 percent market share. Apple’s iOS platform is expected to be running on 15.7 percent of the smartphones shipped in 2011. RIM’s BlackBerry and Windows Phone 7/Windows Mobile will round out the top five with 14.9 percent and 5.5 percent market share in 2015, respectively.
    ……
    Microsoft’s expected gains in the smartphone market aren’t by chance. The company’s market share boost will be due mainly to its recent partnership with Nokia, which will be using Windows Phone 7 as the “principal” operating system in its smartphones likely starting in 2012.

    Always nice to compare with older predictions. Note IDC reports Windows mobile has 5% market share in 2011, whereas they predicted 10%.

    IDC boosts 2010 smartphone shipment projections by 10%
    http://www.computerworld.com/s/article/9183840/IDC_boosts_2010_smartphone_shipment_projections_by_10_

    In 2014, Android will grow to nearly 25% market share, while Windows Phone hits nearly 10% as the market share of all the other major OSs decline, IDC said.

  90. @esr,

    Probably I mixed up the Magic Cauldron with Perens’ analysis as I have read both (long, long time ago, though).

    What I liked about Perens’ work is that he estimates the efficiencies of production, eg, the extremely low efficiency of retail software production. Less than 10%. Which latches well with MS’ margins. I forgot whether the Magic Cauldron made the same estimations.

  91. @Winter:

    I disagree. Open must be about use of to the code, and most certainly not about the methods.

    I figured the whole point of open source was about the process itself, not just the outcome. I think it’s great that Google is using some ideas from open source to help accelerate the development and spread of their system, but I can’t get over the fact that they skipped one that I thought was integral to the whole thing.

    I also feel a little foolish because I feel very odd belaboring this point, and yet everyone else here has no qualms with it.

    It’s not like I actually care whether Google picked the particular parts of open source principles that were most expedient to its success and ignored a few others. That really doesn’t matter to me. They can develop and release in any way they want to as long as they make a good product. I think everyone here is so excited precisely *because* Google is using these ideas to help make a better product.

    But then I don’t hold Google to any idealistic standard either, and it seems wrong to me that they profess one that they only seem to use parts of for its benefits. By the reasons above, if Google found another set of development or release methodologies that would give its platform a product advantage, what then?

    I assume the point of open source is the ideal, not the end product, however great it can be. And I feel kinda stupid saying all of this vis-a-vis arguments that are much more well-adjusted and realistic.

  92. @twilightomni
    “I figured the whole point of open source was about the process itself, not just the outcome.”

    To come back to my analogy. Replace open source with marriage. That is clearly a “legal construct”. Some relations are marriages by law and some are not. And we would all agree that marriage is not about the legal status, but about the “process”.

    We will judge the quality of a marriage by the “process”, and not by the legal ceremony. But would we want to have the process be part of the legal proceedings? Should there be laws codifying how married couples should live together?

    In the same vine, judge Google on their deeds and processes. Do so with any Free and Open Source project. But do not try to tell others they are not Open Source because you do not like their processes. There are countries where they try to regulate how married couples should live, and these are not the countries you want to live in. The same will hold when you try to regulate how Open Source projects should handle development.

  93. So…open source is a process…and don’t legally codify it. But my earlier proposition is that open development is the *process*, and open source is a collection of principles including that (I suppose).

    I agree that open source cannot be legally codified, but if the use of the principles cannot be recognized (“do not try to tell others they are not open source because you do not like their X”) then what’s the point of this discussion?

    I believe that there is common knowledge (not legal, but a defined philosophy) how an open source project should handle development. It’s…get ready for it…open source as well. That’s why I’m confused about all of this in the first place.

    Again, if you think open source is all about release and not about development, then fine. Just say it, and allow me to disagree or express a reasoning that can be proven fallacious or rejected. I’m not trying to regulate what open source is, I’m trying to figure out whether it really is what I think it is or not.

  94. Also, courtship is the *process*. Marriage is neither a process nor a set of principles, but a social construct. The metaphor is incorrect.

    And you can in fact correctly judge courtship, in the proper social context. Just because I don’t have the knowledge to properly judge courtship in China doesn’t mean it can never be evaluated.

  95. There are countries where they try to regulate how married couples should live, and these are not the countries you want to live in.

    Perhaps there is where the confusion lies, because the United States is one of those countries. :) Married couples who do not share a residence for more than 6 months are considered “legally separated,” for example. (YMMV by U.S. state).

    Overall, though, I like your analogy and it makes sense given that is the way the community sees it: the license defines what is and what is not open source, not the methods.

