This article is available under the GNU General Public License (GPL), version 2 or later.
Feel free to contribute encouragements, suggestions, money, or whatever.
Read more detailed warnings about my articles.
These are a few remarks about incorrect things that I think lie in Eric S. Raymond's otherwise fine and deservedly well-known articles. I sent these remarks to ESR first, but he in his full mind and his full right chose to not take them into account. As I fear this makes his articles propagate a few noticeably bad ideas as parasites to the main, correct, ideas that these articles demonstrate, I feel like I must denounce them, which I do here.
Of course, my remarks are best read after knowing ESR's articles, but I've tried to make all my arguments self-contained, so that this article might stand by itself. This does not mean you shouldn't read ESR's articles; you should, because they are very interesting, and they are entertainingly well-written, too.
At the time this reply was originally written, the Cathedral article was release 1.38, and the Homesteading article was release 1.8. At least the latter article has evolved, and ESR took into account some of the present criticism in it. But it looks like from our e-mail exchange, that there are definitely irreducible divergences in our approaches to the free-software phenomenon. So the present article is here to stay.
The first part of this article was mainly written from 1998-05-23 to 1998-05-29. For those who like, there's been an amazing flamewar about it on slashdot. It was modified on 1998-06-08, to take into account ESR's additions in release 1.9 of his "Homesteading the Noosphere" article, as well as adding some material, and adding section names that turn upside down those from Eric's article. On 1998-06-14, I changed "fame" into "reputation", in accordance with a remark from ESR originally appeared in slashdot, except in one place to rhyme with "shame". Only typographical changes or stylistic tweaks were made since.
The second part of this article was begun on 1999-08-07, as a response to "The Magic Cauldron" release 1.15. On 2000-04-21, I completed a first draft of this second part.
I invite discussions related to present article, or to intellectual property in general or even to broader ethical topics to happen on the publicly archived mailing-list email@example.com.
Eric S. Raymond, a respected computer hacker (in the noble meaning of the term), has become famous as an "open-source evangelist", since he wrote the influential article "the Cathedral and the Bazaar", which now has some sequels, currently "Homesteading the Noosphere" and "The Magic Cauldron".
I read "the Cathedral and the Bazaar", shortly after it was published, before it came to be as famous as it is now, and was modified to transform the expression "free software" into "open source software". It is a great article, and I fervently agree with most of what was said, which is precisely why I feel bound to correct points I disagree with. What Eric wrote constitutes indeed a great finding about the dynamics of software development. These findings, as I think Jonathan Eunice points out though in an inadequate way, are mostly, but not quite, independent from issues of software being free or proprietary. And as a free software theoretician, my remarks will be precisely about the not so small gap between that "mostly" and that "quite".
Maybe I can qualify as the Cleric who studies free software theory in the Cathedral, whereas Eric would be the Preacher who goes down in the Bazaar to convert newcomers to the new faith. We're doing different, complementary jobs, and my current job will be to criticize theoretical errors that I think I find in his practical discourse.
My first (minor) remark was about this suggestion in the end of the Cathedral article:
Perhaps in the end the open-source culture will triumph not because cooperation is morally right or software ``hoarding'' is morally wrong (assuming you believe the latter, which neither Linus nor I do), but simply because the closed-source world cannot win an evolutionary arms race with open-source communities that can put orders of magnitude more skilled time into a problem.
While I agree with the last part of the sentence, I do not agree with the first one. What better moral argument is there than the fact that free-software be intrinsically better-fit in the long run to produce good, reliable, software? Surely, this is not a dogmatical statement; but there is no reason why morals should be rooted in a priori dogmas; on the contrary, ethics, like any other forms of knowledge, are better built on a posteriori real-life arguments.
Eric stated in private mail that while he agreed with that remark, he wouldn't modify his article, by fear of confusing readers with notions of ethics being a-priori or a-posteriori, which are alien to the context of the paper, and would require too long an explanation to be understood. Well, that notion is not alien to his conclusion, so if he wanted to strictly stick to that context, he could remove the conclusion altogether. And at least, a footnote with a pointer to some explanatory page of some kind would have helped keep things clear for everyone. Or he could have used a neutral or precise formulation, avoiding both confusion and explanation. By fear of hypothetically confusing readers with overdue precision, Eric chose to actually participate in the propagation of existing confusion about important ethical subjects. It appears to me that for Eric to end up making such a choice, might imply his being the first confused one.
In the sequel to the Cathedral and Bazaar article, "Homesteading the Noosphere", another very interesting article, Eric makes what appears to me as another confusion on the same track, that is the essential source of all the apparent contradictions that his article afterwards attempts to dispel: namely, to think that programs be somehow ownable (well, at least in a way compatible with free software theories).
If I were to conform to one of the models of hacker ideologies described by Eric, I'd rather be of the "anticommercial zealot" type. But then, I think Eric's division of hacker types is biased, partly by a conscious attempt to be humorously provocative, partly as the effect of propagating common confusions.
First off, "anticommercial" sounds incorrect to me: FSF-type of free software hackers are not at all against business, as RMS himself felt he had to remind Eric; they fervently support the idea that software services should be for pay, and that there should be some free market of such services; what they disapprove is the idea that software be owned, and the associated awful tradition of proprietary software licensing. Free software advocates are no more against business than those who fought slavery were against trade. They were against the trade of human beings, not because it was a trade, but because it involved owning human beings. We are against the trade of software not because it is a trade, but because it involves owning software. I don't want to imply that you remove remotely "as much" freedom from human beings by owning software as you do by owning people. Still, I'll argue that ownership of information is indeed a threat to civil liberties. At least, it should be acknowledged that it is love of liberty, not hate of trade, that motivates us free-software tenants.
