The Economic Case Against the GPL

Is open-source development a more efficient system of software production than the closed-source system? I think the answer is probably “yes”, and that it follows the GNU GPL is probably doing us more harm than good.

I mean “efficiency” here in the precise sense economists use it. Of two systems of production, the more efficient is the one which produces more units of output for a given input of factors of production. Now, let’s divide up all possible worlds according to whether the answer to our question is “yes” or “no”.

In all worlds, markets seek efficiency, because investors are constantly seeking the best return on capital. Thus guarantees the most efficient system will win, eventually. The flip side of this is that markets will punish those who adopt the less efficient mode. They’ll be outcompeted. Capital will flow away from them.

If we live in “Type A” a universe where closed source is more efficient, markets will eventually punish people who take closed source code open. Markets will correspondingly reward people who take open source closed. In this kind of universe, open source is doomed; the GPL will be subverted or routed around by efficiency-seeking investors as surely as water flows downhill.

If we live in a “Type B” universe where open source is more efficient, markets will eventually punish people who take open source code closed. Markets will correspondingly reward people who take closed source open. In such a universe closed source is its own punishment; open source will capture ever-larger swathes of industry as investors chase efficiency gains.

In a Type A universe, reciprocal licensing is futile. In a Type B universe, reciprocal licensing is unnecessary. In neither universe can the GPL’s attempts to punish what we regard as misbehavior have more than short-term, temporary effects. At most it can speed or slow movement on the efficiency gradient, not reverse it.

For the GPL to actually determine the mode of software production, we would have to live in a universe where the difference in efficiency between open and closed-source development is so vanishingly close to zero that over typical project lifetimes it is less than the cost of an enforcement lawsuit. This seems as wildly unlikely as flipping a coin and having it land standing on an edge.

I think we live in a type B universe – that is, one in which the GPL is unnecessary rather than futile. Mind you, I am not claiming the GPL is entirely useless. It’s a signaling behavior, like wearing a crucifix or yarmulke or pentagram – it helps build trust groups. But it has costs, too — it creates a lot of needless fear from potential allies and users who suspect they won’t be able to control their exposure if they let it in.

This fear is only exacerbated when we actually sue to enforce it. It’s obvious to pretty much everyone in the open-source community that the RIAA is slitting its own throat by suing music downloaders, alienating future customers wholesale. It’s not obvious why the Software Freedom Law Center’s current lawsuit against Cisco is any smarter – we stand to lose not only Cisco as an ally, but any corporation that estimates (rightly or wrongly) that their own potential exposure to an SFLC lawsuit might be greater than their potential efficiency gains from open-sourcing.

So the correct question to ask is this: Is the GPL’s utility as a form of in-group signaling worth the degree to which fear and uncertainty about it slows down open-source adoption? Increasingly I think the answer is “no”.

The GPL may be a community-building signalling device, but it is also a confession of fear and weakness. To believe that it matters, you have to believe that you live in a Type A universe where closed-source development is such an attractive proposition that you have to punish people for trying to move to it.

So maybe an even more fundamental question to ask is this: Does the open-source community believe in itself, genuinely believe it has a more efficient system of production? And if it does, does it make sense to choose a license that implies the opposite?

147 thoughts on “The Economic Case Against the GPL

  1. I don’t know rms, but I have read some of his thoughts on this. It seems to me that the GPL is not an economic thing at all, but a moral thing. I don’t suppose rms really cares whether the GPL is efficient, what he cares about is that he can freely share his ideas; that being reward in itself. His concern is not that the GPL allows for better software development, his concern is that it is just simply wrong, in a moral sense, not to have software source code freely available, and freely modifiable.

    Of course, I hardly need to tell Eric Raymond this, given that you were the person who dragged the hacker community away from such self indulgent nonsense onto a more credible economic framework. However, my point is that there are more economies at work in the world that the trade economy that your article focuses on.

  2. The assumption in the model is that the circumstantial space is homogenous, and unaffected by the actions of agents within it. In fact the actions of other agents are a part of such a space, and affect the outcomes for subsequent actions.

    For instance, evolutionary theory now recognizes that there is no one optimal reproductive strategy – and that what is optimal for an individual may be dependent on the strategies of others. In particular, it seems that going against the strategy of others can be best. There are ‘free rider’ effects, for instance. There are also tensions between individual and social benefits, and between short and long term outcomes.

    I don’t know much about the dynamics of OSS development or the incorporation of OSS into proprietary software. But it does seem to me that having a barrier – even a weak and largely symbolic barrier – to the privatization of OSS could deter a fair amount of destructive behavior.

  3. Could you suggest a license that gives off the open-source friendly signals and doesn’t cause fear on either side? MIT? Apache?

    I can’t see that throwing it in the public domain would be the solution. That would just allow businesses to combine the benefits of a strong open-source community and corporate secrecy. Of course the open source movement would stagnate in that environment. But in the short run it would make sense for companies to take advantage while the value was there.

    I need to learn more about this issue.

  4. Eric, while your argument is interesting and, based on my limited understanding of economics, certainly has merit, you’re not going to change anyone’s mind, for one reason: you’re trying to counter religious arguments with logical ones. Those who believe the GPL is a Good Thing do not care what its economic benefits or hindrances are. They only care about Freedom!!! – as long as you don’t drag those dirty rotten developers into the picture.

  5. So just for kicks, let me run through all the possible permutations here:

    1.) [+(c)/+OS/+GPL] Patterns can be property, so copyright law is morally justified, but creators of content should certainly be allowed to choose the rights they will reserve and allow any degree of openness. A range of choices in this regard in a good thing (Lessig)

    2.) [+(c)/+OS/-GPL] Patterns can be property, so copyright law is morally justified, but open source is more economically efficient and completely open source is more efficient than GPL, thus making GPL bad, even if not the worse case. (Raymond)

    3.) [+(c)/-OS/-GPL] Patterns can be property, so copyright law is morally justified, and all open source, including GPL, is communist and evil. (Gates)

    4.) [-(c)/+OS/+GPL] Patterns can not be property, so copyright law is an immoral use of force, and closed source is bad. However, GPL is a legitimate use of force in “self defense” so it is good. (Stallman)

    5.) [-(c)/-OS/+GPL] Patterns can not be property, so copyright law is an immoral use of force, thus closed source is bad, and non GPL open source is also bad because it may aid the enemy who believes in closed source. (The Gnazi Party?)

    6.) [-(c)/+OS/-GPL] Patterns can not be property, so copyright law is an immoral use of force, making both closed source and GPL bad. (Hastings)

    There is probably someone specific famous person who publicly espouses position #5 and someone else more famous than me who publicly espouses position #6, but apparently neither are famous enough for me to know who they are. :-)

    There are also two more permutations that I can not immediately make sense of:

    7.) [-(c)/-OS/-GPL] ???? The Amish?

    8.) [+(c)/-OS/+GPL] ????

    @esr – You seem to believe in IP but come to a lot of the same conclusions I do based on ideas about market efficiency. It occurs to me that if you were to come to believe (as I do) that decisions that maximize market efficiency are equivalent to decisions of maximum moral correctness, that you might move from a +(c) to a -(c). Or do you already believe that, but think that the idea of treating a pattern as a property right can create greater economic efficiency in areas other than computer software?

  6. If I choose the GPL, I can make you pay me for permission to make a proprietary fork. If I choose a permissive license, I can’t make you pay up. Seems like in that case, the GPL is the economically superior license.

  7. 1. I am not sure who the this article targets. Its the developers choice which license he wants to work under. Whether its the GPL, BSD or any other variant. The question is not which license is more “efficient”, but which makes sense for the work the developer has done.

    2. Your picture of the free market is painted from an “ideal world”, in reality free markets as of 2009 have led us to an economic collapse, not an efficient outcome whichever way you cut it. Since the system has allowed the current credit crisis, it seems to me there is a systemic issue with free markets the way its currently setup.

    Both these arguments need to be better addressed.

    Cheers,
    Krishna

  8. Correction: That’s the financially superior license. HUGE difference.

  9. @esr: First of all, let me just say that I have real gratitude for you. “How to become a hacker” was the article that inspired me to want to become a hacker, though I’m not one yet, by any means.

    But I think you’re wrong about this, for the following reasons.

    If we did not have the GPL or refused to enforce it, nothing would deter an unscrupulous company from stealing open source code and not contributing their own improvements to it to the community. Any improvements to the software from the open source community would quickly be incorporated into the proprietary version, while any improvements made by the hypothetical unscrupulous company would remain proprietary. Once a few of these companies started enjoying success, every other profit-maximizing company would be forced by the market to follow suit or perish, and we’d eventually have a clusterfuck of companies stealing open source code while the open source community dwindled to nothing because their products could never possibly compete with the proprietary alternatives.

    I think we live in universe C, where, without an enforceable GPL, the most efficient method of software production would be to take what you can (from the open source community), and give nothing back. YARR!

  10. @Sean Hastings – Your list just made a lot of things very clear.

    @esr – As there is no such thing as a free market, greedy and corrupt monopolies will try and push the market toward maintaining their status-quo. You have to look no further than MS’ FUD for an example of this. Hence there was a need for the GPL in the beginning of the OS industry. (The status-quo being the “type A”, with the real world being the “type B”)

    I do think that you are correct in saying its usefulness is limited, however, and the main benefits from the GPL are no longer needed.

  11. > If I choose the GPL, I can make you pay me for permission to make a proprietary fork. If I choose a
    > permissive license, I can’t make you pay up. Seems like in that case, the GPL is the economically
    > superior license.

    Doesn’t the GPL technically forbid proprietary forks? Even if it doesn’t, it seems to me that one person getting payouts to make a proprietary fork is a violation of the spirit and intent of everything the Free Software stands for.

    Given the internet uproars that tend to happen when things like that occur, i’d expect a proprietary fork bribe would result in a fractured community, a stagnant project and destroyed reputations.

  12. Jon B,

    If I am the original copyright holder, I can do anything I want. If I can make some money by selling permission to ignore copyleft, that seems like a pretty damn good economic case *for* the GPL. Frankly, I have no idea what esr is talking about as he is indirectly suggesting that this possibility should be ignored.

  13. http://people.debian.org/~bap/dfsg-faq.html

    Q29 answers that. Basically, if you are the copyright holder, you are also allowed to release it under other licenses. So you can allow people to make proprietary versions if you want. Note, this only applies to your own code. For, say, the Linux kernel, it is impossible because there’s so many contributors and all of them hold copyright over their pieces.

    I do not believe the GPL is useless. Because it stops people from stealing control, which is pretty much why Microsoft still exists. Linux is a valuable product, and making sure that people aren’t creating proprietary forks is actually beneficial for all.

  14. 1) The privileges of copyright and patent are unethical.
    2) Intellectual work can be property just as much as material work, but this is delimited by natural privacy, not unnatural monopoly.
    3) The GPL is a close approximation to the neutralisation of copyright and patent (for covered work) and the restoration of the public’s liberty otherwise suspended.
    4) While people prefer their liberty, business models that rely upon its legal suspension and the submission of people to such law, are doomed despite considerable economic advantages of monopolies.

    The case is thus ultimately ethical, not economic.

  15. Hi esr,

    interesting post.

    The race about GPL isn’t against proprietary code. This one is already won, for evolutionistic reasons.
    The race about GPL is against BSD.

    Only if BSD was more efficient would we abandon GPL.

    So, we live in a type B universe, but GPL is useful, mainly against BSD.

    David

  16. Oh, and concerning your last question :
    “So maybe an even more fundamental question to ask is this: Does the open-source community believe in itself, genuinely believe it has a more efficient system of production? And if it does, does it make sense to choose a license that implies the opposite?”

    I, for one, believe we have a more efficient system of production. And I choose GPL to well signify that I’m ready to fight for this.

    It’s a signal of fight. That’s why concurrents are afraid.

  17. So you suggest me to put my code out there, let any who wish to take it, then close it up if *they *want & look at them making loads of money outta my code & look at worn out me every morning in my broken mirror? OK bye.

  18. I personally love the ISC license for its simplistically; just the bare minimum attribution and disclaimer against damages and stuff. However, my threshold for using this ends right when it comes to web software; if it sounds like a stupid reason, so be it, but I have been “burned” in the past over license loopholes and/or general assholes, that I can’t help but to use the GNU AGPL at the very least as a legal guard against people forking it and keeping their modifications secret (even under the regular GPL, it’s arguably possible to keep your code closed off, which is why AGPL was made).

  19. >It occurs to me that if you were to come to believe (as I do) that decisions that maximize market efficiency are equivalent to decisions of maximum moral correctness, that you might move from a +(c) to a -(c). Or do you already believe that, but think that the idea of treating a pattern as a property right can create greater economic efficiency in areas other than computer software?

    It’s a little trickier than that. I do in fact believe that correct ethics coincides with market efficiency; it has to, otherwise ethics would not be sustainable. I’m a “not sure” on the (c) question; I used to have the position you ascribe to me, and would still prefer to, but for the awkward problem that IP doesn’t seem to have Schelling points.

    I would still prefer to hold (+c), because I fear that in a (-c) world creative work would be drastically underproduced. So my thinking time on this topic is devoted to trying to find Schelling points on the enforcement gradient. I regard the question of whether such points exist as equivalent to the question of whether IP rights are necessarily a creature of government fiat.

  20. >Could you suggest a license that gives off the open-source friendly signals and doesn’t cause fear on either side? MIT? Apache?

    BSD is the classic choice.

  21. (BSD) BSD: What Apple (MacOS) did to FreeBSD
    (BSD) ArgoUML: What Gentleware (Poseidon for UML) did to ArgoUML

    (GPL) Linux: What RedHat did to Linux
    (GPL) MySQL: What Percona did to MySQL

    Which license you think is better for people in general? More number of people? More beneficial to end users? I think GPL is the clear winner here; the companies that build business on GPL software are not losers. More_permissive_than_GPL (BSD, Mozilla et al) licenses are only as safe as the probability of someone not ripping off their work.