  96. @twilightomni
    “So…open source is a process…and don’t legally codify it. But my earlier proposition is that open development is the *process*, and open source is a collection of principles including that (I suppose).”

    What you are looking for is Cathedral/Bazaar development, or any other description of the development process.

    Open Source is any project/code that is licensed under one of the recognized Open source licenses. That license defines what is legally open source. And this gives the legal limits who can participate (everybody) and when and how.

    The process is what uses the copyright license to actually do the work. You can use the same processes without an Open Source license, but that does not make it open source and you lose the “protection” of the law. And you can use any process with your Open Source license, and that remains Open Source. Because of the rights granted to the receivers of the code.

    Why this matters? I remember some cases where CD title music libraries were build by a community and then taken private, excluding the contributers from the database. The contributers had no legal right to keep their own copy of the database. There are numerous software project examples from the pre-open source times.

  97. Also, courtship is the *process*. Marriage is neither a process nor a set of principles, but a social construct. The metaphor is incorrect.

    Marriage is also a contract. It’s a contract entered between two parties in view of the state. Theremore, a marriage is a kind of legal construct.

    This is similar to an open source software. Open source software is a work, sure, but it is a work protected by a copyright, which is another legal construct. The open source license is a permission — a waiver of rights — given by the copyright holder. The license is not, and cannot be, a contract.

    BTW–the process of a marriage isn’t the courtship — that’s process of developing the relationship. The process of a marriage is what the two people do and how they live once they are married.

  98. Then you are talking about the relationship component of marriage, and not the legal construct.

    So open source can likewise be defined here as a collection of processes and adherence to social and legal constructs. But surely no one argues that both open source and marriage exist at a philosophical level as a collection of ideas and not merely societal constracts?

    And that if those principles are not met, then what is the point? You may, for example, define a marriage that barely meets the legal standard (that is, a contractual obligation, with no love, indeed, no actual physical or mental interaction – let’s say permanent absence between spouses with just enough proximity every 6 months to prevent legal separation). And so we have created a marriage that is definitely not a marriage in any sense of common use of the idea. And what good is that?

    Aside from possibly the tangible economic benefits, but at that point we have stopped talking about the merits of marriage itself, and started talking about ancillary benefits that come through the contracts but not from the idea itself.

    Isn’t the philosophy the more important idea here?

    “Open source” by any definition I’ve seen is a set of principles, not conformance with a legal license. The license codifies one particular principle (I shallowly refer to as remix/reuse). A single release of source code can be released under a so-named “open source license”, but an individual release is not correctly described as “open source.”

    A project is either open source, uses some principles of open source, or does not. Projects need not be open source if that is not convenient for them. Or, projects may surely use the principles as they see fit. But does selective use of open source principles make an open source project? Which principles are sufficient or integral to the description? And does “open development” constitute as one of those principles?

  99. @twilightomni
    “And that if those principles are not met, then what is the point? ”

    What is the point of legal marriage? Not when all is love forever.

    The legal part of marriage kicks in when either the love or the forever part is gone. You learn to appreciate the legal aspects when either spouses die or when there is a separation. Who gets the kids? Who gets the house?

    Why do people marry? Because when the mother dies, the father gets the children. The mother gets the house when the father dies (or vice versa). The house is split in a divorce etc.

    Why do people want an open source license? Because when the developers dies, the users get legal hold of the code. And when the developers separate, everyone ends up with their own legal copy of the code.

    And how the source is developed is something different. Certain processes are better done with an Open Source license. But over all, that is an orthogonal dimension.

  100. >Open Source is any project/code that is licensed under one of the recognized Open source licenses.
    >That license defines what is legally open source.

    Not directed at Winter, more using his comment as a reference point. So if we decide that “Open Source” is ultimately defined by the license/outcome and not the process or the “openness” of the development model, an interesting question arises. How far away from the current version are you allowed to keep closed before you aren’t open source anymore?

    Say for example that Mozilla or Google has the following branches:

    Software 4.0 (development major version closed source)
    Software 3.6 (current released minor version OSS)
    Software 3.5 (prior release minor version OSS)
    Software 3.0 (prior release major version OSS)
    Software 2.0 (prior release major version OSS)
    etc etc etc, and assume that as new versions are released, the line where it switches from closed to open moves up the stack.

    We seem to have established that this is “Open source”. Tomorrow Mozilla or Google or whoever decides that their business interests are best served by moving that close line below the 3.6 branch of software, just like their 4.0 branch now. Does this still meet our definitions of open source? What if they close 3.5 so that only major releases are OSS? If they close 3.0?