Next, the term "zealot" appears biased to me; I'd rather say that some people are more or less prone to make compromises with software hoarding themselves, and to encourage or discourage such compromises from others (which are two independent facts, at that!). Extremism and fanatism are by no means attributes to be exclusively or necessarily attached to free software purists who would refuse any compromise; they are to be found among people of any opinion on any topic, among free-software advocates in a proportion probably neither greater nor lesser than among proponents of software hoarding. They are not associated to any particular set of opinion, but to the fact that opinions be held dogmatically or not.
As of pragmatism, I wouldn't confer this attribute to someone on the ground of one's readiness to accept any compromise. Accepting degrading long-term compromises I would call "realpolitik", which I think has proven a useless strategy against blatantly immoral enemies (see the history of fascism or communism), while strictness has often proven to be a winning attitude (e.g. towards segregation in South-Africa). I am certainly not suggesting that no compromise whatsoever should ever be done, but that their long-term implications, and not merely their short-term advantages, should be taken into account as part of the tradeoffs involved in the compromise. The Cathedral article, with my above remark holding, is in my opinion the best demonstration that the GNU way of developing pure free software is the most pragmatic way indeed to develop software. Another confirmation that GNU is a better fit is the constant splitting of BSD-style licensed projects, such as BSD Unix variants or the X project, not to speak about the perpetual start over from nothing typical of proprietary software.
Any apparent contradiction detected by ESR between free software practice and theory seems to me as due to a consequential misreading of free software theory, which he reduces to the common consensus on free software licenses. But licenses are legal and moral documents, not economic treaties; they only describe what people are free to do or not; they do not try to describe what people will actually do, or to explain why they will do it, whereas studying the latter questions is precisely what theory is about.
Hence, I wouldn't interpret the divergences between free software attitudes reported by Eric as essential contradictions in approaches to free software, but merely individual differences in interests; everyone has one's own personal attitude, that does not necessarily contradict anyone else's attitude; much like every programmer writes programs that (hopefully) differ from anyone else's programs, without contradicting them. When considering the fact that development and use of free software does involve scarce human resources of developer and user time, one might be tempted to assert that any difference is then a potential contradiction (in the choice of use of these resources). But then, this doesn't make free software differences any more contradicting each other than proprietary software differences are; much less so, indeed: free software can but let appear essential incompatibilities between programs trying to grab the same (computer, human) resources; but proprietary software also introduces oppositions of monopolies each trying to exclude everyone else with artificial barriers of usage and development fees, and absence of source. Free software hackers may be rivals; proprietary software holders are enemies. In software as in any other topic, Liberty is not about everyone being able to do everything with no limit, but about anyone being able to anything with no artificial limit.
Second, Eric sees a contradiction between the theoretical claim of Liberty in free-software licenses, allowing anyone to hack anything, and the gregarious practice by which hackers follow strict rules of code sharing. But I see no more contradiction here, than in the fact that in countries where anyone is free to participate in any association, individuals do not each create dozens of associations of which one is the only member. Liberty is a matter of civil rights. The right to do anything in given limits does not mean all the things that are in said limits will actually be done! It means that people will have hard enough a time dealing with natural constraints of the physical and social world, and won't have to suffer the additional burden of artificial human rules, beyond those that define the mutual limits of Liberty.
Because it contradicts his convictions, as well perhaps as (again) by fear of repelling some readers, Eric dislikes the idea that software be not ownable, and tries to avoid it, to the point of not even mentioning its being the central focus of most people's free software theories (when they have one, that is).
It appears to me at least that the idea of software not being ownable naturally stems from the same sources of classical liberalism and anthropology that Eric invokes as inspiration to his article. Ownership is justified towards scarce available physical resources, as the only way to have a given resource consistently controlled without introducing permanent fights among people. But software, being just a particular kind of ideas, is nothing like resources: arbitrarily many people can hold an idea without deprival to anyone, and it is forbidding people to act according to their own (in-mind copy of an) idea that introduces violence and attempts to Liberty.
This justification of ideas, information, and software not being ownable does support the establishment of a market in commercial services about ideas, information, and software, in as much as such services do involve spending physical resources. These services include research (not necessarily finding) of new information, bug tracking, improvement, and upgrade of existing information, technical support, educational courses to help handle delivered information, delivery of physical hardware or media implementing information, guarantee of availability, relevance, and/or accuracy of information, insurance of risks related to using information, etc.
Claimed ownership of ideas, which is what free software people call "information hoarding", is but protectionism applied to information processing services, whereby people exclude each other from the market, preventing each other from offering one's honorable software services by levying protection money in the form of license fees, under the threat of prosecution. The case against protectionism has been successfully argued since the seventeenth century, and I don't feel the need to repeat it here. That some people choose to go in protectionist ways is bad enough; what looks really bad to me is that governments artificially grant them an advantage, by legally enforcing patents, copyrights, unfair licenses, so-called "author's" (actually, publisher's) rights, and more generally all that stems from "intellectual property rights". Some people (not Eric) may speak in favor of those IPRs with the conservative argument of tradition having "proven" the concept; however, that would be misapplying the argument to a domain where tradition is bound to fail, since it never had time to adapt to the novel predominance of information in society and economy.
Some may say that liberty is efficient since it is natural, or that it is natural since it is efficient; while protectionism is inefficient since it is artificial, or that it is artificial since it is inefficient. Both are valid and essentially isomorphic points of view on economy in general, and software economy in particular. And it is the only theory of free software economy that I've heard of. I can but refer to my own page (in French) about free software, and my Manifeste de la Libre Information (which has been half translated into english) for a more thorough discussion of these subjects.