  22. >Your picture of the free market is painted from an “ideal world”, in reality free markets as of 2009 have led us to an economic collapse, not an efficient outcome whichever way you cut it.

    Er…no. We got to what’s not actually an economic collapse (yet, anyway) through political failure, not market failure. Political causes include, for beginners, the inflation of a property-assets bubble through a policy of holding interest rates on fiat currencies artificially low.

    Politicians would love you to blame the free market for their mistakes; that way they can keep exploiting you by pretending to shield you from market failures. Don’t be suckered.

  23. I don’t use the GPL because of its viral nature. I want my software to be freely usable, whether or not the user wants his use of it to be freely usable. I’ve been using Apache, but I have no attachment to it. Somebody wanted to publish one of my libraries at drupal.org, and they have a GPL requirement, so I modified the license to allow the user to choose which license they publish it under, MPL, GPL, LGPL, or Apache, patterned after the Mozilla Public License ( http://www.mozilla.org/MPL/ ). I’ve actually been tempted to move to public domain, but again, that limits people who require GPL. And I like to require attribution, so I get some credit. The Creative Commons Attribution license is the best match I know about ( http://creativecommons.org/licenses/by/3.0/ ).

  24. I agree that the GPL should be irrelevant (or futile), however, I see it’s value as in providing a strong part of the incentive to move into a situation where it’s irrelevant.

    I don’t agree that it’s obvious to all involved that open source is the most efficient way to develop, not by a long shot. And that by working (short sightedly) against their own best interests, others will also critically damage the work that is trying to be done. I feel that the GPL provides a necessary legal safety net until utopia arrives.

    I also don’t agree with suing people over GPL problems unless such a suit is widely agreed to be strategically necessary.

  25. I’ve actually been tempted to move to public domain, but again, that limits people who require GPL.

    No it doesn’t; public domain means that anybody can do anything with the code.

  26. Imagine a “Type A” universe where closed source is always the most efficient system of software production and markets will always punish people who take closed source code open.

    Then imagine a “Type B” universe where open source is always the most efficient system of software production and markets will always punish people who take open source code closed.

    And then imagine a “Type C” universe where both open source and closed source exist together, and the most efficient way to make money in software industry is to base your closed source program on an existing non-GPL’ed open source program.

    The “Type A” and “Type B” universes obviously don’t exist, they are just figments of imagination. The reality is that we live in a “Type C” universe, and financially successful companies like Microsoft and Apple use open source software in their closed source programs.

  27. “The GPL may be a community-building signalling device, but it is also a confession of fear and weakness. To believe that it matters, you have to believe that you live in a Type A universe where closed-source development is such an attractive proposition that you have to punish people for trying to move to it.”

    Nonsense. Your fundamental “highlander” argument that there can be Only One True Ultimate licensing approach does not model the spectrum of human motives.
    Some motives target the wallet (proprietary), others the mind (open source), others the heart (free).
    Let the legal system treat them equally and de-conflict the arguments that arise.
    I submit that, in spending time arguing an economic case against the GPL, rather than cheerfully disagreeing and respecting it, you’re aiding the proprietary cause.
    Bill Gates and ilk love you, sir.

  28. >And then imagine a “Type C” universe where both open source and closed source exist together, and the most efficient way to make money in software industry is to base your closed source program on an existing non-GPL’ed open source program.

    There’s no difference between your supposed type C and type A.

  29. We live in a Type A universe. Monetizing the software product through secrecy rent is the most efficient way we’ve thought of so far to deliver excellent software that’s easy to use to the masses. I’ve been through this quite a bit before.

    People get into open source for a bunch of reasons — morality, personal satisfaction, ego boost. From an economic standpoint it doesn’t make sense. But the market is only one axis of life…

  30. >We live in a Type A universe. Monetizing the software product through secrecy rent is the most efficient way we’ve thought of so far to deliver excellent software that’s easy to use to the masses. I’ve been through this quite a bit before.

    Meanwhile, Firefox’s U.S. market share sustained upwards of 20% last month and is still rising, IE is down to 66% and falling, Microsoft’s sales revenue is dropping faster than IE’s market share, and frantic FUD and bribery by Microsoft have failed to stamp out Linux in the netbook market, where plenty of end users are getting along with it just fine. You’re setting yourself up for an “I told you so!” of epic proportions.

  31. esr: Marketshare == revenue?

    How DO firefox make their money? I genuinely don’t know.

    I’d LOVE there to be an epic “I told you so” :)

  32. There seems to me to be an assumption that the answer to A/B is “yes” or “no”, but that isn’t necessarily so, and quite possibly isn’t so in our reality. There could be plenty of states which are non-optimal but stable for long enough to be of interest.

    The right analogy might be the (classic) iterated prisoner’s dilemma game; in this reference, closed-source would be explotative, BSD cooperative, and GPL friendly but retaliatory (e.g., tit-for-tat). Interestingly, if I understand correctly, there is *no dominant strategy* for this game – no strategy, even once established, can hold the field against all comers – but the “friendly retaliatory” class of strategies wins tournaments with a high degree of regularity.

  33. >How DO firefox make their money? I genuinely don’t know.

    Probably mostly from Google. It’s in Google’s business interest to be Firefox’s default search engine, and they pay for the privilege,

    In any case, where Firefox gets their money is not very relevant to my argument. If we live in a type B universe, Firefox will attract either profits or Google-like subsidies because open-source development is the most efficient way to produce a browser; if we live in a type A universe, it won’t. The mix of revenue vs. subsidy is an interesting datum for other reasons, but doesn’t bear much on the efficiency question.

  34. >The right analogy might be the (classic) iterated prisoner’s dilemma game; in this reference, closed-source would be explotative, BSD cooperative, and GPL friendly but retaliatory (e.g., tit-for-tat). Interestingly, if I understand correctly, there is *no dominant strategy* for this game – no strategy, even once established, can hold the field against all comers – but the “friendly retaliatory” class of strategies wins tournaments with a high degree of regularity.

    Clever, but it misses my point that in the B-universe case defector behavior is self-punishing – that’s the premise. It isn’t, in the tournaments you’re talking about; that’s exactly why tit-for-tat is adaptively useful in the tournaments. Now go reread my original argument; there’s a level of it you missed.

  35. Eric, without constant influx of money from Google, the Mozilla Foundation would probably have folded long ago. Firefox may still exist, but be a much more niche product. Sun’s pledge to open source the world (or at least that part of it they still have control over) sure proved fruitful for them; currently they’re set to become an arm of Oracle.

    As for “stamping out Linux in the netbook market”, that really isn’t the goal; the goal, as is always the case for Microsoft, is control of the platform and hence the market; this they’ve achieved. According to Microsoft’s numbers upwards of 90% of netbooks are shipped with Windows. They’re pretty unreliable in this regard, but I’d lay long odds it’s more than half. Even before Windows XP was an option for netbooks, the number one user question on netbook help forums was “how do i get windows on this thing?” And I still can’t find a netbook for sale in stores with Linux on it.

  36. and frantic FUD and bribery by Microsoft have failed to stamp out Linux in the netbook market, where plenty of end users are getting along with it just fine.

    In December 2007, the netbook market was about 6M machines, and 80% Linux.

    In December 2008, the netbook market was 24M machines, and 96% Windows XP

    As of April 2009, the netbook market is about 33M machines, and 97% Windows XP.

    Source on this is Gartner.

    Which means that the peak number of Linux netbooks was in December of 2007 (approximately). Call it 5 million machines. It’s now about 1M machines.

    Which is an 80% reduction in total machines using Linux in this space. I don’t know about you, but I’d find an 80% reduction in installed base alarming.

    In my own circle of acquaintances and friends, there are a lot of people who bought Linux netbooks and then wiped them down and installed XP on them, largely because of what software they wanted to use on the platform. It’s roughly 15% Linux users in the netbook group I’m part of. The driving force is “I use Windows everywhere else, so there’s less weird crap if I standardize.” This, of course, causes the Linuxistas to sputter and begin the sales pitch.

    I would guess that two thirds of these folks have now tried Linux and switched back.

    Microsoft makes about $15 per copy of XP sold in netbooks, according to their latest earnings call. Which runs to $15 * ~18M = $270M per year. That’s (more or less) found money for them, since their real XP revenue stream at this point is service contracts for the businesses that remain on XP. At some point the netbook market is going to saturate; whether it does so before there’s a new hardware revolution in the space is a matter of great speculation.

    Does any Open Source OS development/maintenance team have a quarter of a billion in annual cash flow? I honestly don’t know. I also doubt that the cash flow they’re getting now is sustainable. I’m guessing saturation hits in two years.

    The consumer price delta between a Microsoft and Linux netbook is averaging about $70 at the consumer level. Linux machines are typically $50-70 cheaper than XP ones in this space – which comes out to about 1/6 to 1/5 of the total purchase price.

    Which means that, given a clear choice in the market, people are willing to pay 18-20% more for a netbook with an obsolete version of Windows rather than use Linux. As mentioned, in the netbook user group I’m in, roughly 2/3 of the people have tried Linux and switched back.

    As to future trends:

    There’s going to be a lot of pressure on Microsoft from vendors to keep the XP license going rather than sell Windows 7 Starter Edition for about the same price. There’s going to be a lot of pressure from Microsoft to put in Windows 7 Starter Edition, which allows the user to input a credit card number and unlock additional, pre-installed features.

    Microsoft has demonstrated that the chainsaw they took to Vista to make Windows 7 has substantially reduced hardware requirements and boosted performance – Windows 7 RC1 runs fine on netbooks with all features enabled, even with 1GB of RAM; the power management in RC1 isn’t as good as it is in XP, and there are missing drivers, but those should both be fixed by this summer.

    I think Microsoft will keep moving the XP non-delivery date back in 6 month intervals (they’ve done it twice already), offer both, watch how people respond to Starter Edition, adjust what goes into Starter Edition, and fine tune the ‘try it for free and upgrade’ message. I think that if they shift the ‘locked’ versus ‘unlocked’ features slider a little bit, it might work.

    There will, of course, be a thriving market in utilities to unlock the features in Starter Edition, followed by a malware tsunami bundled to those utilities. Microsoft will capitalize on this, as the malware people will do more for the message of “get it legit” than anything Redmond could do with an ad campaign.

  37. Shantanu Kumar refers to
    > (BSD) BSD: What Apple (MacOS) did to FreeBSD

    And he does so in a context which makes it clear he regards this as a bad thing.

    Not all of us do.

    Yes, large parts of the interesting bits of OS X are closed-source. But it’s still done way more to put a good unix on the desktops of average users than all the Linux coders of the world put together, and in the process it’s contributed a lot of invisible-to-Aunt-Tillie-but-nevertheless-important code back. I call that a win.

    If FreeBSD had been licensed under GPL, Apple couldn’t have used it as the basis for Darwin and thus OS X, and all those Mac users would not now have a high-quality operating system with the best tools available.

    I will absolutely defend the right of individual developers or development groups to choose GPL on their own. Their code, their rules. But that’s not the same thing as saying it’s either the most ethical or the most efficient system on the macro scale. (They don’t quite mean the same thing, but I’m with Eric in asserting that they’re very tightly correlated.)

  38. >Source on this is Gartner.

    See, that’s a problem right there. Historically, Gartner has been deep in Microsoft’s pocket on stuff like this. Those figures are what Microsoft would like you to believe; I’m not convinced they have any but the most tenuous relationship with reality.

    The 15% among your peer group is closer to credible, but certainly doesn’t justify gloom and doom about Linux’s situation. Heck, that’s already larger market share than the Mac has over a broader category.

    Then there is the irreducible fact that, by its own report, Microsoft’s sales and revenues are dropping. This backs up my theory, published here previously, that the netbook makers have used the Linux threat to screw M$ to the wall and it’s actually making effectively no net profit on those netbook sales after hidden subsidies like “marketing support” are figured in.

  39. Ah… I see. Unlike the normal IPD, the C/C (both cooperate) payout is higher than the D/C (I defect, you cooperate) payout, and thus cooperative strategy may well be dominant.

    I still think, though, that you may be underestimating network effects and the likelihood of long-term stable non-optimal states. XP survives because the perceived switching cost is higher than the benefits of switching, not because it’s better. A program’s most important asset is its userbase, not it’s code. The continued existance of XP, despite the best efforts of the OSS community, Apple, and Microsoft (!) would seem to pretty well prove the issue.

    You may be right about GPL/BSD, though; the only issue I can think of offhand is that GPL programs seem much less likely to fork – I’m not sure if that’s a consequence of the license or cultural.

  40. I don’t see the prisoner dilemma (that’s probably me), but I do see a tragedy of the commons type possibility – where the overall system optimal is for people to develope in the open model, but individuals can get a competatitve advantage by exploiting the situation for their own short term gains at the cost of everyone else (which would be the normal situation now).
    Again, GPL in this case provides a mechanism to guard against this exploitation, providing an individual punishment for going against the overall environment where there wouldn’t be one otherwise.

    Is that world the same as A, B or the supposed C?

  41. @esr: I think you missed the point on the Universe C argument. In Universe C a company can take their code open source without being punished. It’s a world where some people see the benefit of proprietary software and some people see the benefits of open source. It’s a world where choice exists and no single choice is right for everyone. You painted two Universes as either black or white. The real world is shades of grey. Where open and closed source both exist for a reason and are both successful for a reason.

    “Er…no. We got to what’s not actually an economic collapse (yet, anyway) through political failure, not market failure. Political causes include, for beginners, the inflation of a property-assets bubble through a policy of holding interest rates on fiat currencies artificially low.”