    And this is a serious question its not snark. Somewhere obviously I think we have a line. No one would say Windows is open source if Microsoft decided that all EOL versions of windows would be released under open source licenses would they?

  101. > Somewhere obviously I think we have a line. No one would say Windows is open source if Microsoft
    > decided that all EOL versions of windows would be released under open source licenses would they?

    In the magical fantasy land where that happens, I would call them ‘friendly.’

    Much like how John Carmack has historically released the source to id’s engines after some threshold of profitability (for id and for id’s licensed partners) has been passed. I wouldn’t call id ‘open source,’ but I would say as a profit-seeking company they have been very friendly to the idea and the community.

  102. @twilightomni:

    Love has nothing to do with marriage in the legal sense, nor in the process sense. Marriage is definitely a process that it is lived day by day, and it works only if the two people involved make it work. Anyone who is married will agree with me.

    As for open source: Winter seems to have the right idea, but I’ll add that just as marriage creates certain legal rights (one can decide many legal and medical matters for an incapacitated spouse, for example) open source license create certain legal rights as well — if a copyright holder decides to close up the source for future releases, one can fork the existing codebase.

  103. >if a copyright holder decides to close up the source for future releases, one can fork the existing codebase.

    Could they, if there were contributions involved that weren’t explicitly assigned to them? Maybe after scrubbing anything questionable? I know that some projects/companies require copyright assignment to them, but I didn’t think most do.

  104. And this is a serious question its not snark. Somewhere obviously I think we have a line. No one would say Windows is open source if Microsoft decided that all EOL versions of windows would be released under open source licenses would they?

    That’s only theoretical. I’m pretty sure that isn’t going to happen anytime soon. But, even so, it’s possible to have an “open source version” and a “commercial closed source version”. Think VirtualBox OSE vs. VirtualBox PUEL, or the OpenOffice.org/StarOffice split, or even the former licensing scheme used by Trolltech for Qt (sort of).

  105. @Morgan Greywolf

    You must never have been married to be able to type that. Marriage is definitely a process….

    Love has nothing to do with marriage in the legal sense, nor in the process sense. Marriage is definitely a process that it is lived day by day, and it works only if the two people involved make it work. Anyone who is married will agree with me.

    That is correct on both counts and I do agree. It’s a mistake for me to say that marriage is not a process; I mean to say that the process is not the only part of it.

    Although love has nothing to do with the legal component nor the process component of marriage, I do think you’d agree that you certainly hope love will be in there somewhere, as in the absence of any other benefit it would be a little unfulfilling without it.

    But this just me conjecturing, as neither have I started my own open source project, nor gotten married.

  106. > They’ve had a year to do it now and Android tablet prices are still crazy.

    A year is enough time to build a factory for final assembly, PCB stuffing, etc. It’s not enough time to bring up a new chip fab or display glass line. Apple’s sales of iOS devices keep outstripping their most optimistic projections. I don’t believe that Motorola and Samsung are charging any more than they have to, since their development costs have to be amortized over a minuscule portion of Apple’s unit volume. Has the Xoom even moved 100K units? I doubt it.

  107. Marriage is a terrible analogy for software development, at least in the U.S., because it is a mixture of three completely different things.

    — It’s a special personal relationship between two people (more than 2 in some societies)
    — It’s a legal thing that is defined in civil law, and provides rights and responsibilities to those who use it.
    — It’s a religious thing to some (not all), covered by religious laws.

    Any given marriage can have one, two, or all three of these. Examples:

    — The United Church of Christ recognizes gay/lesbian marriages. A lesbian couple who live in a state that does not recognize them in civil law may be married in their own eyes and those of their church, but not their civil government.
    — A couple married in a courtroom may not be legally married in the eyes of their church, even if they are religious. For example, they may be divorced but otherwise practicing Catholics.
    — A couple that has been separated but not divorced may be married in the eyes of the law and their church, but no marriage exists in terms of their personal relationship.
    — A couple who have lived together for a significant period and even have a child together may still not be married in the eyes of their church (if any) or the law (depending on common-law marriages statutes), yet may have a strong “marital” relationship.

    This area is a minefield, and I wouldn’t use it as an analogy because it will just create more confusion.

  108. twilightomni said After all, isn’t philosophical strictness the reason for the creation of GNOME in the first place?.

    And I view that as an invitation to snark, just a little, when I say:

    Well, that explains a lot about their UI/UX decisions, doesn’t it?