In my opinion, the taboos found by Eric can then be explained by the natural constraints that apply to programming activities, which have already been hereby told about, for the most part: while programs are not ownable and will never be, the development of programs, as well as any software-related work, as name indicates, do involve resources being spent, and hackers whose works will survive and predominate are those who'll optimize the use of these scarce resources.
Programming activities first involve knowledge and understanding of the goals and means of programs, which can themselves be subdivided into external goals (as observable by ostensibly non-programming users) and internal goals (as observable by programmers who'll maintain, upgrade, and otherwise modify the program). Such knowledge and understanding are of course non-ownable information, and Liberty is the regime under which they will be best fit. But to acquire and coordinate knowledge requires the use of communication channels according to agreed upon protocols; and reliable, trusted, high-bandwidth, well-known, communication channels are a scarce resource.
The part of technological hardware costs in the total cost of communication is rapidly decreasing, as demonstrates the Internet, whose advent indeed has played a major role in the diffusion of information that made the free-software culture possible; this hardware cost of communication has always been a major concern before the Internet, or outside of it; even on the Internet, the WWW, despite its many shortcomings, has played a major role in dramatically reducing the technological cost of creating accurately identified information delivery points. Surely, the predominance of technological costs in communicating usable information outside of the Internet has helped hide the cost of information hoarding behind that of mere communication, so that information monopolies were established without people caring or even noticing, without their even being conscious of the (dreadful) impact of such monopolies on technological advance.
But with the Internet, the technological cost of distributing information has become near zero, and the predominant cost in communication is now the human factor. For humans have limited time to read and write information; they need landmarks to guide themselves in the ocean of available information; they need references in distribution of information, to which they can discharge the worries of finding accurate information; they need focal points in exchange of information, which they can trust to provide useful information, and take into account the feedback. A tightly knit collection of such focal points, where programmers can exchange their software contributions, I will call a programming project. A project has quite a physical existence, as opposed to mere programming objects that are pure immaterial information. A given project may (more or less regularly) release objects, but is not to be identified with any of them: indeed, such released objects may be forgotten, or shared by millions, and reused divergently by different projects isolated in space and time from the original project that produced them.
Software Objects are never assimilable to property, no matter how hard one tries. Ideas can't be owned, stop. Ownership of something is natural when and only when that something has an intrinsic property of mutual exclusion. This is the case of physical objects and services, but not of ideas. Trying to enforce artificial mutual exclusion is an attempt to Liberty, and doomed to both eventually fail and readily bring Oppression. On the other hand, Software Projects are as intrinsically ownable (and ownand -- to be owned) as any physical phenomenon; they have an identity in the physical world, independent of the parts of the noosphere that have been explored already and that are intended to be explored later; they consume physical resources, mainly hackers' hard work.
As I conceive things, the main point that Eric missed in his Homesteading article is then that the combined interest, trust, reputation that lies in a project, that makes people invest their programming time, which I'll call its promise, constitutes the scarce resource that free-software developers are concerned about.
An early version of this paper used the word "fame" instead of "reputation", which was a bad choice, as Eric pointed out in a letter to slashdot. I specifically did not mean and do not mean to specify irrational and uncritical admiration by the masses, which is fame in its the lowest and grossest form, but neither do I mean to restrict reputation to the peer reputation that ESR writes about, which is its highest and finest form. What I mean is an embodiment of all the actual or potential, positive or negative, rational or irrational, <foo> or un<foo> expectations from current and future (etc.) people who might or might not know about the project. Positive such expectations, as well as material inducing such expectations or countering negative expectations, do take some space in people's mind, in people's books, periodics and web pages; they are what attracts resources towards a given project, and they do occupy resources themselves.
To illustrate the fact that free software projects hold resources and value by themselves, without the artificial use of software hoarding, we may take a look at the example of RedHat software, a company that only writes free software (all redhat releases, that are software, are free), that many other vendors do propose at ridiculous pricing, yet sells enough copies of its Official RedHat CD to prosper; I imagine that this is mainly because Official CDs come with the interest, trust, reputation that constitute the promise of the RedHat project, as identified by its trademark; the fact that RedHat does propose additional services like support, packaging with a few proprietary add-ons, and enhanced packaging with paper documentation, must play a role, too, of course; still I remain convinced that these do not remotely account for all the difference in price people are ready to pay between official and unofficial versions of RedHat CDs, since few people need or use those additional services much, except for the paper documentation, that others would no doubt reproduce, too, if it could fill the gap.
Actually, if we replace in Eric's article any reference to ownership of programs into reference to ownership of projects (Eric himself sometimes uses this correct expression), then the article becomes essentially correct. Then, we can't say that programmers homestead the noosphere, the sphere of programs, which is false; programs are inherently immaterial, and no amount of homesteading or any activity can create ownership in the noosphere beyond any "magical" point. Instead programmers homestead the "dual of the noosphere", the sphere of programming projects, that are physical explorations of the immaterial noosphere.
So bad for Eric's pride of having found yet another nice two word slogan: "homesteading the noosphere". But correctness of ideas is more important than nice slogans, and I'm sure Eric is able to find a nice slogan that reflects correct ideas instead of incorrect ones. Actually he is, since after reading my arguments, he dubbed the dual space the "ergosphere", sphere of physical work. This sphere is actually bigger than my original dual, since work may or may not be about exploring the noosphere; the ergosphere is indeed the sphere of services by excellence, to which applies the classical liberal theories of economy that justify free market. Free software theory is but a natural extension to these theories.