    So your telling us that giving out of work pizza delivery guys $500,000 dollar mortgages, with ARMs flying every where didn’t get us here. It wasn’t greed that said let’s not regulate the market because it will regulate itself. And then oh my, would you look at all these bad mortgages that we have to eat now… It wasn’t greed that made people redevelop instruments that had no oversight and thus no controls on it? It wasn’t companies like AIG losing trillions in those instruments? It wasn’t because of people like Madoff and his billions of dollars. It wasn’t a matter of holding the interest rates low that completely fueled the bubble, it added fuel but there’s much more to it than that.

    Yes the politicians are partly responsible, but they by no means hold all the blame. They do what we allow them to do, if we don’t stop them they think everything is kosher.

    “Politicians would love you to blame the free market for their mistakes; that way they can keep exploiting you by pretending to shield you from market failures. Don’t be suckered.”

    Ummmm, do you have your tinfoil hat on? The government is too disorganized to be malicious on purpose. It’s malicious because of it’s inefficiencies and the bureaucracy.

    “Probably mostly from Google. It’s in Google’s business interest to be Firefox’s default search engine, and they pay for the privilege,

    In any case, where Firefox gets their money is not very relevant to my argument. If we live in a type B universe, Firefox will attract either profits or Google-like subsidies because open-source development is the most efficient way to produce a browser; if we live in a type A universe, it won’t. The mix of revenue vs. subsidy is an interesting datum for other reasons, but doesn’t bear much on the efficiency question.”

    But it is important where they get their money. It matters because where and how the money flows in is just as important to efficiency as what type of license your product has. You say that it doesn’t bear much but don’t back it up.

    The reason that where and how the money flows in from is important is, as businesses mature and grow, their revenue streams need to stay stable, and in a free market environment if you don’t have a steady solid source of revenue your business fails to get new investors.

    Mozilla is a good example, when Netscape morphed into Firefox there wasn’t business model for it established, they generated a good product, but now it’s plagued with bloat (it’s using ~127 mb of memory as compared to IE7 with the exact same windows open coming in at a whoping ~27 mb), this might have partly played a part in Google’s development of Chrome. Now what happens if/when Google decides to stop funding a competing browser? Where’s Mozilla going to get the majority of their funding? Who’s going to continue to support you after losing your largest source of funding? What type of proposal are you going to write? How are you going to sell your product and your self after the above scenario? Once you’ve begun to lose funding, what are you going to cut, or if you get new funding what are you going to do with it. How does the funding play a part in the efficiency of your organization? When one source dries up does that change the game? Where your funding comes from is incredibly important.

    I think some of your arguments are too vacuous to be useful or true, but your general analysis of the dangers of suing your customers and scaring off the rest of society are definitely accurate.

  42. You address efficiency, but I guess I have another question: Which is more likely to produce what end-users need or want? I think it really all depends on the user.

    You dig into the guts of any system you work on, if I’m guessing right. My mom-in-law wants to check email, keep up with her sisters/kids and other equivalent tasks. To her the computer is a black box that is full of deep magic that she doesn’t understand, but she’s happy to use it, regardless. I suspect that many users out there are like that. To her it’s a closed system; open source would be meaningless from her perspective, and I think also from the perspective of a majority of users out there. They are not interested in tinkering with the guts; they just want nifty, kool features to play with.

    So which is better, open or closed? if you are at the ‘guts’ level, open, if you are at the user-interface level, why not a closed source model, where software interface houses produce front ends for people like my mom-in-law.

    To put it another way, this allows for UI designers to have tools they can rely on and not have to design themselves, while the underlying components they rely on are built in an open-source fashion. The back end people pool resources in building the tools. The front end people are more agile and responsive to market-driven pressures(making what people want), allowing faster turn around. This makes it easier to compete against the Microsofts and Apples of the world. (Faster OODA Loops, hehe)

    First time responding here…sorry if I fail at any conventions or customs of which I am not aware.

  43. >Ah… I see. Unlike the normal IPD, the C/C (both cooperate) payout is higher than the D/C (I defect, you cooperate) payout, and thus cooperative strategy may well be dominant.

    Exactly so.

    >I still think, though, that you may be underestimating network effects and the likelihood of long-term stable non-optimal states. XP survives because the perceived switching cost is higher than the benefits of switching, not because it’s better. A program’s most important asset is its userbase, not it’s code.

    Quite right. But I showed ten years ago, using the victory of TCP/IP over proprietary networking as a paradigmatic example, that first-mover monopolies are unstable because the incentive on the monopolist always favors squeezing the customer harder, especially as the monopolist faces competition from its own old product versions. Eventually, the portion of the rent the monopolist can collect from a customer always grows to exceed his transition costs out of the monopolist’s jail, at which point *boom*.

  44. The 15% among your peer group is closer to credible, but certainly doesn’t justify gloom and doom about Linux’s situation. Heck, that’s already larger market share than the Mac has over a broader category.

    Not really; this group self selects for “Willing to tinker with their system”, which most Windows (and fewer Mac) users are willing to do. It’s the 2/3 who have tried Linux and moved back (and the number of meetings where we have guys in the back of the room walking people through “scrub Linux, install Windows”) that I consider the true talking point.

    There are two people I know of who installed Linux and removed Windows. There are about 45 who did the reverse.

    Then there is the irreducible fact that, by its own report, Microsoft’s sales and revenues are dropping.

    A closer read on their earnings call shows that there’s a big slowdown on Vista adoption, and that XP revenues dropped by about 10%. The big hit is in the games division. We’re leading up to releases of Windows 7 and Office 14; this is going to put a damper on sales.

    And, of course, there’s the economic credit crunch, which, when factored in across the economy as a whole, Microsoft is STILL beating.

    This backs up my theory, published here previously, that the netbook makers have used the Linux threat to screw M$ to the wall and it’s actually making effectively no net profit on those netbook sales after hidden subsidies like “marketing support” are figured in.

    Both the founders of ASUS and Acer say that the only systems they can sell XP on are netbooks per Microsoft’s marketing contracts. Microsoft specifies some hardware specs (limits on hard drive size, RAM) that they go out of their way to show customers how to get around.

    I’m guessing that the ‘marketing support’ Microsoft gives this space is pretty minimal. The ability to say “Comes with XP Home preinstalled” and a sticker. Microsoft is preparing for the day when this space is saturated, and they can move it to a newer version of Windows. It’s a free rider benefit.

    In some ways, Microsoft benefits if Linux takes about 10-12% of the netbook market among competing distros. They can go to the EU and say “Look, we’re competing against people who are giving their stuff away for free; no monopoly abuse here!”

    Also, like Jeff Read, I can only find Windows netbooks in stores to try out. I joined the netbook users group in the hopes of trying out a few of them to see which ones had keyboards I hated the least and usable screens for what I do.

  45. Well, I, for one, applaud your challenging of assumptions. There is little doubt that the various incarnations of the GPL serve a useful purpose, as have and do the BSD licenses. It is, however, quite valid to examine with dispassion this instrument of license and its consequences.

    For some the GPL has become like the Declaration of Independence, enshrined and revered as timeless in its perfection. History will more likely treat it as an antecedent in an evolutionary process, rather than an ever-enduring monument of fulfillment.

  46. >So which is better, open or closed? if you are at the ‘guts’ level, open, if you are at the user-interface level, why not a closed source model, where software interface houses produce front ends for people like my mom-in-law.

    Well, in my mom’s case, Windows would have cost money. Linux didn’t.

  47. The free market is a framework for organizing productive human labour. I’m not sure that you can apply this framework to intellectual creation, such as computer programs, because of the ease of replicating information that doesn’t occur with goods or services (some services (business administration, musical performance) produce information, but what can’t be replicated here is the time and energy of the worker). Rather than using the idea of “efficiency” which may or may not apply in this case, you need to look underneath the idea and see if the mechanisms still work in the case of computer programs (and considering the arguments about the prisoner’s dilemma or the tragedy of the commons as other commenters above have mentioned).

    The point of the GPL being coercive is to fight against other people being coercive. I’m sure that many people who support the free software movement would love it if copyright laws were abolished, so that there was no need for the GPL. It’s the restriction of liberty that makes the actions of the RIAA so contemptible. It seems like sophistry to complain that the FSF is trying to restrict the liberty of Cisco, when the only respect in which it is trying to restrict liberty is to stop Cisco restricting others with the use of copyright laws.

  48. Matt Says:
    > If FreeBSD had been licensed under GPL, Apple couldn’t have used it as the basis for Darwin and thus OS X, and all those Mac users would not now have a high-quality operating system with the best tools available.

    Where is the economic case here that favours FreeBSD? Does Matt’s statement imply any Non-GPL license is nothing but economically a glorified “ThrowAway” license? Even though both GPL and Non-GPL software makers can earn from support, it seems Non-GPL software are prone to rip-off, while GPL is not. Even though GPL software can be encumbered (e.g. RedHat Enterprise Linux) using trademark and stuff, but then it forces something like CentOS to be eventually possible.

    @ESR: Notwithstanding my high regard for you as an Open Source subject matter expert, I feel a post on “Economic case against Non-GPL” would be rather interesting, especially as a comparative study with empirical data.

  49. >The free market is a framework for organizing productive human labour.

    No, the free market is what you have when nobody is using coercion. It includes a monetized part, which is what you’re thinking about (trade of goods and services mediated by cash or cash equivalents). It also includes a non-monetized part that includes all voluntary cooperation – including, in particular, all open source cooperation.

    Almost all creative work takes place within the free market. There are exceptions, but they’re pretty unusual.

  50. Also, like Jeff Read, I can only find Windows netbooks in stores to try out. I joined the netbook users group in the hopes of trying out a few of them to see which ones had keyboards I hated the least and usable screens for what I do.

    I find the MSI Wind U120 to be the best balance of ergonomics and price in this regard. The keyboard is a bit larger than standard netbooks, and the touchpad isn’t infuriating.

    It also appears to be marketed for Windows use only, as its wireless device lacks a kernel driver, save for an unsupported experimental module you can download and compile yourself.

  51. >I don’t see the prisoner dilemma (that’s probably me), but I do see a tragedy of the commons type possibility – where the overall system optimal is for people to develope in the open model, but individuals can get a competatitve advantage by exploiting the situation for their own short term gains at the cost of everyone else (which would be the normal situation now).

    Is it? It appears to me that the “competitive advantage” is largely an illusion. Because when you take a project closed, the open-source parent doesn’t go away, and you end up on the wrong side of a few-developers against-many competition.

    Attempt this thought experiment: Try to come up with a sound business case for a closed fork of, say, Apache or Firefox or Linux. Now imagine what you’d say to a VC that asks you what market you think you have, and how you plan to stay ahead of the parent project.

  52. @Jeff Read

    So far, the best netbook keyboard I’ve gotten to play with is the Samsung NC10; the Samsung N120 is an NC10 with a larger keyboard (pulled from the NC20 12″ not-quite-a-netbook.) that I haven’t gotten a chance to play with yet.

    I’m deeply interested in the NC20; I want to actually work on one for an hour or two before plunking down the $550 for it. Main selling points: 98% full size laptop keyboard (which will still feel cramped) and a 1280×800 screen. Runs the VIA Nano CPU, has user accessable hard drive and RAM.

    Yes, it comes with Windows, which for me is a feature, since most of what I need to do on the road is print and use Excel. (I’ve got spreadsheets that beat OpenOffice Calc into bloody mush.)

  53. There is a class of so-called “open-source development” where the license may be totally OS, and the community may in fact be totally open to contributed work, but where pragmatically, the vast bulk of the work is done by people on a company pay-roll. I’m not talking about nefarious conspiracy, here; there may be many reasons why things work out this way, but sometimes that’s just the way things are. Matt Asay has often described Alfresco this way, and in his view born of this experience, the open-source license primarily serves to drive adoption, not contribution: free-as-in-beer access, the right to read the code as documentation, and perhaps the possibility of contributing actual patches or features.

    Personally, I’d rather call this sort of work “glass-house development,” just so we can distinguish it from contribution-based open source. But whatever we call it, in such a situation, the “in-group signaling” aspect of the license may have substantial financial value. 

    Economic efficiency seems, in practice, to include things like “cool,” “what everyone else is doing,” and the like: fashion sense. And fashion sense may have no more to do with financial sense in the software development arena than it does in, say, sports cars or blue jeans.

  54. I think Linux would not be nearly as successful with the BSD licence rather than the GPL – they have real copyright laws to make people play nice, and have successfully used them in this manner.

    (And Matthew Garrett DMCAing the MPAA was just beautiful.)

    Although the vitality of a project has much more to do with the vitality of the project. X.org is held together by being the place to be for a current X server, not by copyleft. The BSDs aren’t lacking participation.

    What annoys me is free software advocates who defend the existence of copyright itself on the grounds that it’s the basis of the GPL. Er, no, software freedom is the basis of the GPL. I feel they’ve somewhat missed the point.

  55. @Ken Burnside – the best netbook keyboard and form factor I’ve played with yet is the Sony Vaio Not-A-Netbook – near-fullsize keyboard, wide 1600×768 display, hideous price tag. OH GOD I WANT A DEVICE OF THIS FORM FACTOR.

    (Preferably not made by Sony. Vaio hardware makes Dell laptops look robust. I’ve seen the battered wrecks my NT admin colleagues have taken back from dung-flinging execs to get fixed. NEVER BUY A VAIO. I’ve heard the hardware quality has increased in the past few years, but I’m still certainly not putting my own money toward it.)

  56. A further thought: the world is not, of course, made of Platonic ideal rational economic actors. In practice, free cultural exchange (fanfic, Photoshop manipulators, etc) seems to be on terms similar to CC by-nc-sa – you are socially expected to give credit, you are socially expected to allow further remixing, you are socially expected not to charge for it. Danny O’Brien writes of such social perceptions here: http://www.oblomovka.com/wp/2008/08/07/copyright-fraud-and-window-taxes-no-not-that-windows/ I noted it here: http://rocknerd.co.uk/2008/08/09/just-spell-my-name-right/

  57. If I understand the post correctly, esr is saying that the market will adopt the system with global maximum utility (the best of all possible systems). Intervening parts of the state space with low utility are nothing more than “bumps in the road”: they won’t effect the final outcome because the market will always find a way to traverse them.