  109. @William B Swift:

    if a copyright holder decides to close up the source for future releases, one can fork the existing codebase.

    Could they, if there were contributions involved that weren’t explicitly assigned to them? Maybe after scrubbing anything questionable? I know that some projects/companies require copyright assignment to them, but I didn’t think most do.

    That depends entirely on the license. Above the kernel, Android is Apache-licensed. That means that anybody, not just google, could start a private branch where they make changes and only give away or sell binaries. For the kernel, they either need to keep shipping the sources, or not bother shipping binaries (let people get the kernel image elsewhere).

  110. @Some Guy:

    Quite simply, they can’t do it. Apple’s got an economy of scale advantage that the also-rans can’t begin to touch. Apple is far and away the biggest buyer of flash ram, for a start.

    For the sake of argument, I will assume that you are correct about Apple’s size as a flash memory (it’s not RAM, BTW) buyer.

    Every semiconductor company knows that Apple will throw them under the bus for another nickel. Even if Apple were the greatest customer ever (which I can assure you they are not), any semiconductor company will get very nervous if their top customer takes more than 30% of their shipments, or if their top 3 customers take more than 50% of their shipments. It is in the flash vendors’ interest to diversify their customer base. To the extent that Apple paid millions of dollars upfront to lock in shipments, you can bet your bottom dollar that a substantial portion of those payments immediately went to wafer manufacturers and packaging and shipping houses to lock in shipments there, and from there, a portion of those payments was cautiously invested in increased overall capacity.

    On a more personal note, my own personal favorite flash devices, which were unobtanium at this point last year due to the secretive iPad rollout, are now in stock at half a dozen different domestic semiconductor distributors.

    But it’s a huge mistake to think that iPad production is fundamentally different than smartphone production is fundamentally different than dumbphone production, etc. With the possible exception of the screen, and maybe a few specialty components like gyroscopes, most of the components are either usable in any of the above, or the manufacturers can decide to tweak their product mix to build more of this kind for the iPad, and fewer of that kind for the cellphone. Sure, the iPad takes more memory, for example, but the industry can catch up relatively quickly, and no chip company wants Apple to be their master.

    The semiconductor business is heavily cyclical. Nonetheless, the companies do an excellent job overall of managing the cycles. It’s only when Apple sneaks around behind their backs, not letting anybody know what’s going on and then putting out a few million of a brand new product category that they get really surprised. but now that they understand Apple a bit better, they’ll be watching more closely. Here’s how cyclical the flash industry is, for example:

    In an effort to reduce the oversupply of NAND flash memory as ASPs are projected to fall by approximately 65% in 2008, JV partners Toshiba and SanDisk are attempting to cut production by 30% with production stoppages.

    July 14, 2010
    Toshiba, SanDisk look to boost flash production
    Their planned ‘Fab 5’ fabrication site will turn out 210,000 wafers a month

    As far as the larger items like screens, CPT (probably the largest LCD manufacturer you never heard of) had significantly reduced sales in February. I’m sure they’d love to make more money. They’re in a cyclical industry, too, and in the midst of the post-Christmas trough, from the looks of things.

    In short, your central thesis about Apple’s economy of scale rings quite hollow. Apple’s just a design/marketing company, and Foxconn and others will build for anybody, and all the component vendors will sell to anybody. Yes, companies will give a bit of a discount to get a big customer like Apple, but not as much as you might think, because losing such a customer is too painful, especially if they weren’t paying enough for the vendor to be able to stuff some under its mattress to prepare for that day.

  111. > That depends entirely on the license. Above the kernel, Android is Apache-licensed.

    Except for the parts derived from GPL code, of course.

    “Nothing to see here folks! Move along…”

    No source code for your open source Android device?

    “Remain calm! Don’t panic!”

  112. @Life as we know it:

    That depends entirely on the license. Above the kernel, Android is Apache-licensed.

    Except for the parts derived from GPL code, of course.

    You’re a troll without a cite. And I don’t mean “he said that she said…”. For any non-Honeycomb Android device, the source is out there. Link to a piece of code, and a GPL-licensed piece of code and explain how they are identical, why exactly there is a license violation, and exactly why we haven’t heard of a lawsuit about it yet.