Note that contrarily to what Eric suggests, I think that this ergosphere is not alien to the `cyberspace' (misapplied term imposed by bad journalists, to my disgust, too) of electronic locations and contents: the `cyberspace' is indeed a (tiny) part of the ergosphere (hence not at all intersecting the noosphere), with every web site participating of the effort of exploring the noosphere. This gives publication of "home pages" a territory occupation value that is fully effective, and not just metaphoric, and justifies the raging wars that sometimes happen around property of registered internet domain names.
Once we correct the above confusion between ergosphere and noosphere, everything in Eric's article becomes an immediate consequence of the above considerations, and difficulties vanish. People can and do own projects, that are more or less promiseful. The dynamics of project ownership is the fluid phenomenon that Eric's article accurately describes. And the homesteading theory applies perfectly to software projects.
A project is not to be gratuitously forked, because division splits its capital of promise in several parts, meaning less return on programming investment, unless other causes (like re-merging with different projects, or ending of major internal dissension, or taking into account divergent goals) have at least one of the resulting projects increase its promise above that of the original project. Even if sharing of code between split projects can limit the damages of forking, it costs in proportion to the projects being divergent (hence of the fork being justified), so is only an attenuator to the effect of promise division by project forking.
The mere fact that projects may be forked proves that they are not in the noosphere, since forking is not a concept suitable to immutable platonic ideas. The two (or more) projects resulting from a fork start with the same programs, at the same point of the noosphere; this is incompatible with ownership of programs, which again includes an idea of mutual exclusion; but this is compatible with the idea of owning noosphere-exploring projects, that can happen to be exploring the same part of the noosphere, though with different resources, and a different past and/or future history.
As of Eric's second free-software "taboo", against doing modifications outside of an acknowledged project coordination, it can be considered as a simple corollary to the above, with every unofficial modifier, or everyone somehow not contributing his change back to the main project constituting his own split project. A few rare unofficial patches may simply get merged (eventually) into the official project's sources; but people developing patches that don't merge or are not published at all will have to either quit the project altogether and stick to an old release with all its bugs and limitations, or have to constantly adapt their patch to the moving target of the main project. In a free-software conference in France this year (1998), commercial developers explained how this requirement to constantly spend time adapting patches was enough of a worry to justify not keeping patches secret, and instead contributing them, so that maintenance be externalized from the inside developers to the outside main project coordinator.
If, unlike Eric, we do believe that software is intrinsically not ownable, we are led to question the fact that free software implies a gift economy, as stated in Eric's article. To begin with, since software is not ownable according to us, the notion of "gift" is rather moot when applied to software: how could someone give something one doesn't own? On the other hand, gift is fully meaningful to us when applied to project resources, and programmers' time in particular.
Thus, we may see that indeed, a lot of free software currently written and distributed is done during (an extension to) the "copious free time" of hackers that thereby give away precious software programming time without being paid back in money. So surely, it may be a fact that current free software culture is mostly a gift culture. But even that fact is not obvious, since a lot of free software too is done as part of paid research, development, and deployment work.
Now, software hoarding, by shifting revenues from developers to monopolers, from legitimate work to idle racket, completely biases the software world: capitals are diverted into artificial economical privileges and spent enforcing or going round unjust laws, instead of being invested in useful activities. As a result, proficient programmers are not economically acknowledged (whether they are free-software conscious or not), and development of most free (of rights, open-source, "libre" in French) software has to be free (of charge, not paid according to its value, "gratuit" in French); which explains why free software hackers are forced to give away, willing or unwilling, their programming time: because the hoarding mechanisms mostly exclude them from mainstream software economy.
Free software developers may have something of a "gift culture", this is in no way opposite to a free market economy: it's just that the stakes were shifted, and competition takes place about available but scarce resources, not about resources that are either unavailable or abundant. Should law be corrected and stop supporting software hoarding by acknowledging artificial "ownership" in the noosphere, my bet is that development of then freed software would constitute a flourishing free market of software services that, for better and worse, has little to see with gift economy.
To me, the reason why the "industrial-capitalist production model" does not match "production of software" well, is not, as suggests Eric, because software is a world of abundance rather than scarcity, but because software is nothing that is physical and ownable; software is something intrinsically immaterial and sharable, to which the very notion of production does not apply. What the production model perfectly matches is production of software-related services, and this model justifies freedom of information rather than protectionist information hoarding.
I, among others, proposed that the right phenomenon to which to compare free software development be theoretical scientific research, which Eric did in latter versions of his article. Indeed, both free software development and academic research proceed from the same generalized phenomenon: systematic creation of new information that describes generic processes. The two phenomena are also historically related, since free-software hackerdom was born in Internet-connected universities, by students and teachers in computer science, mathematics, physics, chemistry, et al (and this relation is not fortuitous, since it were those scientists who invented everything about computer science, to begin with computers themselves). In such fields, people explore a space of purely informational objects, that intrinsically can't be owned; but they do fight for reputation and "promise", for attracting other people's interest in their own subfield, so as to have their secondary problems solved, and so as to obtain subsidies. My opinion is that unlike what Eric suggests, the sharing of adaptative patterns between academia and hacker culture may be due both to their having a continued common history and to their sharing similar structural constraints, without incompatibility, and with a deep correlation instead. All in all, I conceive that whatever to say about hacker culture should apply to theoretical scientific culture as well, and conversely.