    If we construe ‘market’ as broadly as possible, I agree, but I think that the GPL is part of that.

    The (almost) totally closed source software ecosystem was at least meta-stable. It is locally attractive. With almost no open source code to work with, and no established community, it took heroic hacking and sustained massive effort to get to where we are today.

    I believe that the (almost) totally open source software ecosystem would be stable. This is the ground state for a “Type B” universe, and a place where the GPL is probably irrelevant.

    It is the intermediate part of the space, where we are now, interests me. There is a mountain range of unstable compromise and conflict between the two pure extremes. In various ways, the GPL provides a lower pass through the mountains.

    If a small/weak/young OSS project gets competition from a closed fork with a strong backer, the OS community could die off (does anyone have any examples? I don’t). The GPL could thus prevent a project from being closed up when it otherwise would be. It is also a publicity tool, beyond a mere badge, that people notice and think about in part because it may impact them legaly. Further, it gives some of the less enlightened participants in the market a compelling reason to contribute openly now, regardless of the eventual merit of OSS vs CSS methods.

    As the network effects of increasing OSS adoption create an environment that is increasingly more favourable to OSS than CSS, the GPL will become less important, but is has been important and it is important in at least some areas even now.

    If the Linux Kernel went BSD tomorrow, I think we’d see a series of closed forks with support for X, Y, or Z feature or device, that each withered over time, doing little lasting harm, but wasting time and effort. Not a huge problem, but not desirable.

    If EMC2 (an excellent Linux based CNC controller) went BSD tomorrow, I think there’s a non trivial chance that a savvy commercial operator could pull 90% of it’s users away with a closed fork, which would have a severe chilling effect on the project. (The risk is less now than it would have been two years ago). Probably not fatal in the long run, but rather harmful.

    Yes, the fear of GPL related legal problems is slowing down OSS adoption, but I we get more in return than the post gives credit for.

  58. I think the type ‘C’ discussion is important, as it deals with the false dichotomy presented in your post. There is an ‘in all cases’ clause implied for your comparisons of efficiency. There is no particular reason to suspect that either one will be better all of the time.

    Let P = proprietary, O = Open, E(x) = percentage of time where x is optimal. As presented, type A is when E(P) = 100, type B is when E(0)=100. I suspect that E(P) > 0, E(O) > 0 and E(P) != E(O). It doesn’t seem particularly troublesome to believe that E(O) = E(GPL) + E(O-GPL) and that E(GPL) > 0.

  59. I agree that a pure Type-B universe would be the most efficient way to create software in a world where belief was a binary choice. Complete belief vs. total disbelief. In the real world, that just doesn’t exist. I actually know a few open source developers that don’t believe it is the best way to build software but as long as they get paid, it doesn’t matter to them. Many of the investors who are funding open source projects are gambling on open source which is different than believing. I argue that at this point in time, it is more efficient to have something like the GPL get more “non believers” to participate in open source, which is why the GPL is so popular.

    So – the easy and obvious answer to the “… fundamental question … Does the open-source community believe in itself …” is: No. Remember, the community is a lot more than just the developers. Will they ever? I’m sure the majority of participants will eventually, but never 100%. The key word in the who gets punished argument is “eventually.” So even if the market “decides” that Type-B is best in the end, that doesn’t mean it’s best now. The current “best” will evolve over time. I think the GPL is less a signal to others in an effort to build trust groups, and more a sign of what a significant number of the open source community’s trust level is. It’s like having between 60% and 80% of the people in the world (depending on the survey) claim to believe in religions based on peace and cooperation yet not having a world that looks remotely peaceful or cooperative.

    Doug

  60. Is it? It appears to me that the “competitive advantage” is largely an illusion. Because when you take a project closed, the open-source parent doesn’t go away, and you end up on the wrong side of a few-developers against-many competition.

    Counterexample: Wine, the closed-source forks of which are infinitely better and easier to use than the open source main project.

  61. This blog entry relays a report from Dell that about a third of the customers for Dell’s Inspiron 9 netbook are ordering Linux and their rate of return is no higher than for XP.

    That seems more plausible to me than Gartner’s MS-funded 3% figure, or Ken’s peer-group estimate of 15%.

  62. I hadn’t noticed serious problems with the free Wine, but I’m not the typical use case… are there actually multiple closed forks now? (And are those companies selling code, or just support, when you look at the fine print?)

    On an amusing side note, I noticed the other day that a major Macintosh game (Spore) is in fact just the PC version running under a proprietary version of Wine…!

  63. ESR:
    >Try to come up with a sound business case for a closed fork of, say, Apache or Firefox or Linux.

    I suspect that there isn’t a good one for OSS on that scale. But my point about EMC2 was to say that there may be one for smaller weaker projects, such as Wine.

    Even if the cases merely look good while being bad, some people are stupid. It’s better for everyone if dumb things are obviously dumb, and so almost never get tried. That way, users don’t get frustrated, investors don’t loose their money, and people don’t have to swim home when the ship sinks. We get to spend the opportunity cost on implementing better ideas (such as contributing to OSS :-).

  64. >But my point about EMC2 was to say that there may be one for smaller weaker projects, such as Wine.

    Yup, it’s theoretically possible. On the other hand, I don’t know of anyone ever successfully pulling this off, and every attempt I actually know about has failed. I think this is one of those maneuvers that looks easier and more lucrative than it actually is.

    >It’s better for everyone if dumb things are obviously dumb, and so almost never get tried.

    Agreed. This is why I worked so hard at spreading the meme that closed-source development is obviously dumb (as opposed to, say, evil and immoral). I think I did a pretty effective job – and my way has the advantage that, unlike the GPL, it doesn’t scare corporate general counsels into not wanting to have anything to do with us.

  65. Rats are an incredibly successful life form. This makes them so much better fit for survival that pigs, green frogs and even humans are doomed to go away. This is the analogy of the reasoning in esr’s blog post. Fortunately the world is more complex than his simplified Darwinian model implies. There are ecological niches for software, just as there are niches for various life forms. Picking a license affects the survivability of a software project in the niche it targets. In some cases the natural selection has already picked the winner based on which license the project was using. For some projects this has been GPL, for some it has been a BSD style license and for some it may have been a proprietary license (or even mixed licensing models). The relative success of Linux compared to its BSD competitors may partly be due to the choice of licenses. If you listen to the BSD crowd, Linux has fewer device drivers and inferior code quality. That leaves, luck, better project management and a more suitable license as explanations for the success.

  66. Of course, the Stallmanites don’t care about scaring corporate general counsels – since they think, by and large, that corporations are nasty eeeeevil things that should be stamped out.

  67. Jacob, you’re omitting what I believe is the true cause for Linux’s continued success over the BSDs: marketing. The Stallmanites and other GPL zealots have succeeded in denigrating BSD over licensing issues to the point that only the cognoscenti pick BSD. Oh, and those poor, deluded Mac users…

  68. > Yup, it’s theoretically possible. On the other hand, I don’t know of anyone ever successfully pulling this
    > off, and every attempt I actually know about has failed. I think this is one of those maneuvers that looks
    > easier and more lucrative than it actually is.

    A few years ago I was using Cedega (one of the wine closed forks). While i wouldn’t call it a failure, it was at best a draw. In the end, for the uses i wanted it for at the time (gaming on linux) i needed both the current stable version of wine and the current version of cedega. Cedega had a shiny interface and (more importantly) better directx support. Wine had better support for everything other than directx. at a rough guesstimate i’d say i used Cedega 75% of the time (the interface means that ties in support go to Cedega) and wine the rest of the time.

    Note that this is based on experience from a good 3-4 years ago. Riffling through the forums after writing the above i found threads that allege that cedega has fallen behind versus wine (good source is http://www.cedega.com/forums/viewtopic.php?t=9255 however i saw a couple of threads with complains. The poll itself is immaterial, too small a sample space). So assuming this report is universal and accurate (forums being a soapbox for the vocal minority) i’d say cedega is ultimately a case in ESR’s favor. Closing the source lead to short term leads and long term stagnation.

    Economically, the question is if you were a a company looking for cross-platform support would you have gotten better bang for your buck by going with Wine or going with the closed source cedega?

    Amusingly enough I saw some accusations that the problem was that transgaming (the company behind cedega) was spending all it’s time helping EA with porting games to the mac. So from the perspective of transgaming i’d say economically they did pretty well out of the whole deal. Not so much for their customers though.

  69. Didn’t Microsoft make use of a lot of BSD licensed networking code? That must have created a lot of embrace-and-extend risk.

    >In all worlds, markets seek efficiency, because investors are constantly seeking the best return on capital. Thus guarantees the most efficient system will win, eventually. The flip side of this is that markets will punish those who adopt the less efficient mode. They’ll be outcompeted. Capital will flow away from them.

    That “all” is a very strong claim. I’ve seen straightforward economic models* where more efficient (in the long run) firms are outcompeted by less efficient firms (the market can stay irrational longer than you can stay solvent).

    My own mental model of this includes an “activation energy” barrier; in this case the GPL was necessary to move us from a “Type A” universe to a “Type B” universe.

    (* I think it was Daron Acemoglu’s “Good Jobs versus Bad Jobs”.)

  70. >Jacob, you’re omitting what I believe is the true cause for Linux’s continued success over the BSDs: marketing.

    I don’t agree.

    I think it’s the relative size of the developer communities that makes the big difference. And that, in turn, traces to the BSD culture – much more hierarchical and rigid; fewer places for casual hackers to fit in.

  71. There exists a type “C” world (the real world). In that world, the benefits to society as a whole are larger than the benefits that can be captured by a company from Open Source. However, the benefits to each company of owning the code are greater than 0.

    In this world, society as a whole could get – say – a trillion $ in value from Windows and OSX being free. However, Microsoft and Apple would each make nearly nothing compared to what they do today. Their investors and they would prefer to own it, only generate $100 billion in value but have that value flow to them rather than making much less.

    In this world, however, you don’t want to pass a law just expropriating all software as free & public because you don’t want to reduce the incentive for future inventors or cut the incentive future inventions from MSFT and Apple and others.

    As a result, in this world – you create a special license – a Public License, if you will, where coders who don’t want their code to be owned and are OK supporting open source allow their code to be used in other projects run similarly. Maybe this license is administered by a large producer of open source software. Perhaps you name this foundation after an African mammal.

    I would submit to you that this type C world happens to be a lot closer to the real one than either of the strawmen you propose.

    (For the economists among you, there are great externalities not priced in to the decisions he’s talking about)

  72. >There exists a type “C” world (the real world). In that world, the benefits to society as a whole are larger than the benefits that can be captured by a company from Open Source. However, the benefits to each company of owning the code are greater than 0.

    You seem confused. In distinguishing type A from type B universes I was just talking about relative production efficiency; all this stuff about positive externalities would be interesting if I had been talking about other things.

  73. I don’t buy it.

    “In all worlds, markets seek efficiency, because investors are constantly seeking the best return on capital.”, has been proven false several times over. Humans (investors) are not rational in their choices. Only in Econ 101 courses is this true.

    Check out the book “Predictable Irrational” by Dan Ariely for a bunch of easy to understand refutations to the idea that markets make the most efficient choices. I believe the name of the area of study is behavioral economics.

    You can search for his web site and videos for some hint as to what is in the book, but I definitely recommend buying the book.

    Ciao!

  74. I just don’t (for one example) see that Microsoft was punished economically for what they did with Kerberos. I think it probably helped sell AD tremendously. But maybe the economic theories do not take into consideration that type of monopolist’s actions?

  75. >“In all worlds, markets seek efficiency, because investors are constantly seeking the best return on capital.”, has been proven false several times over. Humans (investors) are not rational in their choices.

    You’ve misunderstood what the behavioral economists are trying to tell you. No blame; most people who have read popular accounts do.

    Yes, human beings are predictably irrational in a number of different ways. They still seek to maximnize ROI; they’re just not necessarily very good at it as individuals. Fortunately for economists, the market rewards behavior that is operationally indistinguishable from being good at it, so over sufficiently long timescales behavior approximates classical rationality pretty well.

    The flip side of this (which is the cutting edge of behavioral-economics research just now) is that phenomena like risk-reward asymmetry and the endowment effect seem to be more strongly elicited by simple laboratory-style experimental games than they are in real-world behavior.

  76. >I just don’t (for one example) see that Microsoft was punished economically for what they did with Kerberos. I think it probably helped sell AD tremendously. But maybe the economic theories do not take into consideration that type of monopolist’s actions?

    They do, but not in a way that helps address my original question. Monopolists can evade the consequences of rent-seeking and production inefficiency for a while, but the bill does eventually come due – as appears to be happening with Microsoft now.

  77. >You seem confused. In distinguishing type A from type B universes I was just talking about relative production efficiency; all this stuff about positive externalities would be interesting if I had been talking about other things.

    1) Type may be contingent on available software: the universe might have flipped from type A to type B when, for example, gcc became available.

    2) There might be a mixed type C world, say one where open source was more efficient for operating systems and closed source more efficient for applications.

    3) You were “talking about other things”. You made the assumption that closed/open source would be more successful in a type A/B world. Externalities might mean that a closed source business model succeeds despite the world being type B.

  78. Counterexample: Wine, the closed-source forks of which are infinitely better and easier to use than the open source main project.

    This is a very bad example. One of the closed source forks, Cedega, is known to be horribly outdated and even worse compatibility-wise than the official Wine. The other, CrossOver, works fine for the short list of supported applications but completely falls apart once you try unsupported applications. Plus CrossOver does in fact contribute code changes back to the free Wine about 6 months after their proprietary version of it has gone public.