    1. On a mailing list I frequent, one of the regulars turns in this report:

      For those who care, the actually decent Android tablets at low prices are here and shipping. Check out the Dropad A8, AKA Herotab C8. 7″ capacitive multitouch screen that is reviewed as being as good as the iPhone/iPad, 1 GHz samsung processor, Android 2.2, 512M RAM, 4G ROM plus up to 32GB micro SD expansion, USB/OTG ports, 1.3mp front facing camera, HDMI and 1080p video playback. There are video reviews and the thing looks sharp; super fast responsive screen movements, nice screen. A few rough edges that you can deal with by tweaking drivers (the price you pay for the low $$ and early shipping models). I’m sure they’ll get those ironed out and ship with the newer drivers soon. No Android marketplace as shipped but apparently that’s already been patched in the wild. Or you can just sideload.

  113. @esr:

    > Check out the Dropad A8, AKA Herotab C8

    Interesting. The very first google result for either “dropad A8” or “herotab C8” turned up a vendor I’d never heard of, merimobiles.com, selling them for $200.

    Either they’re legit, they’re extremely good at SEO, or there just aren’t enough “herotab C8” pages out there yet. In any case, I wouldn’t expect a name brand premium for this sort of thing to be more than $50 – $100. Expect more B&N Nook Color class devices at B&N Nook Color class pricing, with the price ramping down, and the quality and features ramping up.

    1. >Expect more B&N Nook Color class devices at B&N Nook Color class pricing, with the price ramping down, and the quality and features ramping up.

      Oh yeah. This hardware dissection shows that the C8 is not a single-chip device – the flash RAM, radio, audio codec, and a couple of unidentified chips are outboard of the processor. That means there’s plenty of room for the price to drop as the Nvidia and Qualcomm SoCs become available in Q2 and Q3.

  114. Patrick,

    Have you ever talked to a component vendor who’s doing business with Apple? You’re vastly underestimating just how much they’re willing to do to get and keep Apple’s business. Some people think Wal-mart drives a hard bargain; they’re amateurs compared to Apple’s procurement staff.

  115. > A few rough edges that you can deal with by tweaking drivers

    In the consumer market, that’s what you call a deal-breaker.

  116. > Have you ever talked to a component vendor who’s doing business with Apple?

    I used to do product development for an IC vendor who did significant business with Apple. Apple rolled over in bed and that company faltered and got acquired and I’m working somewhere else now. So, yes, I understand intimately what you’re talking about. However, I also know there are a lot of more well-managed companies that will go so far to work with Apple, and no farther, and I will wager that the sum total of the market cap and production capabilities of those well-managed companies (starting with the big guns like Intel, Samsung, SanDisk, Toshiba, and Micron) far exceeds that of the little guys who don’t know any better.

  117. @Cathy
    “This area is a minefield, and I wouldn’t use it as an analogy because it will just create more confusion.”

    I understand that you point out that marriage is legally complex, with different jurisdictions. There are many emotions and strongly held believes involved. People will fight each other on ephemeral points of interpretation.

    But how is that different from Open Source?

  118. @Patrick – Thanks. It didn’t even occur to me to consider the difference that licenses would have on contributions; I’m not a developer and had never thought about them from that perspective.

  119. @tmoney
    “And this is a serious question its not snark. Somewhere obviously I think we have a line. No one would say Windows is open source if Microsoft decided that all EOL versions of windows would be released under open source licenses would they?”

    The wonders of language. The meanings of a word are all in context.

    We have Open Source code. That is code available under an open source license. Say Symbian or OpenSolaris. A theoretical EOL version of MS windows licensed under a recognized Open Source license would meet that definition.

    We have open source OS’. That would have to mean all versions in use, except maybe the latest development release, must be available under an open source license. OpenSolaris and Symbian would have fit that bill.

    We have Open Source projects. That would be some organization that only releases Open Source code. It might be a shortcut for all the people involved with a specific Open Source software package. So, OpenSolaris and OpenOffice might have been called Open Source projects even though they were hosted by a predominantly closed source company.

    This overlaps with Open source development which is collaborative development around Open Source code. It is mostly here that there seem to be all kinds of philosophical differences and holier than thou posturing (I deliberately will not identify along what dimensions holiness is measured).

    Note that none of these have to be in active use to be called Open Source. You can dump some legacy code under an Open Source license on your website with absolutely nobody downloading a copy. The code would still be Open Source. But there would be no Open Source project or Open Source development.

  120. Oh yeah. This hardware dissection shows that the C8 is not a single-chip device

    If that guy is right about the 2 gig NAND chip then that is surely evidence that the NAND explanation has issues. If there’s a massive shortage of chips why would you stick a 2 gig chip on there and not use it. The obvious extension is why would you do that anyway.

    1. >The obvious extension is why would you do that anyway.