The comparison is perfect, as far as peer evaluation is concerned: peer evaluation is recognized as the best and only sensible way to judge quality in science as well as in software. Actually, I would go as far as to extend the statement to just any information that's exchanged: it is peer review that ensures best the quality of any information, be it scientifical, technical, financial, or concerning any part of life, including everyday-life. It is it that establishes price in a free market, depending on available information. Which is why freedom of information is essential to the very balance of market economy, and why peer review will have to be acknowledged in all fields of knowledge, which requires all current informational privileges, secrets, patents, non-disclosure agreements, intellectual properties, and copyrights, to fall. However, for the sake of brevity, I won't discuss such obvious generalization in this article (I'll again refer for that to my Manifeste de la Libre Information), and I'll focus on the sole software topic. Let us just say that freedom of information can benefit business in more ways than just what free software proponents usually think about.
The notion that acknowledgment of authorship be respected, that a trace of information creation be kept, which constitutes the third free-software "taboo" seen by Eric, is also present in Science, and again, not without good reasons. First, such trace is most useful when doing research that involves analysis of historical and epistemological processes, or exploration of (historically known) related topics. And more prosaically, it is necessary to ensure that proper credit be awarded to authors, so that they get the fame or shame associated to their work, and so that further investments and grants go to deserving projects, that is, so that projects be credited a deserving amount of "promise", that attract corresponding amount of resources. People who would erase such trace (or worse, corrupt it), whereas it costs nothing to keep it, create for other people the burden of lacking useful information (or worse, having not to trust available information), and having to spend resources tracking back correct information; they are flawing the very process of peer review. Such attitude is hence assimilable to vandalism when done gratuitously, or theft when done out of interest; which is why it is strongly disapproved by custom, and should be punishable by law.
I have no major objection to the rest of Eric S. Raymond's "Homesteading" article. I can just say, about the conclusion, that there already exist some written codes about rules for free software development. Hence, the GNU GPL, already provides a framework defining software freedom, which solves the "third taboo" problem of software authentification, through the availability of a trace of successive versions or modifications. Even though in practice, this availability is not strictly enforced, discarded information doesn't usually include authorship trace, that is kept in a CHANGES file, and the GPL provides a legal tool to prevent erasure of such trace. Other common "free software licenses" are not always as good.
As for conflicts, lots of projects have developped their own day-to-day conflict avoidance methods; mostly consisting in communicating through adequate mailing lists, with some way or another of designating people in charge. For instance, the Perl patch-pumpkin is an efficient way to avoid conflicts, combining time-sharing with classical centralism. Methods for conflict resolution has been developed, too, and at least my own TUNES Project includes a formal written Charter describing sensible mechanisms to resolve conflicts, that I believe can be observed as being in actual informal use by other projects.
Of course, the very possibility of a project split is the ultimate solution to irreducible conflicts. That such split be possible without involving the cost of either legal and financial battles or that of a rewrite of software from scratch with incompatible specifications, is a major feature of free software. Surely, gratuitous splits are discouraged; but justified splits that do not prevent further code exchange have often proved a great advantage, accounting for the flexibility of free software (see for instance the GNU Emacs/XEmacs/MULE/MBSK and GCC/EGCS splits).
In reply to my remarks, Eric asserted the following:
In practice, the distinction between noosphere and ergosphere is not important for the purposes of this paper. It is dubious whether the `noosphere' in the pure sense Fare insists on can be said to exist in any meaningful way; one would almost have to be a Platonist philosopher to believe in it. And the distinction between noosphere and ergosphere is only of practical importance if one wishes to assert that ideas (the elements of the noosphere) cannot be owned, but their instantiations as projects can. This question leads to issues in the theory of intellectual property which are beyond the scope of this paper.I believe that Eric has it all wrong on these points, and that my above arguments explained why.
It is precisely because programming happens in the same ergosphere as all services that classical economics applies perfectly to it, and that there needs be nothing "magic" or irreducibly peculiar about the dynamics and ownership of programming projects, and about its associated "taboos". Not only is the situation of software not exceptional, but it can also be generalized into a theory of `Libre' Information as a natural requirement for free markets. And of course, if ideas were not to "exist in any meaningful way", then nothing could justify ownership of them. However, argueing about the precise intended meaning of "idea", "exist", and "meaning" would lead us far away from free software, into epistemology, that I'd be happy to discuss, but not in this article. Finally, I think that Eric is misled when asserting that theory of "intellectual property" is beyond the scope of a paper that claims to discuss free software theory. I do reckon, that once again, Eric may have wonderfully described free software practice; but any attempt to go from description to explanation is in my opinion doomed to fail, if it doesn't take into account free software theory.
All the divergences of opinion between Eric and more radical free software people (including me) are not superficial. Aaron M. Renn showed it in some insightful analysis about free software versus open-source software. Eric has since confirmed that he regards ownership of software as continuous with other intangible property rights, and considers it fully justified by his classical-liberal and libertarian principles. In this he differs sharply and consciously from many advocates of `free software', some of whom (including I) consider themselves as classical-liberal and libertarian, too. Open source software and free software both describe the same practice of releasing software under "free" or equivalently "open source" licenses; but we can hereby differentiate "open source" from "free" as having different underlying philosophies.
Note: it appears that in some versions of this article, I may have written stupid and unfounded extrapolations on what the opinions of Eric may have been, and what would be his justifications for them. I was really silly, and I am sorry of the harm that I may have caused. So after all, this article also reveals my own problems, too. Mea culpa.
Current laws have created the artificial notion of "intellectual property rights", and made its use the only way to defend the authentification of information that is necessary to peer review. These laws have completely perverted economical institutions, as many corporations crucially depend on information license fees, and large fortunes have been built on information monopolies, rather than depending and building on payment of actual rendered services. These laws have created permanent biases in the market, and made mass-media a tool for Oppression rather than for Liberty. These laws are not something I want to compromise with, and the flawed justifications of these laws are to be fought. If Eric's arguments depended on these justifications, they would be opposite to the essence of free-software. But happily, they do not, and I wish Eric would make it obvious, rather than spread confusion in his articles.