  79. To rephrase one of ESR’s points: GPL is based in part on the fear that open source projects will get taken over and pushed aside by proprietary alternatives. But that’s simply not the case: Apache, Mozilla, Perl, PHP, Python, etc. How much more evidence is needed?

    Many who argue for the GPL miss that the market is self-correcting. If companies don’t add value, they don’t get paid.

    The GPL view is narrow: “I’m only going to give you my code if you pay me by giving me yours”. The BSD (etc.) approach is much more open: “I want there to be more good software in the world, even if I only benefit indirectly”.

    Apple’s use of BSD clearly increases the amount of good software in the world. No matter how many proprietary extensions Apple has added, developers are still better off than a fully proprietary system. Evidence: look at how many open source developers have MacBooks rather than Linux laptops.

    While I’m at it: copyleft abuses the word “freedom”. It specifically takes away a freedom that some people consider to be quite important: deciding how to get paid for the fruits of their labor. (IMHO the LGPL is a reasonable compromise, but of course looked down on by copyleft purists.)

    ESR looked at the economics, but I think the moral high ground is much more clear. GPL code can include BSD code, but not vice versa. With GPL code, people must conform to a narrow vision. With BSD code, people are free to do what they want.

  80. Quoting esr:

    In any case, where Firefox gets their money is not very relevant to my argument. If we live in a type B universe, Firefox will attract either profits or Google-like subsidies because open-source development is the most efficient way to produce a browser

    Did you know that Google is making its own browser (Chrome)?

    Did you know that by nearly every measure, (other than ‘availability on *nix’ (*) and the absence of plug-ins (read: no Flash)), it is superior to Firefox?

    (*) I do understand that you’re not going to run a Windows-only browser. Efforts to bring Chrome to linux are underway (and open source).

    That said, the rise of Skia (the graphics engine in Chromium and Android) means that less and less ‘giflib’ will be in-use.

    Skia has its own parsers for GIF, JPEG, etc.

  81. Oversimplification. Unfortunately market is much more complicated than that, and it tends to go towards local rather than global minima (which you require for your argument to make sense). See for example “The tragedy of commons”.

    FAIL.

  82. > If FreeBSD had been licensed under GPL, Apple couldn’t have used it as the basis for Darwin and thus OS X, and all
    > those Mac users would not now have a high-quality operating system with the best tools available.

    Do you think Apple’s choice of FreeBSD might have had anything to do with Jordan Hubbard?
    Do you know the history of ‘mklinux’?
    Do you think its possible that Apple picked the FreeBSD environment simply because it is better?

    “BSD is what you get when a bunch of Unix hackers sit down to try to port a Unix system to the PC. Linux is what you get when a bunch of PC hackers sit down and try to write a Unix system for the PC.”

    Perhaps these opinions will get me banned.

  83. @Scott Lawton

    > The GPL view is narrow: “I’m only going to give you my code if you pay me by giving me yours”. The BSD (etc.) approach is much more open: “I want there to be more good software in the world, even if I only benefit indirectly”.

    GPL associates a “cost” to the software in terms of liability of giving back any derived code, whereas BSD does not. GPL ensures freedom for the code irrespective of how many user-hops it makes. BSD ensures freedom for the immediate user of the code, and no further. So, GPL is narrow for the user but open for the code. BSD is narrow for the code but open for the user.

    > Apple’s use of BSD clearly increases the amount of good software in the world. No matter how many proprietary extensions Apple has added, developers are still better off than a fully proprietary system. Evidence: look at how many open source developers have MacBooks rather than Linux laptops.

    Not using FreeBSD could have harmed Apple (compare Nokia’s liability with Symbian). But, how does FreeBSD benefit from this? How is this a good example for developers to write code under BSD license and still expect to earn money? It is obvious that the reason Apple manages to sell so much is due to the (perceived) value they provide. But then, some also argue that Apple hardware is overpriced. Had MacOS been built upon GPL code, that would have created healthy competition and no-less-value in terms of open code.

  84. >Perhaps these opinions will get me banned.

    No, but trolling me about my policy might. Mind your manners.

  85. I agree with what you have to say. However, within our type B universe, those of us who develop open source software while making reciprocal licenses optional do rather well. We do not alienate allies of any kind, be they companies or other hackers who find some kind of religion in a reciprocal license. However, occasionally, I get an e-mail saying “HOW COULD YOU?!”, referencing my choice of software licenses. Such correspondence has and will perpetually lack replies.

    Most things that I release into the wild are released under the 3 clause BSD license (aka X11 license) that allows people to pretty much do as they wish with the software. A viral license blocks .. so do patents, if someone wants to use my stuff. GPL3 tried to solve that, most home brewed distributions do not have the disposable capital to negotiate a patent license that is agreeable with version 3 of the GPL. Moreover, they don’t have millions to defend themselves in court should they profit from their efforts. I can see why GPL3 was needed, I just don’t agree enough with its language to use it.

    Yet, there are companies that do extremely well with version 2 of the GPL. They require all contributors to accomplish a copyright assignment form ensuring the company remains the only steward of the code. The company can then dual license the program in any way that they see fit. If you gave me a really great program, I’d have no problem giving you patches and my copyright.

    I have to say, again, GPL3 scared me, even after providing some input as to what it should cover. A conversation between RMS and myself entailed a section that would cover medical devices. Due to federal regulations, software on these devices can not be modified en situ.Would you want a device that was designed to save your life hacked by someone who did not fully understand the device?

    However, are you comfortable with the usability quirks of these devices that have the potential to end lives due to someone pressing the wrong button on a an angry fruit salad? Some of these issues have gone on for YEARS, which clearly illustrates why most intelligent people want to live in universe B.

    In any event, these devices were not covered by GPL3. RMS is vehement, lives in self imposed poverty and has dedicated most of his life to improving society. Thank you Eric, in your arguments, you have separated the man behind the curtain from the universes that you described. Economics is not in his plan, he advocates not having children in order to live minimally. I have a 3 year old, we live minimally, we need to eat.

    In conclusion, yes, especially with times being as they are, the GPL is unattractive to many. Its not money that I’m worried about, I simply care not for imposing restrictions that are spawned from reality or paranoia. My intent ended at releasing the goddamn thing, do whatever the hell you want with it :) Yet, hopefully, I get to eat some fruit from what I did .. and viral licenses tend to make that fruit wither on the vine.

    Good rant.

  86. I failed to fully articulate, I firmly believe that once software patents have vacated our society, the GPL will have truly served its purpose and can be retired. I can’t get mad at the FSF or the GPL, they are only trying to help in their own way .. and I really appreciate their efforts. I have a great deal of respect for RMS, when he talks, I listen and digest what he has to say.

    So, I guess its possible to appreciate someone but not blindly swallow their ideals, no matter how sweet they might seem. I point this out only because many fail to draw the same conclusion.

    Sorry for the double bang .. perhaps you can install some WP plug-ins that (1) thread comments, (2) allow people to edit comments up to 15 minutes after submission? If you like, I’ll email you the plug-ins, I use them on my blog with great success.

  87. Interesting. I almost wrote that the real problem is not that the open source version would outlast a proprietary fork (not even a debatable point — where are proprietary X servers now?), but rather that in the meantime there’d be a whole load of wasted effort going into making these proprietary forks which would be better spent improving the open source version. Then I realised that’s kinda contrary to how things work in the real (non-software) world; even though capitalism means a whole lot of duplicated (“wasted”) effort, it works so well because the improvements we get from having competition far outweigh that inefficiency.

    So here’s an interesting vector for discussion, if anyone’s interested: could competition from a proprietary fork actually help open source projects improve? Have there been any cases where this has happened?

    (I don’t have an opinion on this, yet.)

  88. “BSD is what you get when a bunch of Unix hackers sit down to try to port a Unix system to the PC. Linux is what you get when a bunch of PC hackers sit down and try to write a Unix system for the PC.”

    With the expected result: BSD is beautiful, but its support for various flavors of PC hardware and peripherals is still lacking, even by modest Linux standards.

    As for Apple, Mac OS X is Mach + BSD layer inherited from NeXTStep, not FreeBSD. The fact that they hired some FreeBSD hackers is probably due more to the cold-hearted engineering concerns of BSD code being more compatible with what they have, rather than inherently superior.

  89. Lewis:

    I’m not sure, but I do think that many software projects don’t (short-term!) fork enough when exploring design space. Especially on proprietary software, there’s a tendency to assign 5 developers to work together on a project when you’d be better off having them develop 5 different implementations and throwing four away.

    Some version control systems seem to encourage this sort of behavior; Git in particular, but it may tend to be a feature of many of the newer systems – I’m not really familiar with the state-of-the-art in VC, but I do get the idea there has been some paradigm shift from the glorified CVS systems that were the standard some years back…

  90. I think people who make proprietary forks seldom are clear sighted enough. They spend all their effort on their own fork and none on the Open Source branch. The smart thing is to just keep the really differentiating bits proprietary, and let the majority of your work go in the Open Source version. This keeps the code base more uniform, making it easier to maintain the fork. Forking off and not keeping in sync with the main project is just plain stupid, unless the main project is dead in water.

  91. With the expected result: BSD is beautiful, but its support for various flavors of PC hardware and peripherals is still lacking, even by modest Linux standards.

    It runs on the hardware I own, or otherwise want. I don't see the problem. And yes, it is well-structured.

    As for Apple, Mac OS X is Mach + BSD layer inherited from NeXTStep, not FreeBSD. The fact that they hired some FreeBSD hackers is probably due more to the cold-hearted engineering concerns of BSD code being more compatible with what they have, rather than inherently superior.

    MacOS X borrows FreeBSD's virtual file system, network stack and components of its userspace. Apple continues to integrate new code from and contribute changes back to FreeBSD.

    Prior to Hubbard's arrival, MacOS X had a user-level based on netbsd. After, it was switched out for FreeBSD.

    Because it was superior.

    Apple could have run linux. As it stands, MacOS X (and therefore BSD) outships linux by a wide, nearly unrecoverable margin.

  92. Like pretty much any economic argument, this assumes that “profitable” is the same thing as “good”.

    You may think that commercial success is the only important measure of success, but not everyone thinks so–a project may be a resounding success even if nobody’s getting rich from it–and if you’re going to make that assumption, you should make it explicit.

    How is an ecosystem of free software valued here, if none of it is sold, but it’s still used extremely widely?

  93. Apple picked FreeBSD and then hired Jordan Hubbard, not the other way around.

    I agree 100% that Apple would not have chosen any GPLed base for their OS, because it would not have allowed them to do what they wish with the code. Since I'm typing this on a Mac Pro, and have several other Macs, I'm not about to fault them for the decision. The BSD license allowed Apple to bring a robust, easy to install and manage Unix system to the masses – something that GPL advocates still wish they could truly do.

  94. OK thanks to those *invisible-to-Aunt-Tillie-but-nevertheless-important code*, BSD outsmarts Linux. Thanks for clearing it up :P

  95. That's all well and good if you make it VERY clear in the code that you are receptive to proprietary forks. Most 'managers' see GPL3 and run like hell, never bothering to ask. If I worked for …. TomTom, I would not even consider GPL3 code. No way in HELL could I afford the litigation and settlement to satisfy reciprocal patent licenses.

    We had 3 million USD in funding to set up a shop that did nothing but pay people to work on open source software. GPL3 came out, that money quickly vanished.

    The point is, since more than one version of the GPL exists, please specify the version you are describing (or all).

  96. Our economic collapse was due to unregulated prospectors on wall street, the world depression has nothing to do with software licenses. When the news says 'recession' it really means depression. When the news says 'no recession', it means recession. The GPL is more of a picturesque 'ideal world' than anything.

    In my ideal world, people are just automatically rewarded not just for their talents, but contributions to society. Unfortunately, that is not the case, it won't be the case for some time .. in the mean time, we have to eat :)

    So, use my code, make some money .. if you run in to me, buy me a beer. :)

  97. @Shantanu Kumar

    > GPL is narrow for the user but open for the code. BSD is narrow for the code but open for the user.

    Agreed! RMS (et al) clearly care about code. I care about people.

    I believe that developers (who get to choose how to charge for their contribution) and users (who get a wider variety of software) both benefit from unencumbered licenses. Incidentally, of course I agree that developers have the right to choose GPL, or BSD, or proprietary. That's part of the free market too.

    > Apple … But, how does FreeBSD benefit from this?

    At the very least: by creating a larger ecosystem for *nix, which helps Linux too. Individual developers (FreeBSD or not) benefit from an additional hardware choice (that has proven popular, despite the higher purchase price). Finally: I would guess that Apple has contributed useful code … though I don't follow the issue, and my argument doesn't depend on it.

    >How is this a good example for developers to write code under BSD license and still expect to earn money?

    Again: larger ecosystem means more companies who hire (whether employee or contract) *nix developers. My tiny company included. In the old days, we wrote Mac-specific software. Now much of what we do can be done by contractors on their Linux boxes and will "just work" on my Mac. The major developers and key contributors often get hired (at nice salaries) by large companies, e.g. Guido (Python) by Google.

    > But then, some also argue that Apple hardware is overpriced.

    The great thing about a free market: nobody is forcing those people to overpay! Just buy something else.

    > Had MacOS been built upon GPL code

    That never would have happened. Plain and simple. Many companies, large and small, simply avoid GPL even if that means writing something from scratch. These companies won't pay for custom extensions, and generally won't hire the developer.

    Of course some companies do fine with GPL, and (as noted) use it as a competitive weapon. That's part of the landscape — but I'm not convinced it's increasing and I'm positive it's not going to be universal.

  98. IMHO the the economic good = profitable. While there are other kinds of good, which aren't economic. The discussion of those would not be, obviously, called, the "economic case", but some other case.

  99. So we start from the assumption is that either open or closed source licences are a marginally better basis for efficient software development. (If the benefits were not marginal, their coexistence would have been much briefer.) While this may be the case, there is a non-trivial cost in switching development from one to the other, at least without the legal author's agreement to open-source or to dual license respectively. At the very least, this cost is that of the code that has to be rewritten and debugged under the new licence.