      I was wondering that myself. My best guess is that this thing is a rather hacked-up version of somebody’s reference design and it was cheaper to patch the design than to spin a new board. If true, this theory has a couple of implications.

      First, as you say, “What NAND shortage?” But I never believed that fairytale anyway – there’s barely any part that’s cheaper to pump out in volume if you’ve already sunk the fixed costs for a fab.

      Second, there are burn-it-and-ship-it reference designs floating around out there, including PCB artwork, that are so cheap (or free) that production of this class of tablet is limited only by the parts-supply chain. Not hard to believe; if I were ARM or Nvidia or Qualcomm I’d have engineers writing those to give away so I could boost sales.

  121. Not quite on-topic, but illustrative of why MS can be ignored in this field. When reading this, I got a deja vue feeling. That calcified behavior was exactly what brought down Nokia. They have the solution in hand every time, but simply fail to actually grab it. So, indeed, two turkeys will not make an eagle. And turkeys they have proved to be.

    The problem with Microsoft…
    http://tech.fortune.cnn.com/2011/03/29/the-problem-with-microsoft/

    The Microsoft (MSFT) team working on Courier was equally jazzed. “We had a breakout product that had the potential to really delight the user,” says Rebecca Norlander, a star programmer inside Microsoft who quit last June after 19 years with the company. Just as important, Courier held the promise of catapulting Microsoft into mobile devices, a lucrative field that had eluded Microsoft for 15 years — and where rivals Apple (AAPL) and Google (GOOG) had made significant inroads.

    So when Robbie Bach, who led the company’s entertainment and devices division at the time, presented his idea to CEO Steve Ballmer and Microsoft’s senior leadership, he expected enthusiasm and additional funding for the project. There was just one problem: The Courier prototype borrowed from Windows, Microsoft’s vaunted computer operating systems, but had an operating system all its own. (That’s what Apple did with its iPhone and iPad — it built a new operating platform based on its existing Mac OS X.)

    Bach learned a hard lesson about the power and might of Windows within Microsoft. Not only would Bach not receive the extra funding he sought, said Ballmer, who personally delivered the blow, but there would be no Courier because it was unnecessary. The best of Courier, where appropriate, would be folded into the next version of Windows, Windows 8, due at the end of 2011 or in 2012 — or maybe even Windows 9. Several months after its death, Bach announced his retirement. (He wouldn’t comment for this story.)

  122. @esr:

    First, as you say, “What NAND shortage?” But I never believed that fairytale anyway – there’s barely any part that’s cheaper to pump out in volume if you’ve already sunk the fixed costs for a fab.

    Well, there was a shortage last spring/summer, and these things happen periodically, but not usually to that extent. Free market in action — there are always some available at some price, except back then, for some flash devices, there weren’t.

    One problem is that the raw materials (the wafers) cost a lot, too, as does keeping the fab running. Hence the article I quoted earlier about reducing output in 2008. And once you’ve built the part, it’s inventory (interest) cost if you don’t sell it as quickly as possible. So the manufacturers are careful about managing production. But to do it properly, they need good forecasting, and that’s what Apple fucked up for them. (But they’ll be watching now.)

    The interesting thing is that Apple’s move to secure more parts by prepayment, while extremely helpful to the industry, is purely self-serving. By being so secretive about what they were doing with the iPad, and needing to buy a lot of parts on the spot market, Apple must have significantly increased their own BOM cost, in addition to making it impossible to build enough to satisfy initial iPhone 4 orders.

    BTW, it’s not only the component manufacturers who really, truly want to diversify their customer base — the assemblers do, too. So, if you can do a modicum of design (not much, as the rushed hacked-up boards show) and a modicum of marketing, all the other pieces are in place for you, too, to start shipping competitively priced Android products. So I stand by my theory that Samsung and Mot both thought there was a lot of pent-up demand at the very high end for a non-Apple product. They should be figuring out the truth about now.

    1. >So I stand by my theory that Samsung and Mot both thought there was a lot of pent-up demand at the very high end for a non-Apple product. They should be figuring out the truth about now.

      /me shakes head

      The stupid…it burns…

    2. >BTW, it’s not only the component manufacturers who really, truly want to diversify their customer base — the assemblers do, too

      I get why the assemblers want more different buyers compete for their business, in general. But it seems screwy to me that Foxconn is in trouble. You’d think the shift from shipping Nokia dumphones to higher-margin devices like the iPhone would be good for their profitability rather than a problem.