Note however, that despite my being a claimed opponent to "intellectual property rights", I am annoyed at best when Eric, when pointing to this article from his page (for which I am most grateful to him), defines my position as "anti-IPR". My fighting those so-called "rights" is not the premises that define my position, or even a main characteristic feature of it, but just one of its many consequences. My position is that of a classical liberal, a defender of Liberty at large, and Freedom of Information in particular. IPRs are but ones among many attempts against liberty that I fight, although ones that are backed by many prejudices and important lobbies. Surely, fighting prejudices which serve to justify huge financial interests is a difficult task. It requires being all the stricter since the task is hard. I do not disagree with the notion of being efficient at promoting free software, and this may mean that some articles avoid some controversial issues. But I oppose to actions that are expedient on the short-term, but backfire on the long-run, because they accept to support many wrong ideas rather than fight them, in their quest to have just one good idea succeed.
All in all, I much respect Eric for his insightful insider's analysis of free or open-source software practice, which had never been done before; but I deplore his spreading what I believe are misconceptions about free software theories. I want free software ideas and their justification to be argued for or argued against for what they are; and this is only getting more difficult when the topmost literature about free software practice contributes to such misconceptions.
Now, of course, we need not agree on the theory so as to work together in practice. I fully back Eric's open-source software movement, in as much as it tries to promote free software practices in the industry. But I will also participate in other movements, so as to promote the acknowlegment of Freedom of Software and Freedom of Information by laws -- or perhaps more accurately, from laws.
This is my response to "The Magic Cauldron".
Eric Raymond at last gifted us with the long awaited "Magic Cauldron", third part of his announced tetralogy of articles about free (er, I mean "open-source") software. And once again, be him thanked for his contribution to the general understanding of open source phenomena, even though this third part is less ground-breaking than its fantastic prequels. However, once again too, I wish to clearly point out a few divergences that I have with respect to the analysis and presentation he makes in his articles.
To begin with, let me say a few things about magic.
Magic is what we feel when we face something wonderful and unexpected, something that we understand is good, be that we don't understand how it could be produced or reproduced. There is a lot of magic in the world. For instance, to me, there is some kind of magic in what Eric writes, and I mean not just the open source part. There is also a higher kind of magic that makes the things we love possible: and I mean not just free software, but civilization at large, life in general, and the very universe to begin with. Why is there a universe, and what justifies its physical laws? Why can we observe life in the universe, instead of a chaotic mass of meaningless events? Why do men build civilizations and technologies, instead of just eating each other, armed with sticks and stones?
It is important to note how magic has to do as much with our ignorance as with our understanding: it isn't in the finite and relatively small field of things we understand, nor is it in the infinite field of things which we ignore but it is at the finite but huge intermediate field of things of which we know, but that we don't know. Thus, any sufficiently complex phenomenon is not even magic, or undistinguishable from magic: it is utterly unspeakable and unfathomable. It takes a complex but sufficiently simple, speakable, phenomenon, like technology, to appear as magic to anyone. Now, technology is by definition is precisely a body of distributed knowledge about techniques: it works because every bit of required information about it has been mastered by someone, somewhere, and been made publicly available.
Knowledge is an evolving concept. When we extend the field of our knowledge, a small part of what once appeared as magic is explained; its magic vanishes, and is replaced by science. We may then produce, reproduce, modify, or otherwise tame phenomena that formerly were exclusive attributes of Gods, and henceforth adapt our behaviour to take advantage of this knowledge. Such is the way that civilization progresses. Of course, most of what is magic will forever remain so, for our means of explanation are limited; and we cannot usually predict what we'll be able to explain someday. But when we really care about some phenomenon, when it affects the way we live, then we must strive to understand it, and dispell the magic in it, so as to be actually able to reproduce the phenomenon, and to take the best advantage out of it. Let there be no remorse or nostalgy in dispelling magic, for if we manage, this paradoxically won't decrease the overall magic in the world, since with any interesting elucidation of some phenomenon springs more and more questions about other associated phenomena: ``the more we know, the more we know that we know not''; together with science comes more magic, not less magic.
The above remark isn't meant as a critique to Eric's choice of presentation, but rather as a complement to it. I do think Eric indeed tries his best to develop a science of free software, and he achieves more than most, although his style may mislead many a reader from a few essential ideas about free software. I hope to give a useful complementary point of view, that allows people to see what Eric doesn't show.
Happily, ESR goes away from this non-explanation of a gift culture: the "gift culture" pattern had some description power, but it lacked any explanation power, and was unsuitable not only for economical or political analysis, but for psychological analysis as well, as I showed in my criticism of ESR's previous article.
I think that the new problematic "How do I make money at this?" is appropriate, and that it has always been. I refer to my article on free software (in French) for a paper that discusses this problem in detail. Let us just recall the fundamental arguments of this paper: what creates value is services; Free Software means a Free Market in services; Intellectual Property means monopolies in services; usual market arguments apply for Free Software and against Intellectual Property.
I conclude this paragraph by insisting that although ESR's "realpolitik" is to not directly criticize the political economic model behind proprietary software, the principle of this line of argument is just like not critizing the USSR for its attempts to liberty, only for its economic outcome. The truth is that the economic failures of proprietary software are directly related to their attempts to liberty, and that the benefits of Free Software are but those of a Free Market.