    The higher the transition cost, the more difficult the transition, the less efficient on the short and medium term can the software development be.

    Non-GPL licences have no impact on the cost of replacing a closed-source product with an open-source one as they do not apply to the former, but waives any cost of using open source products and innovations into closed-source products. Therefore, if closed source is beneficial, this type of licence will accelerate the adoption of the closed model. But if open source is the better model, then, non-GPL licences will allow more breathing space to closed source developers. See the case of Apple, who managed to save a significant amount of effort in developing the proprietary OS X, by basing it on an open source foundation. With GPL, they would have had either to write their version of UNIX from scratch or to open-source the end result.

    With GPL licences costs become symmetrical. Open source projects need to redevelop any proprietary code (e.g. GNU/Linux and Unix), but closed source projects are also forced to redevelop any features of open source software they want to include into their product. Thus the costs of porting from close source to open source stay the same, but the costs of the reverse switch reach a similar level.

    If the closed source model is better, this will of course increase costs now and prolong the agony of open source. However, if open source is the better model, while immediate costs stay the same, on the long-run open source would benefit from the diminished ability of closed source projects to piggyback its own advances. This diminished ability would translate in some proprietary entities "seeing the light" or exhausting their results faster, thus releasing their customer base and developers to open source sooner.

    The business argument above, as any business argument has a philosophical basis. I believe that the I&IP government regulations, as any government regulations are only making things worse and actually provide benefits to the regulatees. I&IP rules have been advocated as an attempt to encourage more useful thoughts to become publicly available, by creating an artificial scarcity – and thus an economic value – for certain types of mind products if they become publicly available. In fact, the spread of ideas has been hampered, as R Stallman found out when he noticed that the price of commercial UNIX increased steeply after it was improved using his (and others') public domain contributions to Unix. GPL is the instrument used by individuals to privately repeal copyright (and to some extent patent) law in relation to their work. Other open source licences, do challenge the entitlement of government regulations on I&IP, they do not abolish them, rather they suspend them for a particular version of that piece of software.

  100. Confusing typos:

    "proprietary entities "seeing the light" or exhausting their RESOURCES faster"

    Other open source licences, do NOT challenge the entitlement of government regulations on I&IP

  101. Folks, I have a suggestion. Whenever scores of intelligent people are engaged in a seeming endless debate and always re-occuring debate about whether that bird on the picture is a duck or a goose, then perhaps it's time to change the level of zoom.

    Let's zoom out, for example: is there are clearly more efficient licencing policy for ALL intangible products, including f.e. high-budget Hollywood movies?

    Let's zoom in: could one kind of licencing be more efficient for one kind of software, and another kind more efficient for another kind of software?

    i.e. what if the Schelling points are not _around_ but _within_ the category called software: say, between broowsers and office suites?

  102. Intellectual property is a misnomer. Please show me some living being that is solely responsible for the shaping of their intellect? Such a person would have to be birthed on a deserted island and abandoned immediately, blind and deaf so that even animal influence did not help shape their intellect. To be thorough, such a person should also have no senses, so that smells, sounds, tastes and textures could not contribute to their intellect. If such a person emerged, I would happily pronounce the primitive grunts coming out of such a being as their intellectual property.

    What remains is copyright, patents and trademark .. which can not be grouped. Some can be challenged, some can not. Some can be renewed, some can not. Please do not use this misnomer, however convenient it may seem.

  103. >> GPL is narrow for the user but open for the code. BSD is narrow for the code but open for the user.

    > Agreed! RMS (et al) clearly care about code. I care about people.

    One downside I see to GPL is — if the individual or the company who wrote the code gets hit by a bus or goes out of business (respectively), then the code itself cannot be used by people unless they are willing to accept GPL (assuming they cannot contact the original author anymore).

    >>How is this a good example for developers to write code under BSD license and still expect to earn money?

    > Again: larger ecosystem means more companies who hire (whether employee or contract) *nix developers. My tiny company included. In the old days, we wrote Mac-specific software. Now much of what we do can be done by contractors on their Linux boxes and will "just work" on my Mac. The major developers and key contributors often get hired (at nice salaries) by large companies, e.g. Guido (Python) by Google.

    What if the code under BSD license is not related to *nix in general? Examples: Bio-technology related analysis tool / CMS / Telecom product — I purposely picked these examples because such things generally they take a lot of time and effort to build right. Can you stop another (bigger, with more resources) company to build over the open source version and start selling a closed-source version that adds more value? BSD license will prove anti-competitive for the original author, and is clearly unsuitable for such a case.

    > Many companies, large and small, simply avoid GPL even if that means writing something from scratch. These companies won't pay for custom extensions, and generally won't hire the developer.

    There is good reason why more companies are building business around Linux rather than FreeBSD. Similar stats exist for MySQL vs PostgreSQL. These same companies also have such developers on their payroll, and they pay for custom modules / features / integration done. Or dare I say, it is the eco-system that GPL helps create.

  104. >Intellectual property is a misnomer. Please show me some living being that is solely responsible for the shaping of their intellect?

    Stupid argument. Someone else built my house; this doesn\\'t make it any less my property.

  105. Interesting discussion!

    With GPL licences costs become symmetrical. Open source projects need to redevelop any proprietary code (e.g. GNU/Linux and Unix), but closed source projects are also forced to redevelop any features of open source software they want to include into their product.

    That's a good point. GPL is an antagonistic license. It's us vs. them; GPL vs. proprietary. The FSF wants all software to conform to one model. (Note: that will never happen.)

    Apache/BSD/MIT etc. are cooperative licenses. They want more good software to be available, even if some of it happens to be proprietary. (Some in the cooperative community merely tolerate proprietary software; others think it has certain benefits.)

    Incidentally, Apache/BSD/MIT allows GPL code to take without giving back. GPL'd derivative works prevent the original author (and community) from using the changes on the author's preferred terms. (Authors could prevent this by creating an anti-copyleft license, but that's antagonistic rather than cooperative, so not consistent with the overall goals.)

    on the long-run open source would benefit from the diminished ability of closed source projects to piggyback its own advances

    Maybe: but only if antagonism beats cooperation.

    the spread of ideas has been hampered, as R Stallman found out when he noticed that the price of commercial UNIX increased steeply after it was improved using his (and others') public domain contributions

    GPL and GFDL also reduce the spread of ideas, and encourage re-inventing the wheel.

    I can understand the arguments for GPL; I just think many advocates ignore the cost.

  106. I would say combination of games and open source is still almost unexplored area. I believe that involvement of users in gamemaking process is a key factor here.

    I am going to start a new business based on this idea: a powerful open-source runtime that allows anybody to create games from Wesnoth-like strategies to NetHack-like adventures with minimal effort, and all this combined with optional paid content created by professional artists and composers.

  107. I think the argument only makes sense if you assume a truly free market, i.e. one where corporations don't get huge government subsidies in the form of draconian copyright and patent monopolies. The GPL was created specifically as a defense against these things. .

  108. >Let's zoom in: could one kind of licencing be more efficient for one kind of software, and another kind more efficient for another kind of software?

    In fact, my own analysis ten years back suggests that some software – such as games — has the economics of a perishable packaged good. That is, short shelf life, no support requirements, no interoperability issues because it doesn't talk to anything else. Open source licensing doesn't make a huge amount of sense for these products.

    I'm fully aware of these special cases. I don't think they invalidate the general analysis at all.

  109. I'd say you're right, but it's not between browsers and office suites: the best browser in the world is still proprietary. I'd say they lie between where people are willing to dick around with their computers and where they are not: the infinite customizability of Linux is a boon in the server room but an absolute showstopper on the desktop. These boundary lines are in the process of being severely gerrymandered by Apple, too: an increasing number of people, such as scientists, who at one time were willing to use an OS with lots of technical knobs in exchange for the power that Unix gives you, take one look at Mac OS X which has Unix power and ease of use and switch, never to return. The Mac's status as the ultimate no-compromises desktop OS is motivating an en-masse transition from Linux to the Mac as the default development platform of the hacker elite.

  110. I used to be what some call a "freetard", an ardent apologist for RMS and the Free Software ideology. I used to believe in GPL as a license that protects the pool of software licensed under it as a common that cannot be "subverted" or "exploited" by proprietary vendors. I used to believe it's all about freedom. Today I know better.

    Freedom is fundamentally based on human volition, which is a part of human nature – the ability to choose for oneself what to think and how to act. Thus anything that compels one to act against his own will or choice is tyrannical. Realizing this I also realized that a lot of what humans established to supposedly "protects freedom" actually by definition does exactly the opposite. I'm talking about governments. Without its ability to coerce people to act against their will it would not be a government. It is by definition a monopoly on coercion. Thus it is by definition tyrannical. A pro-freedom government is an oxymoron.

    What does this have to do with GPL? Well, just like proprietary software licenses GPL depends on government to function as it's supposed to. I there were no copyright laws, as draconian as they are, copyleft would be meaningless. Without copyright, there is no virulence since copyright assumes as liable even those who did not receive a copyrighted work directly from the provider, something that does not hold in reality. It is a man made law that does not correspond to natural reciprocity that occurs in voluntary human relationships. If you give someone a program and ask him not to copy it to someone else, yet he does, only he is liable for the breach of this contract, not the one whom he copied it to. The contract is between him and the provider, not between the one who he gave it to and the provider.

    This simple fact actually breaks both GPL and many proprietary licenses without government. All that's left are contracts which users actually have no problem entering into and abiding by which are unlikely to be draconian ones that prohibit such natural things as sharing with your friends. A lot of software would probably still be binary, yet distributed under terms similar to BSD. FOSS code would predominantly under contracts akin to BSD as well.

    My point is this. Free Software movement has little to do with freedom other than they talk about it a lot. Free Software is merely a welcomed choice that emerged in response to there being a lot of people in the market who were fine with just scratching their itch and then just giving the resulting code out to others to use and a lot of other people who felt more confident about software that they can use without being threatened force for sharing or modifying it for their needs. BUT that threat itself would not hover over everybody's heads if a coercive monopoly used to make it didn't exist (the government). The incentives just wouldn't be there. The market would quickly adapt to the ACTUAL needs of the market without constant corporatist exploitation of the coercive monopoly (the government) to get their way DESPITE what the market wants.

    The true source of the ethical problem then are not so much the proprietary software vendors as the government itself and more crucially, the belief in that we need a coercive monopoly to have a civilized society (and how is a society based on threats of violence as accepted norms "civilized"?).

    By default, then, for me more liberal licenses like BSD win. They're closes to what a FOSS license in a truly free market would be. GPL and proprietary licenses are broken without government. Also, we certainly don't need a freedom fighter who would if he could just force people not to make certain kinds of agreements between each other that he doesn't approve of which is what RMS is. He stated clearly multiple times that proprietary software should be outlawed (essentially saying that a coercive monopoly should be used to coerce his view of how others should behave).

  111. Great! 5 pages of comments and the next last one is right on my train of thought. What I wanted to add is that the size of the universe plays a role, more specifically the number of people using computers. (I stick to software applications here) If this number is infinite, my belief is that it is a B universe. Is earth big enough? I beg to doubt it. What I am after is large enough groups of users with programming ability. A closed-source company has no chance against a large enough (and somewhat organized) group of committed experts of usage and programming working on an open source solution. And you need a seed. In the case of semiconductor simulation software (which I know a bit) there are perhaps a few seeds, but not enough users with zeal to get near the amount of man-years that the closed-source companies put in. Users cannot afford to work with anything but the state-of-the-art, and Bob's your uncle. But the closer you get to programmers (application-wise), the less crap is tolerated from the users, and the university becomes more B-like. So, earth is an AB universe, which Shenpen hinted using much fewer words.

  112. "Source on this is Gartner.

    Which means that the peak number of Linux netbooks was in December of 2007 (approximately). Call it 5 million machines. It's now about 1M machines."

    No, those are just sales in that period, in the USA only, and only in a certain subset of physical shops that have agreements (err, 'incentives') not to sell non-windows machines anyway. It's from NPD anyway.

    It is 'current sales' not 'installed base'.

    So GNU/Linux machines are still increasing in installed base, even if you just consider that limited set of US-related data.

  113. With software development is easy to measure efficiency. If you only have to write it once, it is more efficient than writing it more than once, all other things being equal (and they often are). Market support or supposed code-quality have nothing to do with it.

    In the current 'real world' only the GPL and similar licenses provide the legal support for enforcing this ideal of efficiency, even though that was not their aim.

    BSD licenses are fine, but they enable short-term individual gain over society by adding inefficiency – i.e. code taken proprietary and kept secret. The is an inefficiency that society pays for, but the individual might gain from. Lying and stealing can be an effective short-term evolutionary gain for individuals but it comes at a cost to society; and often society gets fed up with the behaviour and gets some pay-back in the end.

    Also back in the real world, efficiency isn't everything. A big factory producing cheese is more efficient than a cottage industry creating cheese, but I don't think anyone needs to guess which model might add more to the value of a society and it's culture.

  114. (reposted from another thread)

    Responding specifically to your statement that, "we don't need the GPL (anymore)", what license would you choose for the linux kernel?

    I admit that this is at best a rhetorical question, as it is at least as difficult to re-license the kernel with a non-GPL license as it would be to relicense it under GPLv3.

    That aside, I'm still interested in your opinion and thoughts.

  115. I tried to reply yesterday, but apparently I bungled something and the comment was lost.
    I do agree property is indeed a misnomer for exclusive thought distribution rights and this is why I used the phrase 'intellectual and industrial property' only as a title for a particular type of regulations. Still, I do not agree with your argument for the reason best put forward by esr.