    1. >Perhaps a mild NAND flash shortage coming up now, due to events in Japan:

      Yes, that actually occurred to me as I was writing the comment you were replying to :-)

  123. Having worked at one of the semiconductor suppliers of Apple (won’t name names, suffice it to say it’s in the top 5), I will tell you we bent over backwards for Apple, they we such an important customer. Of all the customer audits we had, I remember that one best, because of the conniptions management exhibited in the days prior to it.

  124. This seems newsworthy:

    Playtime is over in Android Land. Over the last couple of months Google (GOOG) has reached out to the major carriers and device makers backing its mobile operating system with a message: There will be no more willy-nilly tweaks to the software. No more partnerships formed outside of Google’s purview. From now on, companies hoping to receive early access to Google’s most up-to-date software will need approval of their plans. And they will seek that approval from Andy Rubin, the head of Google’s Android group.

    1. >This seems newsworthy

      As a commenter notes, the only named source is the CEO of Nokia. This reads like a hatchet job, not journalism.

  125. @esr:

    But it seems screwy to me that Foxconn is in trouble. You’d think the shift from shipping Nokia dumphones to higher-margin devices like the iPhone would be good for their profitability…

    The assembler gets paid for, well, assembly. End-product value is not as important as assembly complexity. Complexity will drive revenue somewhat, but profit can actually be eroded by too much complexity, if assembly takes longer than anticipated, and/or if production problems cause the yield to drop, resulting in rework or scrap. But, all in all, from an assembly perspective, a tablet is probably no worse (maybe even less complex) than a smartphone. You’d have to carefully analyze some teardowns, but there’s a lot more room in the form-factor.

    So, a lot of Foxconn’s revenue and profits are going to be driven by unit volumes, rather than retail value per unit. Since Android rather than Apple is eating Nokia’s lunch, replacing Nokia units with Apple units is not a net positive in terms of absolute numbers of units, and there will probably be an accompanying decline in revenue and profit.

    While Foxconn has 50% of the non-captive assembly business, a lot of Android phone are manufactured by the other 50% of assemblers, or manufacturers do their own captive in-house assembly. For example, ZTE has 70K employees. Obviously, that’s not just marketing and design.

    1. >The assembler gets paid for, well, assembly.

      Hm. Well answered, sir!

      Your analysis implies that the oncoming wave of Android SoCs is not going to be particularly glad tidings for the assemblers, because their whole point is to drive down build complexity. Hadn’t thought about it this way before, but one of the effects of those chips may be that more manufacturers move their assembly in house.

      Now I get where that article was coming from. Being a Foxconn executive is going to seriously suck over the next several years. Between Nokia’s volume going away and assembly tending to go captive again, that is. If I were them I’d be looking at trying to climb the value chain.

  126. @Matt Minton:

    Yes, any company with half a brain will support Apple extremely well. But will also charge enough to pay for that, and not simply sell things to them at a loss and hope to make it up in volume. I wouldn’t be surprised if Intel has a dedicated Apple support team, but I would be surprised if Apple extracted significantly better pricing from them than Dell or HP.

  127. > If I were [Foxconn] I’d be looking at trying to climb the value chain.

    Agreed, but that’s where being too far in bed with Apple could be problematic. Don’t want to piss off the customer, you know.

    In any case, this whole subthread started because Some Guy said “Apple’s got an economy of scale advantage that the also-rans can’t begin to touch.”

    Which is patently untrue. Foxconn (and others) would be more than happy to assemble for any of the also-rans, at a price competitive to what they’re losing from shrinking Nokia. As far as components go, everybody in China/Taiwan/Korea knows exactly what everybody else in China/Taiwan/Korea is paying for components. There seems to be no such thing as a valid NDA, which means that transparency rules the day. Sure, a smaller vendor (of which, BTW, Samsung, HTC, and ZTE are not) might expect to pay a bit of a premium over what Apple pays, but it will be a small premium, and the smaller vendor expects to take a lower margin both coming and going, based on the fact that he has a much lower design/marketing/sales overhead, and doesn’t need Apple’s insane margins to keep the lights on and his shareholders happy in any case.

  128. Patrick said: I wouldn’t be surprised if Intel has a dedicated Apple support team, but I would be surprised if Apple extracted significantly better pricing from them than Dell or HP.

    I would also not be surprised – because unlike (as far as I know) HP or Dell, Apple’s actually gotten Intel to make them special Apple-only processors.

    (EG. the Penryn C2Ds in the iMacs in the E8x35 range, and the Mac Mini P7350/450, both of which only had VT-x for Apple.)