The thing is that, as I argued before, there has always been a market about Free Software. However, it was just not a market about information (which isn't a service), and only a small part of it is a market in development services (because the displacement of capitals due to proprietary software dominating the industry limits the free exchange of development services). It is a market of mindshare (or timeshare in attention and development). The gift culture was never a good explanation at any level, and we're glad that Eric dropped it. Geeks never bring gifts, they lack social skills for that.
We need to begin by noticing that copies of computer programs like all other kinds tools or goods or services are ultimately sought for their use value. (Which Bastiat calls "utility" in his Economic Harmonies, whereas he reserves the term "value" for the exchange value. As Bastiat brilliantly argues, the confusion between the two kinds of "values" is at the heart of most economic fallacies and confusions.)
The use value of the copy of a program is its value as an aid to decision, as a source of enlightment, as a component of an automated device. All value is ultimately use value, mediately or immediately, effectively or potentially. Any "value" that doesn't derive from use value is a fallacy. You can sell something only if it has some indirect use to someone; if it doesn't, then the sale is at the very least an error for the end customer; and if that error is artificially maintained by one of the sellers, then the sale is a fraud.
The Factory Model for software production is very appropriate even with free software to describe how mass-production software is developed, despite extremely low replication costs. Of course it doesn't cover services in general. Economics in general talk about services, not products; manufacturing products is but a particular case of services. The fundamental false premises with which the tenants of Intellectual Property corrupt the minds of the public are that licenses be a product, that licensing be a service. Development is a service, and sales of such service follow the usual economic models. Software distributions (in the form of boxes containing CDs or other media) are products, and sales of such products follow the usual economic models. Licensing is but a racket, and sales of licenses follow the models of racketeering, not those of economics.
ESR justly explains that the real deserved money is in services. That's obvious: services are where economics tells us that value has always resided and will always reside. Not that obvious things needn't be repeated once in a while. Thus, let's thank ESR for reminding us about the obvious. But let's also remind that it's painfully obvious, and see how it confirms economic theory: software services, including development, provide value, and can be exchanged for other services, to everyone's benefit; licenses provide no intrinsic value, they are but a racket, and are a disservice to customers as well as a deterrent to the market in general, to the long-term detriment of providers as well as of customers.
The term "free" is not misleading. It's the very same "free" as in "Free Market", for it's all about having a Free Market of services, as opposed to a collection of service monopolies.
ESR gives us once again the very same fallacy as in his "Noosphere" article.
Information does want to be free of rights, and information delivery does want to be free of charge. If the acts of thinking, rethinking and communicating information do not hurt any third party, then they should be freely admitted. Political Liberty 101. If the marginal cost of replicating information is almost zero, then the marginal price of the corresponding service will also be almost zero, in a free market. Free Market Economy 101. Any attempt to make replication costs artificially high only introduces economic inefficiency and political oppression. Protectionism 101.
Now, not only does information want to be free, it also wants to be useless. This is the phenomenon well-known to hackers as "bit rot", as any jargon-file reader (and writer) should know. As John Perry Barlow puts it, "Information is Perishable". This means that although (the right to access, use, etc) information itself will cost nothing in a free market, there is a wide range of associated services that will cost something, and that represent a huge market for people to honestly earn money, without the need of licensing rackets. Such services include timely distribution of information, distribution of reliable information, guaranteeing the value of information, giving information adapted to the specific needs of a customer, regularly updating available information, etc, etc.
To reuse the examples of ESR, the treasure map will ultimately cost zero, because its value will ultimately be zero, right after someone finds and claims the treasure. Same with the Swiss account number: once someone uses it and empties the account, the number becomes useless. In both cases, not just the information, but the timing is valuable. Time is money. Everyone knows it. Giving the information to ones before to give it to others is a valuable service. Keeping a secret is a service. The right to use information by itself isn't a service, it needn't and shouldn't be "protected" by law. Giving the opportunity to use the information is a service, and that is what should receive the same legal protection as any service, by removing protectionist barriers against it.
There is no tragedy of the commons with information, because when a "free-rider" uses available information, he consumes his own resources, not anyone else's; he doesn't empty any common pool. What would be a tragedy of the commons would be that people be entitled a marginally free access to a customer service. Mind you, that's exactly what happens for customer support using the proprietary model of software licensing, and anyone having used software support in the 80's and in the 90's can tell how the tragedy happened and resulted in extremely poor "support" being given to common customers.
Information is not physical resources. It obeys different laws. ESR rightly shows how there is a phenomenon of "Inverse Commons" with information. Information gains by being spread, not restricted. Intellectual Property's very nature is to prevent the spread and use of information, despite its alleged initial goal of having it encourage publication.
ESR justly reminds that paying authors for the value of the information they produced is a mirage that can't ever be implemented. It's a mirage that maybe everyone once dreamt about (I did) when thinking about the economics of authorship with one's mind corrupted by the intellectual property myths. But Eric's reminding is no explanation. Now, the explanation is simple: paying authors for the value of information is a myth, because information has no intrinsic value. Authors should be paid for the services they render, that is, the act of creation. The fee for such services will be in direct relation not with the hypothetical value of produced information, only with the value (including a part due to rarity) of the author's proficiency. Judging the "value" of produced information, which is best done by peer review of freely available information, is but the way to evaluate the value of an author's proficiency; that's exactly how universities judge academic researchers, and that's exactly how free software vendors judge free software developers. Free publication is essential to a fair evaluation of authors' proficiencies, and thus is an essential component of a Free Market that employs these proficiencies.
It is trivial to state that closing source is a form of protectionism that locks the market of services around a given program. If the size of the market is the same, protectionism always benefits the copyright holder who hoards the source. However, the size of the market will not be the same. In presence of competing software, free or not, hoarders will have to lure customers into their captive market.