    I think property is indeed a misnomer because ownership is meaningful only for scarce assets. Thought scarcity operates in reverse to what generally happens to physical assets. Normally, a resource is easily available (not scarce) until its use causes depletion. Until they are shared, thoughts can be scarce (i.e. they can exist only in the minds that conceive them independently). Once they are shared, their availability becomes unlimited. Potentially, anyone can acquire a certain thought without depriving anybody others of it. It is a tragedy of commons in reverse, perhaps we should call it 'a tragedy of privacy', where potentially desirable thoughts perish precisely because they are not made public.

    Obviously, marks fall into a totally different category, as they relate to identity, and impersonation (identity theft). Brand regulation makes sense to the extent it protects the unique identity of a brand, and thus avoid market distortion.

  116. "Maybe: but only if antagonism beats cooperation. "

    Less to the point, open source and closed source development approaches to profit-making are divergent. Closed source focuses on scratching somebody else's itch, i.e. about maximizing revenue from sales; open source is about scratching your own itch, i.e. cutting own costs by sharing the total cost of development with other stakeholders. While the different focus is not always a cause for antagonism, it does create conflict.

    Cooperation is possible where closing the source does not add to the cost of further development or where open source does not prevent closed source development. Closed source does add to the cost where there is an working open source equivalent or it is in active development, while GPL should not prevent closed source to run side by side or on top of open source.

    "GPL and GFDL also reduce the spread of ideas, and encourage re-inventing the wheel."

    The difference between closed and open source is that in open source you need to reinvent the wheel only when you try to replicate closed source. While in the closed source world re-inventing the wheel is needed at least when you try to replicate features provided by your direct competitors; moreover, often there is some cost associated to building on a different developer's code which financially is equivalent to a partial re-invention of the wheel.

    The CATB provides several engineering arguments for the superiority of open source development, which automatically translate in an increased economic efficiency of the model. If they are valid, on the long term open source will prevail in the areas where code is a tool, a cost, rather than a consumer good, such as a game (even Stallman does not "believe there is a moral imperative to make those sorts of works [art and opinion] free"). However, many closed source entrepreneurs take Keynes's view that "on the long term we are all dead". It is irrelevant for Bill Gates if Linux achieves "world domination" as long as he has been able to make a living for three decades by developing MSDOS and Windows as closed source.

    What remains still debatable is the cost of entering the market. In a non-GPL world the cost is dramatically lower for the less efficient development code. This means that the less efficient development mode will be used longer in more areas and thus, the over all efficiency of code development (and usefulness) will be lower and advances would be delayed. This is the moral/engineering basis for the business case. In market terms, this situation results in lower general productivity, i.e. less growth both at macro-economic level and for the coding community specifically ( I cannot use the term 'software industry' because in the open source development it is merely one segment of the coding community).

    Without GPL, on the short term closed source development has less development costs, so more money can be spent not only on marketing (a legitimate tool) but also on lobbying for regulatory distortions of the market such as software patents or DMCA, which can further delay the success of the more efficient development model.

  117. If not property, what, exactly, would you call the ability to profit from the fruits of one's labors? Advocating the abolition of intellectual property protection – and denying its existence is merely argument by redefinition, in this debate – is nothing less than advocating that those whose work product is intellectual, rather than physical, do not deserve to profit from their work. I see this as profoundly evil.

  118. This is true only if you include GPL code as "closed source" for the purposes of BSD-licensed development. The GPL is a licensing roach motel: code gets in, but it doesn't get out.

  119. RMS,

    I think the need for software is more complex than you think, and there is utility on the restrictive licenses like the GPL. I'm working on 3 new pieces of software that I plan on releasing under GPL. Why? More and more employers are asking for code samples. As the market kinda sucks right now, I can't be certain of my job situation, so I want to be ready with a couple of medium- to small-scale software apps that I can give away as code samples on my web site. I want to GPL them so that others can look at the source code, but also so people can't just wholesale copy my source and claim it's theirs. Also, if the community finds it useful, I know my donation will remain a donation, and never become part of something proprietary. Those are my goals, much above high adoption rates or efficiency.

    I agree with you that we live in a universe more like B than A, but I think it's really a mix of both. For instance, I don't think there's a heck of a lot of efficiency to be gained by a hardware manufacturer open-sourcing the firmware to be loaded into their device at boot time, or a manufacturer of a simple calculator to release their source. Then again, maybe I'm wrong.

  120. Wow… that International Lisp Conference pic is indeed… interesting. I suppose they all use Emacs SLIME on their Macs?

  121. "If you give someone a program and ask him not to copy it to someone else, yet he does, only he is liable for the breach of this contract, not the one whom he copied it to. The contract is between him and the provider, not between the one who he gave it to and the provider.
    This simple fact actually breaks both GPL and many proprietary licenses without government. "

    Sorry, can you elaborate it? I don't grok why going after those who distribute GPL'd software with proprietary mods is anything but the enforcement of a contract, in the same way as if the RIAA would only go after those who buy a legal copy of music CD and seed it on Pirate Bay would be nothing but the enforcement of a contract?

    I think if the RIAA goes after those who themselves obtained an illegal copy and seed it further is indeed governmental intervention, because this person has signed no contract. Only the original seeder breeched the contract: the guy who bought the music legally and seeded it or otherwise copied it.

    Similarly, if person A downloads a GPL'd software, modifies it and distributes it in a not-GPL-compatible way to person B, A breached contract. If B distributes it further, he breached no contract. But A did. RMS is right to go after A but not B.

    Does he actually go after B? Is this the core idea in your argument? If yes, any evidence please that he ever went after B?

    (BTW I'm not a "freetard" I'd just like to see clearly.)

  122. Well that's one Schelling-point within the category of software but I'm thinking of others too. For example, training time per end user.

    For Apache is next to none: one guy maintains 10 servers with 500 sites with a million unique visitors per year. Apache training/study for the guy per end-user: next to nothing, a second, maybe.

    Retraining your accountant to use OpenOffice: significant.

    Firefox: Just. Works. No training needed. No learning needed.

    But again, this is just ONE Schelling-point. What you mentioned is another one. I think there are others.

    We don't need to form a priori theories here – just see what the market does, and try to explain it.

  123. I agree with ESR that the universe is closer to B than A, but if that is the case, why did proprietary closed source software become as common as it has? I'm not trying to refute ESR's points, just trying to explain how we got here. Was the software market not sufficiently free? Has it become so? Is ubiquituous internet access and the rapid collaboration it allows the deciding factor (or just one of many)?

  124. Yes, but I'm not saying RMS does go after them, just that the copyright law assumes this kind of liability which is what allows RIAA/MPAA/BSA to go after people who actually didn't have a contract with the original provider and by same token also allows SFLC to do the same. Even if they never chose such a path before (there weren't that many lawsuits in the GPL world yet anyway, at least comparably to the proprietary software and music) the legal threat still remains. This legal threat is a governmental intervention itself.

    Without it, we may very well be found in a situation where those who don't have a direct contract with the vendor, but instead got the copy second hand freely do some things which the GPL license prohibits in which case I think a lot of the purpose of the GPL would essentially be broken. If this happens a lot it would simply mean the market rejects the GPL terms as a good way of distributing software and goes a more liberal route like the BSD. The fact is today this simply cannot happen due to the ongoing legal threat and thus the market's signals with regards to the GPL have been stifled, just as they are stifled with regards to proprietary licensing.

    I suppose what I'm trying to say would come down to this. I don't believe GPL would fly in a market in which the assumed liability of even second hand copiers didn't exist, a market that's free of government intervention. So it's copyright that gives GPL its teeth much more so than BSD, just like it gives teeth to proprietary licenses. Without this the free market would simply, in individuals pursuing their personal interests, break all of the barriers they deemed worth breaking, including both the barriers of the GPL and proprietary contracts. Software would probably most of the times be free (even if not always gratis) and the best way one could restrict what you can do with it would be not to release source code or with some sort of a DRM (unlikely, since that's easily circumvented as well).

    Regards

  125. First, your analysis is flawed. Even if open source is more efficient, a claim for which you provide no justification in the post, the GPL can speed up open source adoption by forcing codebases to be open, as you note, and if you believe open is better, why allow others to close up your source like the BSD/MIT family of licenses allow? That would be like closed source licenses having a clause that anybody with the last name Smith has a right to the source and will be given the full source code if they ask for it. Why would you allow competitors to take your source and use it in the opposite way of what you think is most efficient? However, the problem with the GPL is that open source is not intrinsically superior and hence dogmatically enforcing openness is not optimal. We don't live in type A or type B, we live in type C, where open and closed source each has its own advantages. Open source allows for an open development process, where anybody can pick up a codebase and contribute to it. On the other hand, closed source allows for authors to own their works and receive compensation for the amount of effort they've put into developing it. However, these two features are fairly exclusive. I've proposed that the best solution is a mix of the two, a mixed-source or hybrid license, I think that is what will win out soon.

  126. I've said this before, but I think in defence of the GPL I can say one thing:

    Ultimately as long as human beings are attached to their creations in one way or the other, there will always be need for some restrictive software license to protect one's hard work. Whether it be a totally proprietary license or even the GPL (which forces changes to code within the GPL framework to be released under the GPL and doesn't allow for proprietary extensions) the underlying human instinct for preserving one's creation is going to exist.

    People who are noble enough to give away their work for free and allow third parties to benefit with no strings attached by giving them the option to relicense such code under a proprietary license and further enjoy its benefits without re-releasing the enhancements under an open source license will remain on the fringes of any society. Even if it can be argued that ultimately such freedom benefits everybody economically and socially, the benefits to oneself in the immediate context are so remote and intangible that not many would even think of it that way.

    Another factor is that no matter what software license you choose, GPL or any other, it's merely a contractual document with no intrinsic power without legal backing. If you cannot take a violator of any license (even the BSD license can be violated, believe it or not) to a court of law, of what use is such a license?

    For example even the BSD license asks people to preserve original copyright notices. In regard to BSD, what prevents anybody from stripping out such copyright completely and further how the hell does one even bring such violations to light if the code is made proprietary by a third party?

    Whether any one contract of business can be legally enforceable in a practical sense in all parts of the world at any given time itself is a big question mark.

    Under the current circumstances, I would argue that even the GPL is too ideal for the world, let alone a more "permissive" license like the BSD license or MIT license.

    For the record I equally appreciate people who release code whether under the BSD-style licenses or the GPL.

    I suspect that people who speak up in favour of BSD licenses rarely release code at all or even if so they create only trivial code which can be given away without a second thought.

  127. You're right on. This is called an Evolutionarily Stable Strategy (ESS) in biology and sociology. The GPL is designed as an evolutionarily stable strategy against closed source software. If you drop the evolutionary stable strategy, then you don't have a stable equilibrium. It may be the case that the OSS model is also evolutionarily stable. They probably are both fairly stable, since they're widely used without any major problems. An interesting question that Eric raised here is whether the barriers that the GPL put up have a higher cost than the lawsuits that can arise from the GPL. I doubt this is very relevant. The important question is which model is gaining more traction in society, i.e. which strategy is more stable, and the answer empirically seems to be the GPL, because it's been much more popular than the OSS licenses.

    Of course this doesn't necessarily mean anything morally or economically, as the GPL gaining ground may just be a fad. But I doubt it. If you drop the micro-scale selfish game theoretic economic models (that Eric might be inclined to as a libertarian?), and instead look at large scale patterns of social behavior, you see that people are actually "against capitalism" to some extent. That is, licenses such as Creative Commons noncommercial and GPL are very popular, presumably because people don't want money to be made unless they get a cut of the money, or just because they're anti-business progressives, or for "group solidarity" (establishing barriers to other groups), or some other reason. These reasons will often be "irrational" in the economic jargon of game theory. However, that doesn't really matter, as the only reason the rational selfish models of capitalism (e.g. Hayek) are justified in the first place is that they are an evolutionary stable strategy, i.e. capitalism is not justified on rationality but stability. So I basically dismiss Eric's argument as being circular reasoning — it uses selfish economic game theoretic models for human behavior, which are justified due to their ability to produce stable societies, to dismiss a competing stable evolutionary strategy (the GPL) which is stable and has a lot of users. Without making any further arguments, this is circular reasoning, as reductio ad marketum doesn't magically make other stable social organizations go away, it at most establishes that a market is one of many stable social organizations. Furthermore, the GPL actively works against other licensing schemes by intentionally working better with the GPL than other licenses — certainly a stable Darwinian strategy.

    Eric's argument could be made stronger by using data, maybe he has personal experience or statistical data that show why the GPL is damaging. But without this data it's not at all obvious why the GPL is worse just by invoking an unsubstantiated economic argument, esp. without considering the social and moral factors that cause people to adopt the GPL. Not that I'm on Stallman's side, as I don't mind closed source software. It's just that a priori one would expect the GPL to be more stable given its barriers and viral nature, and we also see the GPL being very popular in FLOSS software, so it seems backwards to argue against the GPL using an argument of evolutionary stability (e.g. economics) given the GPLs seeming evolutionary advantages, especially without any data to back the claim.

  128. > In all worlds, markets seek efficiency, because investors are constantly seeking the best return on capital.

    Nope. Investors seek the best return on THEIR capital.

  129. You are partially right when you ask "doesn't the GPL technically forbid proprietary forks?". It is forbidden, if and only iff, the original GPL code's copyright holders did not all agree to the fork. Consider MySQL. MySQL is GPLed and they offer the code in a proprietary license. They can do this because all contributors to the GPLed code MUST assign the copyright to MySQL who, being the sole author, can offer the code base in multiple licenses. The GPL does not disallow multiple licenses if the code base that it is offered on has all the authors agree. Linux, today, is not able to do that because of the number of contributors and in order that each and every one is tracked and asked if the code can be offered in multiple licenses, is a hard thing to do.

  130. esr wrote:

    >I would still prefer to hold (+c), because I fear that in a (-c) world creative work would be drastically underproduced. So my thinking time on this topic is devoted to trying to find Schelling points on the enforcement gradient. I regard the question of whether such points exist as equivalent to the question of whether IP rights are necessarily a creature of government fiat.

    I am not sure that the concept of Schelling Points is required for creativity to be properly rewarded without government enforced monopolies on the reproduction of specific patterns. Perhaps I am just not understanding how the concept applies here and maybe you could spell it out for me more clearly.

    I do see mechanisms that reward creation without need for government enforced pattern monopoly, and I see no reason to believe that these mechanisms would not reward creative behavior to the most economically efficient level. Provided that the status conferred by authorship/invention of a given pattern is viewed as a valid property right, I see no need to enforce any monopolies on pattern reproduction.

    Some of these mechanisms are obvious, such as the known creator of a certain pattern being more likely to receive offers for additional work related to that pattern. For example: Eric Raymond is far more likely to be payed to appear at an event to talk about his book "The Cathedral and The Bazar" than is Sean Hastings – even if Sean Hastings hypothetically happens to have some equally interesting things to say about the book, and is known to be a much better public speaker. ;-)

    What may not be quite as obvious, is that these author/inventor related market advantages also exist in the merchandising of a pattern. A copy of CatB signed by Eric Raymond is clearly more valuable than a copy signed by Sean Hastings, and it is also true that an authorized copy of the book produced by Eric Raymond (or his authorized agent) is more valuable than a knock-off copy produced by Sean Hastings. If both Eric Raymond and Sean Hastings are offering copies of CatB at the same price, Eric Raymond will sell more copies.

    There are two basic reasons for this:

    1.) If consumers value the act of creation, they would rather pay the creator of the pattern in question, in hopes that this reward will induce further acts of creation that will be similarly valuable to them.

    2.) There is a status associated with having an "authorized copy" of any pattern, rather than a "knock-off." This is similar to the concept of a "Veblen Good" but the status increase is generally perceived to be with respect to appreciation for the act of creation, not just the price payed.

    Now if Eric is selling CatB copies at a higher price than Sean is, he will still beat Sean in the free market, provided the price difference does not exceed the value that consumers place on the act of creation (as the sum of the above two mechanisms). If the given pattern is thought to be an obvious derivation of previous work, then the act of creation is not highly valued, and Eric can not charge much more than Sean for a copy of Eric's book (signed or not). But if the new pattern is considered to be important and unique, then Eric's fame as the creator of the pattern is greater, and he will be able to charge much more for his authorized copies than Sean the knock-off vendor can.

    So where the current concept of copyright and patent law creates monopoly pricing, without such law, the market provides a competitive environment which produces a true free market valuation for each act of creation and rewards the creator appropriately.

    One last additional point is that where knock-off copies of a pattern are not suppressed, more total copies of the pattern will be produced and distributed. This is true because monopoly pricing will always produce fewer total sales at higher price than competitive pricing. What is interesting here, is that the pattern creator's fame (and thus the described market advantages) increase with each copy of the pattern that is created and distributed. So not only does the free market scenario more fairly value the act of creation, but in many cases, the pattern creator should actually be seen to do much better without any government enforced monopoly on pattern replication.

  131. >Ultimately as long as human beings are attached to their creations in one way or the other, there will always be need for some restrictive software license to protect one's hard work.

    As long as human beings are attached to their sources of income, there will always be anti-competitive behavior of all sorts, unions willing to use force to prevent new players from entering the labor pool, local businesses trying to prevent competitors from setting up shop next door, and old established industries trying to prevent upsetting technological innovation. people often claim property rights over all sorts of things that can not beneficially be considered property – patterns in which anyone might otherwise arrange their own physical property, patterns of words or bits on someone else's computer, useful business processes that a competitor might also adopt, etc.

    There is often a desire by one person to claim ownership of some (or even all) of the actions of someone else. That does not mean that we "need" to view such limited (or complete) slavery as legitimate. It is neither morally correct nor economically efficient to allow people to impose such control over the actions of others by calling it a property right.

  132. You incorporate it into your closed source code secretly and then don't tell the end user so that they do not know that they have a right to see the source, copy, share etc.

  133. What you're talking about is completely different. I wouldn't generalize here and say that anti-competitive behaviour is entirely due to attachment to one's *own* creations.

    There is no such thing as a totally free market or pure competition. The factors you mentioned are very real and practical and there's no way that you can adopt a moralistic stand on something as basic as a survival instinct.

  134. The biggest assumption you have made in the above is that every customer out there is completely educated and knowledgeable about all aspects of the product which itself is flawed in my opinion.

    The two points you mentioned are
    > 1.) If consumers value the act of creation, they would rather pay the creator of the pattern in question, in hopes that this reward will induce further acts of creation that will be similarly valuable to them.

    I know that this is a big hope of the Open Source community's economic model, but I would suggest that lower pricing by "copiers" would definitely win more customers. Look at the official CD sales of most linux distributions – I bet that very few people across the world actually buy the official CDs – they would much rather download from a mirror or if they are limited by bandwidth, from a local distributor who can also deliver the product earlier. I know that would be the smarter thing for the customer to do.

    > 2.) There is a status associated with having an "authorized copy" of any pattern, rather than a "knock-off." This is similar to the concept of a "Veblen Good" but the status increase is generally perceived to be with respect to appreciation for the act of creation, not just the price payed.

    I wonder how many people actually put status ahead of price paid. For example if I got a Linux CD in 3 days from a local vendor at a much cheaper price than the official original CD from the actual distribution owner at a higher price and had to wait 3-4 weeks for shipping, I know which I would prefer to buy even though I deny the "original" owner of revenue and reducing his incentive.

    Ultimately your logic is flawed because people will always assume that there are "others" who will take that route, and in the meantime, the original developer of the software is only getting a very small percent of official sales because of the easy availability through other channels, official or not.

  135. For example even the BSD license asks people to preserve original copyright notices. In regard to BSD, what prevents anybody from stripping out such copyright completely and further how the hell does one even bring such violations to light if the code is made proprietary by a third party?

    With the GPL, what prevents anybody from stripping out the copyright/license?

    Same sort of deal, people can do it, technically, but they are legally restricted from doing so. In the case of BSD:
    <pre>Redistribution and use in source and binary forms, with or without
    modification, are permitted provided that the following conditions are met:
    * Redistributions of source code must retain the above copyright
    notice, this list of conditions and the following disclaimer.
    * Redistributions in binary form must reproduce the above copyright
    notice, this list of conditions and the following disclaimer in the
    documentation and/or other materials provided with the distribution.</pre>

    It is trivially easy to spot license violations in either the GPL or BSD case. Although I believe violations of the BSD-like licenses are far less frequent because the license so so easy to uphold even if you're an evil corporation like Microsoft (hell, boot the Windows Vista disc, click "What to know before installing Windows" and it lists out all sorts of BSD-like copyrights, including one copyright from the Regents of the University of California, Berkeley :-).

  136. @hari

    >The biggest assumption you have made in the above is that every customer out there is completely educated and knowledgeable about all aspects of the product which itself is flawed in my opinion.

    Not sure what I wrote that makes you think that I have any such assumption.

    > I bet that very few people across the world actually buy the official CDs – they would much rather download from a mirror or if they are limited by bandwidth, from a local distributor who can also deliver the product earlier.
    >…
    >I wonder how many people actually put status ahead of price paid. For example if I got a Linux CD in 3 days from a local vendor at a much cheaper price than the official original CD from the actual distribution owner at a higher price and had to wait 3-4 weeks for shipping, I know which I would prefer to buy even though I deny the "original" owner of revenue and reducing his incentive.

    Your comments on delivery time, etc are in fact a different part of a full measurement of "price" – these effects are part of what forces the price of an authorized copy to come down to the market price + the free market value of being an authorized copy. This is a good thing. It reduces the price of the service of providing the data from monopoly levels to free market levels through competition.

    I am not saying that authorized copies of popular patterns will do as well or better without government granted pattern monopoly. Monopoly distorts the free market and can certainly make certain people very rich in doing so. I am saying that the free market solution is more economically efficient – which is to say that it represents the greatest value for all people combined.

    Consider that lack of monopoly pricing means less money for marketing budgets which means that patterns rise or fall in popularity based more on true merit than having information channels clogged by those promoting them. This means lower barrier to entry for new quality patterns and less persistence of old flawed patterns. Which turns out to be an additional incentive for innovation.

    Imagine a world in which hardware vendors could wither put an authorized copy of Microsoft windows on a new machine they ship, complete with a sticker that says "Microsoft Authorized" or, equally legally, choose to ship a machine with an unauthorized copy of windows on it at a cheaper price. Do you think that this leads to a world with more or less innovation in the OS market?

  137. Maximum economic efficiency is that which creates maximum overall human value. This is the same thing as maximum moral good. All human wants create a human value market. Survival and reproduction are the two strongest desires that humans (and all replicating entities) have, but they are qualitatively no different than any other desire that creates value in the market.

  138. Yes, I agree. But I am in way over my head here so my explaining my own assumptions about different categories of software has no value to anyone. But I would be interested if someone with more insight would.

  139. I completely agree. The GPL isn’t but an anti-IP proposal which requires of IP in order to suffice. It is thus self-defeating. (IP=Intellectual property)

    The objection one must cast from a libertarian standpoint to the GPL is that cannot rely on an entirely contractual basis, where high enforcement costs—e.g. of code leaking, contract breach, or even theft and code leaking by someone else (who in turn distributes the code to a third party)—be internalized by individual agents. The GPL’s terms require of state coercion for this purpose. (In this regard, the GPL is not essentially different from DRMs.)

    The crux of the whole debate—as it oft happens in most debates today—is what we “freedom” is about. RMS seems to take a left-leaning stance of the concept: to him, we are entitled of a “right to read,” a right to explore all software’s code, etc. To me, we only have a right to exchange goods and stipulate the terms of that transaction—which of course implies a right to agree that a certain software should not be leaked, and impose a penalty in case it happens. But we don’t have a right to prevent others than those who engaged in that contract to keep it secret. To libertarians like I, the State has no right to coerce individuals to keep others’ secrets.

    Regards,

    PIC

    PS: Please tell me which “written accent” I have. I really wanna know since I’m not an English native speaker… (Don’t judge by my surname)

    ESR says: Sorry – just generalized foreign. One minor usage error of a kind a native speaker would be unlikely to make (should be “libertarians like myself”, “I” is a subject pronoun that doesn’t belong here) and a couple of places where a native speaker almost certainly would have made slightly different usage choices.

  140. Imagine a world in which hardware vendors could wither put an authorized copy of Microsoft windows on a new machine they ship, complete with a sticker that says “Microsoft Authorized” or, equally legally, choose to ship a machine with an unauthorized copy of windows on it at a cheaper price. Do you think that this leads to a world with more or less innovation in the OS market?

    Let’s not go there. Do you really want the distro proliferation in Windows that has plagued Linux?

  141. You make some excellent points. I think Creative Commons would be a better open source software liscense

  142. esr implies that placating to fear is a reasonable economic decision (shall we say, business strategy) in the case of software. Creating fear may be profitable – even if socially unsound. But prostrating to?

  143. It’s an interesting question although I wouldn’t think about it precisely the way posed by Eric.

    Technology makes other factors of production more efficient. Because it is very close to a perfect competitive market (like the financial markets), it by default has negative return on investment. It only has positive return when there exist natural monopolies, or rent seeking. The continued betting on technology by investors therefore is much like picking lottery tickets (for disruptive/monopolistic change) and needs the large payoff (“we all get rich”) to be sustainable. This also requires closed-source or at least a bunch of secret recipes, to justify the investment dollars, since consulting or other open source business models can not as easily scale and don’t have the moat.

    Yet because investment in technology generally has negative return, the converged, rational economics outcome is for software that is prosaic, functional, infrastructure, or behind the cutting edge, to converge to open source unless there is a high enough monopoly wall. Much of the benefit for employees of technology will be from closed source, so this means people working in technology have an incentive to favor closed source (this would be the monopoly / rent seeking behavior that most professions pursue). However despite this preference, over the long run (except for monopolies) much of closed source has negative returns so this drives a general economic convergence towards open source. However in the special case where there is a strong monopoly moat, and the new software could reinforce that moat, it makes much sense to keep the software or at least the secret sauce closed.

    It’s a similar analysis as for low cost producers. As the waters in a monopoly moat lower this eventually provides a comparative advantage for open source which can operate at significantly lower cost. Thus open source generally wins in the long run, since it can produce commodities (roughly defined as software decades after the first cutting edge research happened) at a cheaper cost. Unless the monopoly has natural advantages which allow it to develop new software inside its moat (e.g. Apple’s genius at marketing and aesthetics, Bill Gates’ former genius at conquering the world with embrace-and-extend).

    But a final factor with the explosion of start-ups as well as open source is that people in technology just want to work for themselves on what interests them irregardless of economic signals, down to the salary level where it starts to impact happiness (and since happiness versus wealth generally is log-like with an abrupt bend around the upper-middle class income this makes perfect sense). So the economic forces are herding already a very disorderly bag of cats and will converge only very roughly over many decades.

    To finally get to your argument I’d say it’s interesting but flawed because the big factor in the converged economic outcome is whether monopolies can be built in closed-source land. The GPL prevents new closed-source monopolies from being built on top of open-source infrastructure, whereas MIT license does not prevent this. In other words, the GPL might tip the balance, probably for the benefit of the FLOSS crowd. Another weird side-effect is that the GPL has global effects so even if we converge to a mostly GPL world, it may have been that a closed-source dominant arrangement would be more economical. In other words, the free software guys may work really hard with trillions of man-hours, GPLing everything, and successfully prevent closed source from competing at all in the market economy, only to find out that they are all poor and only IBM is making consulting money off of them, and not even that much money. The socialist-GPL outcome I suppose.

    My new landlord wrote a lot about economics and open source:
    http://en.wikipedia.org/wiki/Eric_von_Hippel

  144. Pingback: GPL, for Goodness Sake

Leave a Reply

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

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