  129. @Sigivald:

    There is no question that Apple pushes Intel hard for features it needs. But are there truly any “Apple-only” Intel CPUs? For example, both the P7350 and P7450 are available in other machines, and you can buy the CPUs.

  130. >Foxconn (and others) would be more than happy to assemble for any of the also-rans,

    Foxconn and the rest of the contract manufacturers charges are based on volume. Anything you build in the millions is going to be far, far cheaper to make than anything you build in the thousands. Motorola and Samsung aren’t just charging arbitrary amounts because they think they can get away with it, they’re charging what they have to, given the small quantities they’re able to move.

  131. >As a commenter notes, the only named source is the CEO of Nokia. This reads like a hatchet job, not journalism.

    Hatchet job or no, I’m betting we see some movement in this direction from Google. Android and the “Android Experience” are going to become too valuable to not have some sort of Android Cathedral Edition branch.

  132. @Some Guy:

    I still don’t buy your argument. There is no question that building 2 units costs less than 1, 3 less than 2, etc. At the limit, assembly cost per unit follows an exponential decay curve (which will never hit zero due to some fixed costs). Nonetheless, small numbers of millions (like Mot and Samsung are hitting) is pretty far down the curve.

    Let me put it this way — the shipment numbers that Mot and Samsung are hitting are logarithmically indistinguishable from the numbers that B&N is hitting with the Nook Color. Assembly on the bigger tablets can’t cost much more (if any) per unit. Component costs (especially screen and faster processors and more memory) could easily be almost double. Why is the price more than triple? And don’t give any handwaving about the Nook’s razor/razorblade strategy — sure B&N wants to make money on the backend, but they don’t seem overly concerned about hackers messing with their precious IP, and at one point they were giving away coupons that made the retail price $200, or only one fourth of the price of the Xoom, so it’s hard to imagine they’re losing too much, if any, per unit at $250.

    Maybe, unlike Apple, Mot and Samsung just haven’t yet figured out their purchasing guys are making a killing on the side.

  133. Patrick, you’re not accounting for the fact that Moto and Samsung have to amortize their *development* costs over far fewer units, too. They’re both billion-dollar business, and they spend a lot more money than you realize to get unit one of anything out the door than a 100-man shop would.

  134. @Some Guy:

    I realize that perfectly well. And maybe they’re going to ride the price curve down gracefully after they shake down the early adopters for extra dough. Or maybe they just significantly miscalculated, and are going to get their lunch eaten from the bottom. No skin off my teeth either way.

  135. I said it before and i said it again….

    Honeycomb UI is a big failure!! It is way too desktopy!

    When you make it Desktopy , Just remember your competition is not the Ipad…. but windows7 netbook/notebook and Macbook Air!

    $800 10 inch High end Honeycomb 3G tablet vs $1000 11.6 inch Macbook Air or
    $600 10 inch Low end Honeycomb wifi tablet vs $250 10 inch netbook!

    Honeycomb is bound to fail this year…better luck next year!

    By the way what is all this stupid argument about android being fully or not fully open source?? That is a 2 year old debate! What you guys should look at is…….

    Will Android be continually be open source(Code available for all)?

    With HoneyComb it is totally close even with a near 2 month realease of xoom Tablet for the masses. So Android was open and now it has become close!

    But what about GingerBread, froyo and the rest? Well that is still open.. But those are all Phone/Mobile OS.

    All of Android tablet Os is close! None is Open!

    Will Google open it? I doubt it…..
    The reason is Google choice of a desktopy UI.

    but we should at least give google the benefit of doubt. we all should wait till Google IO for any chance of it being open or remain close.

    The next version Ice cream will be for the Phone/Mobile OS with element of Honeycomb UI in it. The Phone Os version should remain open……

    unless….As Google do…. DON”T BE EVIL!!!!!!

  136. Google holds back Android Honeycomb; Asus releases the source code
    Control-freak accusations are unfair, Google’s Android boss argues
    http://www.itworld.com/mobile-wireless/155149/google-holds-back-android-honeycomb-asus-releases-source-code

    As if to back up the contention by Google’s Android boss that the tablet version of Android isn’t being penned in so Google can keep control, PC-maker Asus released part of the source code yesterday.

    Asus posted a link on the product page for its Eee Pad Transformer tablet that lets readers download a 97MB file with the source code for v8.2.2.6 of the Android kernel.

  137. 4.5 months later the source still hasn’t been released. But surely ESR has looked into the heart of Google and seen that they’re on the up and up.

    If there’s nothing wrong with Google, there must be something wrong with the universe!

Leave a comment

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