The argument of the "monopoly value" of software doesn't hold. Competitors are able to spend the same amount of money in a competing software project, free or not. If they do not use the considered software, it doesn't give anyone any competitive edge that the software be closed; if they will use the considered software, it won't give them an edge either, if they don't contribute back their adaptations, and it won't give them an edge if they contribute back. In any case, free software won't in the long run change the competitivity of users; it will, however, reduce cost, and thus provide early adopters with an edge.
Closed software doesn't benefit any honest worker. It will, however, benefit cheats and swindlers, and a whole class of managers, who will be able to exploit programmers whose proficiency value is tied to proprietary products, as well as customers who are locked into a monopoly. Said swindlers are powerful enough to lobby governments into enforcing and extending intellectual property.
All value is ultimately use-value. Some product or service has value if and only if someone finds some value at using it. Any other value is a fallacy. If no one sees any value in using something, then that something won't be sold to anyone, and will have no value. Maybe some people will get lured into buying it before everyone realizes, hoping to resell it for a higher price; but that will be only as an error or as a fraud. Sale Value that is not ultimately backed by Use Value is a Fraud. As a result, not only are ESR's Use-Value models valid, Use-Value models are the only kind of valid Economic Models. Similarly, ESR's "Indirect Sale Value" are honest if and only if they are actually "Indirect Use Value", if they correspond to sale of useful services. Note that the "free the software, sell the brand" strategy is exactly the one by RedHat, Mandrake, and other makers of Linux "distributions" all of whose contributed software is free.
Unless you're a swindler, your interest is in keeping software Free, and fighting Intellectual Property. Now, the question of secret vs public is different. For keeping things secret is completely different from information protectionism: it doesn't involve the use of governmental force, only someone's own efforts to keep things secret; the burden of the secret is upon the person who wants to protect oneself, not upon the taxpayer. Similarly, non-disclosure agreements can help distribute exclusive information while keeping any enforcement costs for the exclusion among the signatories of such agreements; it is crucial that they should not bind non-signatories. I accept the main part ESR's analysis, but assert that it is a problem independent from Intellectual Property.
As more and more people understand that Free Software is all about a Free Market of software services, they will demand free software, as customers. The whole problem is about getting them to understand it, whether by practical experience or by theoretical explanation. Actually, these are complementary approaches. I am grateful to ESR for successfully providing a summary of the former. I will continue to assert the latter.
As Frédéric Bastiat explained in the article Abundance and Scarcity of his Economic Sophism, it has always been and will always be the interest of producers as such to have high and/or guaranteed revenues, which can only be achieved by increasing demand and decreasing supply, i.e. by creating scarcity. It has always been and will always be the interest of consumers as such to have low and/or guaranteed prices, which can only be achieved by decreasing demand and increasing supply, i.e. by creating abundance. The common good lies in the interest of the consumer. Producers will continually try to impose protectionist schemes such as intellectual property; consumers will continually push for freedom. So yes, there will always be crooks who sell software licenses, and crooks who turn free software into proprietary software (as seen daily with lots of BSD code). It is up to us as consumers to ensure that our interest prevails, that the Beast die, so we can fully focus on real work at last.
It's all great if you can convince someone that what you do will be useful. But why would one write software without funding? How to fight Rocky's Lemma of Innovation Prevention? How to really innovate?
The answer is simple: developing great software for nothing is a way to show off your proficiency as well as a way to invest your time into tools that you are most able to use. It is a way both to demonstrate and to build proficiencies that will make your work valuable on the market. It is investment.
It's exactly the very same reason why some people do a PhD before to go on the employment market. If they started to work immediately instead of doing a PhD, they would spend three years in lucrative work, whereas they have to pay a lot a be paid little while they do their PhD. But when they are done, they have both demonstrated their skills, and built up new skills, skills in a specialized field where no one else can claim their proficiency. Once again, it's an investment.
Our revolution is but the eternal fight of freedom against oppression. Oppression takes more and more devious forms, because it knows that it loses all its strength when it is unmasked.
All software will eventually become free. System software first, then applications, then Middleware. Only limited use-once software with verifiable output, or games, will remain secret (not proprietary), and even then, not for long. But the revolution won't end there. With software becoming free software, new development models will pop up, new development tools, new programming environments, new programming methodologies, of which we have only seen the premises.
When Free Software becomes mainstream enough, some long-standing technical problems with software development will at least begin to be tackled, that couldn't be because of proprietary software hoarding. Cooperative programming techniques will be developed, metaprogramming will raise, etc. Our fight will not end just because we will have won one battle. Life is a fight, a perpetual fight for freedom against those who will constantly pop up to oppress other people by fraud or by force. But apart from the fight, it is also a lot of hard work, to make Life better with enhanced techniques. Each time we win a fight against those who destroy, we have to focus on building up.
ESR's initial question was:
To many people, the successes of the open-source community seem like an implausible form of magic. High-quality software materializes ``for free'', which is nice while it lasts but hardly seems sustainable in the real world of competition and scarce resources. What's the catch? Is Ceridwen's cauldron just a conjuring trick? And if not, how does ephemeralization work in this context -- what spell is the goddess speaking?
He could have asked the same about Civilization at large: why aren't we just eating each other, armed with stones and sticks? And the answer would be the same, too: because no one owns progress, because by definition, civilization is shared, and what isn't shared isn't civilization.
To people who are interested in the basic principles of civilization and economics, I recommend Frédéric Bastiat's Economic Harmonies (now also available in original French).
Browse back to: